[go: up one dir, main page]

WO2025230256A1 - Procédé exécuté par un équipement utilisateur dans un système de communication sans fil, procédé exécuté par un nœud, équipement utilisateur et nœud - Google Patents

Procédé exécuté par un équipement utilisateur dans un système de communication sans fil, procédé exécuté par un nœud, équipement utilisateur et nœud

Info

Publication number
WO2025230256A1
WO2025230256A1 PCT/KR2025/005719 KR2025005719W WO2025230256A1 WO 2025230256 A1 WO2025230256 A1 WO 2025230256A1 KR 2025005719 W KR2025005719 W KR 2025005719W WO 2025230256 A1 WO2025230256 A1 WO 2025230256A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
model
cube
access point
optionally
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/KR2025/005719
Other languages
English (en)
Inventor
Wei Li
Xiangli LIN
Bin Wang
Ying Li
Meifang Jing
Huiyang WANG
Yinghuan Zhao
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Samsung Electronics Co Ltd
Original Assignee
Samsung Electronics Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Samsung Electronics Co Ltd filed Critical Samsung Electronics Co Ltd
Publication of WO2025230256A1 publication Critical patent/WO2025230256A1/fr
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/06—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas at the transmitting station
    • H04B7/0613—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas at the transmitting station using simultaneous transmission
    • H04B7/0615—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas at the transmitting station using simultaneous transmission of weighted versions of same signal
    • H04B7/0619—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas at the transmitting station using simultaneous transmission of weighted versions of same signal using feedback from receiving side
    • H04B7/0621—Feedback content
    • H04B7/0626—Channel coefficients, e.g. channel state information [CSI]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/309—Measuring or estimating channel quality parameters
    • H04B17/318—Received signal strength
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/391—Modelling the propagation channel
    • H04B17/3913—Predictive models, e.g. based on neural network models
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/08—Testing, supervising or monitoring using real traffic
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/10—Scheduling measurement reports ; Arrangements for measurement reports
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W48/00—Access restriction; Network selection; Access point selection
    • H04W48/20—Selecting an access point

Definitions

  • the present disclosure relates to the technical field of wireless communication, and in particular to a method executed by a user equipment in a wireless communication system, a method executed by a node, a user equipment and a node.
  • 5G mobile communication technologies define broad frequency bands such that high transmission rates and new services are possible, and can be implemented not only in “Sub 6GHz” bands such as 3.5GHz, but also in “Above 6GHz” bands referred to as mmWave including 28GHz and 39GHz.
  • 6G mobile communication technologies referred to as Beyond 5G systems
  • terahertz bands for example, 95GHz to 3THz bands
  • IIoT Industrial Internet of Things
  • IAB Integrated Access and Backhaul
  • DAPS Dual Active Protocol Stack
  • 5G baseline architecture for example, service based architecture or service based interface
  • NFV Network Functions Virtualization
  • SDN Software-Defined Networking
  • MEC Mobile Edge Computing
  • multi-antenna transmission technologies such as Full Dimensional MIMO (FD-MIMO), array antennas and large-scale antennas, metamaterial-based lenses and antennas for improving coverage of terahertz band signals, high-dimensional space multiplexing technology using OAM (Orbital Angular Momentum), and RIS (Reconfigurable Intelligent Surface), but also full-duplex technology for increasing frequency efficiency of 6G mobile communication technologies and improving system networks, AI-based communication technology for implementing system optimization by utilizing satellites and AI (Artificial Intelligence) from the design stage and internalizing end-to-end AI support functions, and next-generation distributed computing technology for implementing services at levels of complexity exceeding the limit of UE operation capability by utilizing ultra-high-performance communication and computing resources.
  • FD-MIMO Full Dimensional MIMO
  • OAM Organic Angular Momentum
  • RIS Reconfigurable Intelligent Surface
  • an aspect of the present invention provides method and apparatus executed by a user equipmenet (UE) and a node.
  • UE user equipmenet
  • the embodiments of the present disclosure provide a method executed by a user equipment in a wireless communication system, a method executed by a node, a user equipment and a node.
  • the embodiments of the present disclosure provide the following technical schemes.
  • an embodiment of the present disclosure provides a method executed by a user equipment in a wireless communication system, including steps of:
  • the first information including model information related to at least one first artificial intelligence (AI) model used for access point selection;
  • AI artificial intelligence
  • the second model is one of the at least one first AI model
  • the UE related information includes channel state information (CSI).
  • the determining, through a second AI model and based on UE related information, second information related to an access point used for the UE comprises:
  • the second information determining, through the second AI model and based on the UE related information and information related to a UE distribution situation in a space associated with the second AI model, the second information.
  • an embodiment of the present disclosure provides a method executed by a node in a wireless communication system, including steps of:
  • first information including model information related to at least one first artificial intelligence (AI) model used for access point selection; and
  • AI artificial intelligence
  • the second information being determined by a second AI model based on UE related information, the second AI model being one of the at least one first AI model, the UE related information including CSI.
  • the second information is determined by the second AI model, based on the UE related information and the information related to the UE distribution situation in the space associated with the second AI model.
  • the number of the at least one first AI model is at least two, and the first information further includes information related to selection of each first AI model.
  • the information related to selection of each first AI model includes at least one of the following:
  • the second information includes at least one of the following:
  • the first value being used for indicating the access point used for the UE
  • the identifier information of the access point used for the UE is obtained based on the first value and a first mapping relationship, the first mapping relationship includes at least one second value and identifier information of the access point associated with each second value, and the first value is one of the at least one second value.
  • the second information is reported in a case where the first value is different from the first value reported last time, and/or the second information is reported in a case where the identifier information is different from the identifier information reported last time.
  • the UE related information further includes at least one of the following: information related to the UE's traffic; and, environment information related to the UE.
  • the method further includes: transmitting third information related to reporting, the second information being reported based on the third information.
  • the third information includes information related to at least one of the following:
  • a reporting period a reporting period
  • a reporting trigger event a reporting trigger event
  • the reporting trigger event includes an event related to at least one of the following: a change of the second AI model; a mobility of the UE; the traffic information of the UE; a reference signal receiving power (RSRP) measured by the UE; a change of the CSI; and, a change of the second information.
  • a change of the second AI model e.g., a mobility of the UE; the traffic information of the UE; a reference signal receiving power (RSRP) measured by the UE; a change of the CSI; and, a change of the second information.
  • RSRP reference signal receiving power
  • the event related to the mobility of the UE includes at least one of the following:
  • an amount of change of the speed of the UE is greater than or equal to a first threshold; the speed of the UE is greater than or equal to a second threshold; and, the speed of the UE is less than or equal to a third threshold;
  • the event related to the traffic information of the UE includes at least one of the following:
  • the traffic type of the UE is changed; the amount of change of the traffic amount of the UE is greater than or equal to a fourth threshold; the traffic amount of the UE is greater than or equal to a fifth threshold; and, the traffic amount of the UE is less than or equal to a sixth threshold;
  • the event related to the RSRP measured by the UE includes at least one of the following:
  • the RSRP measured by the UE is greater than or equal to a seventh threshold; the RSRP measured by the UE is greater than or equal to an eighth threshold; and, the RSRP measured by the UE is less than or equal to a ninth threshold.
  • the second information is reported in a case where a first condition is satisfied, wherein the first condition includes at least one of the following:
  • the speed of the UE is greater than or equal to the first threshold
  • the UE has received fourth information, the fourth information being used for indicating to report the second information related to the access point used for the UE.
  • the method further includes: receiving fifth information, wherein the fifth information is reported in a case where the first condition is not satisfied, and the fifth information includes identifier information of at least one access point and information related to CSI of the at least one access point.
  • the at least one access point includes at least one of the following:
  • an access point whose associated channel matrix has a feature value greater than or equal to a threshold.
  • the fifth information includes an encoded result of each access point in the at least one access point
  • the encoded result of each access point is obtained by encoding sixth information of this access point by using a third AI model based on a self-attention mechanism
  • the sixth information is obtained by fusing the identifier information of this access point with the information related to CSI of this access point.
  • the sixth information is obtained by fusing the encoded identifier information of this access point with the information related to CSI of this access point.
  • the second AI model is determined from the at least one first AI model based on at least one of the following:
  • position information of the UE information related to a spatial position associated with each first AI model; channel state information of the UE; information related to the UE channel state information of the space associated with each first AI model; information related to the signal strength of the access point associated with each first AI model; or information related to the UE's capabilities.
  • the method further comprises:
  • each fourth AI model transmitting seventh information associated with each fourth AI model, the seventh information being related to a UE distribution situation in a spatial position associated with the fourth AI model, wherein said each fourth AI model comprises the at least one first AI model.
  • the at least one first AI model is at least some of AI models in a first model set, and the first model set comprises an AI model associated with each of a plurality of cubes of a serving region.
  • the plurality of cubes of the serving region are obtained by dividing the serving region.
  • the plurality of cubes of the serving region are obtained by dividing the serving region based on the UE related information of a plurality of sample UEs in the serving region.
  • the UE related information comprises at least one of the following:
  • the plurality of cubes of the serving region are determined based on the geographic position information of each sample UE belonging to each UE class, wherein said each UE class is obtained by clustering the plurality of sample UEs based on the UE related information of the plurality of sample UEs.
  • the plurality of cubes of the serving region are obtained by extending a first cube associated with each UE class, wherein the first cube associated with said each UE class is determined based on the geographic position information of each sample UE belonging to each UE class.
  • the extending a first cube associated with each UE class is based on at least one of the following:
  • the first boundary of the first cube is extended to the second boundary of the serving region.
  • At least one of the two first cubes is extended to the blank region.
  • At least one of the two first cubes is extended to the blank region comprises:
  • extension speed associated with one of the boundaries is related to at least one of the following:
  • the extension speed associated with one of the boundaries is related to a ratio, the ratio being a ratio of the density of APs in the unit space extending to the blank region with the boundary being a starting point and the density of the sample UEs in the first cube to which the boundary belongs.
  • an embodiment of the present disclosure provides a user equipment in a wireless communication system, wherein the user equipment includes at least one transceiver and at least one processor coupled to the at least one transceiver, and the at least one processor is configured to execute the method executed by a user equipment provided in any one of the optional embodiments of the present disclosure.
  • an embodiment of the present disclosure further provides a node in a wireless communication system, wherein the node includes at least one transceiver and at least one processor coupled to the at least one transceiver, and the at least one processor is configured to execute the method executed by a node provided in any one of the optional embodiments of the present disclosure.
  • an embodiment of the present disclosure further provides a computer-readable storage medium having computer programs stored thereon that, when run in a processor, execute the method provided in any one of the embodiments of the present disclosure.
  • a computer program product including computer programs that, when executed by a processor, implement the method provided in any one of the optional embodiments of the present disclosure.
  • a method performed by a user equipment includes receiving, from a base station, first information, wherein the first information includes model information on at least one artificial intelligence (AI) model for selecting an access point; determining second information on the access point through an AI model among the at least one AI model, based on UE related information and UE distribution situation information associated with a space for the AI model, wherein the UE related information includes channel state information (CSI); and reporting, to the base station, the second information.
  • AI artificial intelligence
  • a method performed by a base station includes transmitting, to a user equipment (UE), first information, wherein the first information includes model information on at least one artificial intelligence (AI) model for selecting an access point; receiving, from the UE, second information, wherein the second information on the access point is determined through an AI model among the at least one AI model, based on UE related information including channel state information (CSI) and UE distribution situation information associated with a space for the AI model.
  • CSI channel state information
  • a user equipment comprises a transceiver, and a controller coupled with the transceiver, and configured to receive, from a base station, first information, wherein the first information includes model information on at least one artificial intelligence (AI) model for selecting an access point, determine second information on the access point through an AI model among the at least one AI model, based on UE related information and UE distribution situation information associated with a space for the AI model, wherein the UE related information includes channel state information (CSI), and report, to the base station, the second information.
  • AI artificial intelligence
  • CSI channel state information
  • a base station comprises a transceiver, and a controller coupled with the transceiver, and configured to transmit, to a user equipment (UE), first information, wherein the first information includes model information on at least one artificial intelligence (AI) model for selecting an access point, receive, from the UE, second information, wherein the second information on the access point is determined through an AI model among the at least one AI model, based on UE related information including channel state information (CSI) and UE distribution situation information associated with a space for the AI model.
  • CSI channel state information
  • FIG. 1 shows a schematic structure diagram of a wireless network system to which an embodiment of the present disclosure is applied
  • FIG. 2 shows a schematic structure diagram of an exemplary base station according to the present disclosure
  • FIG. 3 shows a schematic structure diagram of an exemplary user equipment according to the present disclosure
  • FIG. 4 shows an architecture diagram of a communication system according to an embodiment of the present disclosure
  • FIG. 5 shows a flowchart of a method executed by a user equipment according to an embodiment of the present disclosure
  • FIG. 6a shows a flowchart of a communication method according to an embodiment of the present disclosure
  • FIG. 6b shows a flowchart of another communication method according to an embodiment of the present disclosure
  • FIG. 7a shows schematic diagrams of cube division according to an embodiment of the present disclosure
  • FIG. 7b shows schematic diagrams of cube division according to an embodiment of the present disclosure
  • FIG. 7c shows principle diagrams of an extended cube according to an embodiment of the present disclosure
  • FIG. 7d shows principle diagrams of an extended cube according to an embodiment of the present disclosure
  • FIG. 7e shows principle diagrams of filling a blank region according to an embodiment of the present disclosure
  • FIG. 7f shows principle diagrams of filling a blank region according to an embodiment of the present disclosure
  • FIG. 7g shows principle diagrams of filling a blank region according to an embodiment of the present disclosure
  • FIG. 7h shows principle diagrams of filling a blank region according to an embodiment of the present disclosure
  • FIG. 7i shows a schematic diagram of a division result of cubes according to an embodiment of the present disclosure
  • FIG. 8a shows a principle diagram of training a cube-model according to an embodiment of the present disclosure
  • FIG. 8b shows a principle diagram of determining a distribution index of a cube according to an embodiment of the present disclosure
  • FIG. 8c shows a principle diagram of training a cube-model according to an embodiment of the present disclosure
  • FIG. 8d shows schematic diagrams of two associated cube lists for determining a cube according to an embodiment of the present disclosure
  • FIG. 8e shows schematic diagrams of two associated cube lists for determining a cube according to an embodiment of the present disclosure
  • FIG. 8f shows a principle diagram of determining a similarity between a UE and a cube according to an embodiment of the present disclosure
  • FIG. 8g shows a schematic diagram of an edge UE according to an embodiment of the present disclosure
  • FIG. 8h shows a schematic diagram of a non-edge UE according to an embodiment of the present disclosure
  • FIGS. 8i, 8j and 8k show several principle diagrams of determining AI models to be used by a UE according to an embodiment of the present disclosure
  • FIG. 9a shows a schematic diagram of an AP selection according to an embodiment of the present disclosure.
  • FIG. 9b shows schematic diagram of several situations in which a UE updates AP selection results according to an embodiment of the present disclosure
  • FIG. 9c shows a schematic diagram of a reporting mechanism according to an embodiment of the present disclosure.
  • FIG. 10 shows a principle diagram of generating report information based on an AI model according to an embodiment of the present disclosure
  • FIG. 11 shows a schematic diagram of an access point selection result according to an embodiment of the present disclosure
  • FIG. 12 shows a flowchart of a communication method according to an embodiment of the present disclosure
  • FIG. 13 shows a schematic diagram of an AI model according to an embodiment of the present disclosure
  • FIG. 14 shows a principle diagram of position encoding based on access point granularity according to an embodiment of the present disclosure
  • FIG. 15 shows a schematic diagram of channel information division according to an embodiment of the present disclosure
  • FIG. 16 shows a principle diagram of an encoder according to an embodiment of the present disclosure
  • FIG. 17a shows a schematic diagram of dividing a serving region according to an embodiment of the present disclosure
  • FIG. 17b shows a schematic diagram of determining an AI model for each cube according to an embodiment of the present disclosure
  • FIG. 18 shows a flowchart of a communication method according to an embodiment of the present disclosure.
  • FIG. 19 is a schematic structure diagram of an electronic device according to an embodiment of the present disclosure.
  • 5th-generation (5G) communication systems it is expected that the number of connected devices will exponentially grow. Increasingly, these will be connected to communication networks. Examples of connected things may include vehicles, robots, drones, home appliances, displays, smart sensors connected to various infrastructures, construction machines, and factory equipment. Mobile devices are expected to evolve in various form-factors, such as augmented reality glasses, virtual reality headsets, and hologram devices. In order to provide various services by connecting hundreds of billions of devices and things in the 6th-generation (6G) era, there have been ongoing efforts to develop improved 6G communication systems. For these reasons, 6G communication systems are referred to as beyond-5G systems.
  • 6G communication systems which are expected to be commercialized around 2030, will have a peak data rate of tera (1,000 giga)-level bps and a radio latency less than 100 ⁇ sec, and thus will be 50 times as fast as 5G communication systems and have the 1/10 radio latency thereof.
  • a full-duplex technology for enabling an uplink transmission and a downlink transmission to simultaneously use the same frequency resource at the same time
  • a network technology for utilizing satellites, high-altitude platform stations (HAPS), and the like in an integrated manner
  • HAPS high-altitude platform stations
  • an improved network structure for supporting mobile base stations and the like and enabling network operation optimization and automation and the like
  • a dynamic spectrum sharing technology via collision avoidance based on a prediction of spectrum usage an use of artificial intelligence (AI) in wireless communication for improvement of overall network operation by utilizing AI from a designing phase for developing 6G and internalizing end-to-end AI support functions
  • a next-generation distributed computing technology for overcoming the limit of user equipment (UE) computing ability through reachable super-high-performance communication and computing resources (such as mobile edge computing (MEC), clouds, and the like) over the network.
  • UE user equipment
  • MEC mobile edge computing
  • 6G communication systems in hyper-connectivity, including person to machine (P2M) as well as machine to machine (M2M), will allow the next hyper-connected experience.
  • services such as truly immersive extended reality (XR), high-fidelity mobile hologram, and digital replica could be provided through 6G communication systems.
  • services such as remote surgery for security and reliability enhancement, industrial automation, and emergency response will be provided through the 6G communication system such that the technologies could be applied in various fields such as industry, medical care, automobiles, and home appliances.
  • the embodiments of the present disclosure provide a method executed by a user equipment in a wireless communication system, a method executed by a node, a user equipment and a node.
  • the embodiments of the present disclosure provide the following technical schemes.
  • an embodiment of the present disclosure provides a method executed by a user equipment in a wireless communication system, including steps of:
  • the first information including model information related to at least one first artificial intelligence (AI) model used for access point selection;
  • AI artificial intelligence
  • the second model is one of the at least one first AI model
  • the UE related information includes channel state information (CSI).
  • the determining, through a second AI model and based on UE related information, second information related to an access point used for the UE comprises:
  • the second information determining, through the second AI model and based on the UE related information and information related to a UE distribution situation in a space associated with the second AI model, the second information.
  • an embodiment of the present disclosure provides a method executed by a node in a wireless communication system, including steps of:
  • first information including model information related to at least one first artificial intelligence (AI) model used for access point selection; and
  • AI artificial intelligence
  • the second information being determined by a second AI model based on UE related information, the second AI model being one of the at least one first AI model, the UE related information including CSI.
  • the second information is determined by the second AI model, based on the UE related information and the information related to the UE distribution situation in the space associated with the second AI model.
  • the number of the at least one first AI model is at least two, and the first information further includes information related to selection of each first AI model.
  • the information related to selection of each first AI model includes at least one of the following:
  • the second information includes at least one of the following:
  • the first value being used for indicating the access point used for the UE
  • the identifier information of the access point used for the UE is obtained based on the first value and a first mapping relationship, the first mapping relationship includes at least one second value and identifier information of the access point associated with each second value, and the first value is one of the at least one second value.
  • the second information is reported in a case where the first value is different from the first value reported last time, and/or the second information is reported in a case where the identifier information is different from the identifier information reported last time.
  • the UE related information further includes at least one of the following: information related to the UE's traffic; and, environment information related to the UE.
  • the method further includes: transmitting third information related to reporting, the second information being reported based on the third information.
  • the third information includes information related to at least one of the following:
  • a reporting period a reporting period
  • a reporting trigger event a reporting trigger event
  • the reporting trigger event includes an event related to at least one of the following: a change of the second AI model; a mobility of the UE; the traffic information of the UE; a reference signal receiving power (RSRP) measured by the UE; a change of the CSI; and, a change of the second information.
  • a change of the second AI model e.g., a mobility of the UE; the traffic information of the UE; a reference signal receiving power (RSRP) measured by the UE; a change of the CSI; and, a change of the second information.
  • RSRP reference signal receiving power
  • the event related to the mobility of the UE includes at least one of the following:
  • an amount of change of the speed of the UE is greater than or equal to a first threshold; the speed of the UE is greater than or equal to a second threshold; and, the speed of the UE is less than or equal to a third threshold;
  • the event related to the traffic information of the UE includes at least one of the following:
  • the traffic type of the UE is changed; the amount of change of the traffic amount of the UE is greater than or equal to a fourth threshold; the traffic amount of the UE is greater than or equal to a fifth threshold; and, the traffic amount of the UE is less than or equal to a sixth threshold;
  • the event related to the RSRP measured by the UE includes at least one of the following:
  • the RSRP measured by the UE is greater than or equal to a seventh threshold; the RSRP measured by the UE is greater than or equal to an eighth threshold; and, the RSRP measured by the UE is less than or equal to a ninth threshold.
  • the second information is reported in a case where a first condition is satisfied, wherein the first condition includes at least one of the following:
  • the speed of the UE is greater than or equal to the first threshold
  • the UE has received fourth information, the fourth information being used for indicating to report the second information related to the access point used for the UE.
  • the method further includes: receiving fifth information, wherein the fifth information is reported in a case where the first condition is not satisfied, and the fifth information includes identifier information of at least one access point and information related to CSI of the at least one access point.
  • the at least one access point includes at least one of the following:
  • an access point whose associated channel matrix has a feature value greater than or equal to a threshold.
  • the fifth information includes an encoded result of each access point in the at least one access point
  • the encoded result of each access point is obtained by encoding sixth information of this access point by using a third AI model based on a self-attention mechanism
  • the sixth information is obtained by fusing the identifier information of this access point with the information related to CSI of this access point.
  • the sixth information is obtained by fusing the encoded identifier information of this access point with the information related to CSI of this access point.
  • the second AI model is determined from the at least one first AI model based on at least one of the following:
  • position information of the UE information related to a spatial position associated with each first AI model; channel state information of the UE; information related to the UE channel state information of the space associated with each first AI model; information related to the signal strength of the access point associated with each first AI model; or information related to the UE's capabilities.
  • the method further comprises:
  • each fourth AI model transmitting seventh information associated with each fourth AI model, the seventh information being related to a UE distribution situation in a spatial position associated with the fourth AI model, wherein said each fourth AI model comprises the at least one first AI model.
  • the at least one first AI model is at least some of AI models in a first model set, and the first model set comprises an AI model associated with each of a plurality of cubes of a serving region.
  • the plurality of cubes of the serving region are obtained by dividing the serving region.
  • the plurality of cubes of the serving region are obtained by dividing the serving region based on the UE related information of a plurality of sample UEs in the serving region.
  • the UE related information comprises at least one of the following:
  • the plurality of cubes of the serving region are determined based on the geographic position information of each sample UE belonging to each UE class, wherein said each UE class is obtained by clustering the plurality of sample UEs based on the UE related information of the plurality of sample UEs.
  • the plurality of cubes of the serving region are obtained by extending a first cube associated with each UE class, wherein the first cube associated with said each UE class is determined based on the geographic position information of each sample UE belonging to each UE class.
  • the extending a first cube associated with each UE class is based on at least one of the following:
  • the first boundary of the first cube is extended to the second boundary of the serving region.
  • At least one of the two first cubes is extended to the blank region.
  • At least one of the two first cubes is extended to the blank region comprises:
  • extension speed associated with one of the boundaries is related to at least one of the following:
  • the extension speed associated with one of the boundaries is related to a ratio, the ratio being a ratio of the density of APs in the unit space extending to the blank region with the boundary being a starting point and the density of the sample UEs in the first cube to which the boundary belongs.
  • an embodiment of the present disclosure provides a user equipment in a wireless communication system, wherein the user equipment includes at least one transceiver and at least one processor coupled to the at least one transceiver, and the at least one processor is configured to execute the method executed by a user equipment provided in any one of the optional embodiments of the present disclosure.
  • an embodiment of the present disclosure further provides a node in a wireless communication system, wherein the node includes at least one transceiver and at least one processor coupled to the at least one transceiver, and the at least one processor is configured to execute the method executed by a node provided in any one of the optional embodiments of the present disclosure.
  • an embodiment of the present disclosure further provides a computer-readable storage medium having computer programs stored thereon that, when run in a processor, execute the method provided in any one of the embodiments of the present disclosure.
  • a computer program product including computer programs that, when executed by a processor, implement the method provided in any one of the optional embodiments of the present disclosure.
  • Couple and its derivatives refer to any direct or indirect communication between two or more elements, whether those elements are in physical contact with one another.
  • the term “or” is inclusive, meaning and/or.
  • controller means any device, system or part thereof that controls at least one operation. Such a controller may be implemented in hardware or a combination of hardware and software and/or firmware. The functionality associated with any particular controller may be centralized or distributed, whether locally or remotely.
  • phrases "at least one of,” when used with a list of items, means that different combinations of one or more of the listed items may be used, and only one item in the list may be needed.
  • “at least one of: A, B, and C” includes any of the following combinations: A, B, C, A and B, A and C, B and C, and A and B and C.
  • the term “set” means one or more. Accordingly, a set of items can be a single item or a collection of two or more items.
  • various functions described below can be implemented or supported by one or more computer programs, each of which is formed from computer readable program code and embodied in a computer readable medium.
  • application and “program” refer to one or more computer programs, software components, sets of instructions, procedures, functions, objects, classes, instances, related data, or a portion thereof adapted for implementation in a suitable computer readable program code.
  • computer readable program code includes any type of computer code, including source code, object code, and executable code.
  • computer readable medium includes any type of medium capable of being accessed by a computer, such as read only memory (ROM), random access memory (RAM), a hard disk drive, a compact disc (CD), a digital video disc (DVD), or any other type of memory.
  • ROM read only memory
  • RAM random access memory
  • CD compact disc
  • DVD digital video disc
  • a "non-transitory” computer readable medium excludes wired, wireless, optical, or other communication links that transport transitory electrical or other signals.
  • a non-transitory computer readable medium includes media where data can be permanently stored and media where data can be stored and later overwritten, such as a rewritable optical disc or an erasable memory device.
  • FIGS. 1-3 below describe various embodiments of the present disclosure implemented in wireless communications systems.
  • the descriptions of FIGS. 1-3 are not meant to imply physical or architectural limitations to the manner in which different embodiments may be implemented. Different embodiments of the present disclosure may be implemented in any suitably-arranged communications system.
  • FIG. 1 illustrates an example wireless network according to embodiments of the present disclosure.
  • the embodiment of the wireless network shown in FIG. 1 is for illustration only. Other embodiments of the wireless network 100 could be used without departing from the scope of the present disclosure.
  • the wireless network includes a base station (next generation nodeB, gNB or gNodeB) 101, a gNB 102, and a gNB 103.
  • the gNB 101 communicates with the gNB 102 and the gNB 103.
  • the gNB 101 also communicates with at least one network 130, such as the Internet, a proprietary Internet Protocol (IP) network, or other data network.
  • IP Internet Protocol
  • the gNB 102 provides wireless broadband access to the network 130 for a plurality of first user equipments (UEs) within a coverage area 120 of the gNB 102.
  • the plurality of first UEs includes a UE 111, which may be located in a small business; a UE 112, which may be located in an enterprise (E); a UE 113, which may be located in a WiFi hotspot (HS); a UE 114, which may be located in a first residence (R1); a UE 115, which may be located in a second residence (R2); and a UE 116, which may be a mobile device (M), such as a cell phone, a wireless laptop, a wireless personal digital assistant (PDA), or the like.
  • M mobile device
  • PDA wireless personal digital assistant
  • the gNB 103 provides wireless broadband access to the network 130 for a plurality of second UEs within a coverage area 125 of the gNB 103.
  • the plurality of second UEs include the UE 115 and the UE 116, as well as subscriber stations (SS, for example, UEs) 117, 118 and 119.
  • SS subscriber stations
  • one or more of the gNBs 101-103 may communicate with each other and with the UEs 111-116 using existing wireless communication techniques, and one or more of the UE 111-119 may communicate directly with each other (e.g., UEs 117-119) using other existing or proposed wireless communication techniques.
  • the term “base station” or “BS” can refer to any component (or collection of components) configured to provide wireless access to a network, such as transmit point (TP), transmit-receive point (TRP), an enhanced (or “evolved”) base station (eNodeB or eNB), a 5G base station (gNB), a macrocell, a femtocell, a wireless fidelity (WiFi) access point (AP), or other wirelessly enabled devices.
  • TP transmit point
  • TRP transmit-receive point
  • eNodeB or eNB enhanced (or “evolved”) base station
  • gNB 5G base station
  • gNB 5G base station
  • AP wireless fidelity access point
  • Base stations may provide wireless access in accordance with one or more wireless communication protocols, e.g., 3GPP 5G New Radio (NR), Long Term Evolution (LTE), LTE Advanced (LTE-A), high speed packet access (HSPA), Wi-Fi 802.11a/b/g/n/ac, etc.
  • 3GPP 5G New Radio NR
  • LTE Long Term Evolution
  • LTE-A LTE Advanced
  • HSPA high speed packet access
  • Wi-Fi 802.11a/b/g/n/ac etc.
  • the various names for a base station-type apparatus and functionality are used interchangeably in this patent document to refer to network infrastructure components that provide wireless access to remote terminals.
  • the term "user equipment” (UE) can refer to any component such as a mobile station (MS), subscriber station (SS), remote terminal, wireless terminal, receive point, or user device.
  • MS mobile station
  • SS subscriber station
  • remote terminal wireless terminal
  • receive point or user device.
  • a user equipment-type device and functionality are used interchangeably in this patent document to refer to remote wireless equipment that wirelessly accesses a BS, whether the UE is a mobile device (such as a mobile telephone or smartphone) or is normally considered a stationary device (such as a desktop computer or vending machine).
  • Dotted lines show the approximate extents of the coverage areas 120 and 125, which are shown as approximately circular for the purposes of illustration and explanation only. It should be clearly understood that the coverage areas associated with gNBs, such as the coverage areas 120 and 125, may have other shapes, including irregular shapes, depending upon the configuration of the gNBs and variations in the radio environment associated with natural and man-made obstructions.
  • one or more of the UEs 111-119 include circuitry, programing, or a combination thereof.
  • one or more of the gNBs 101-103 includes circuitry, programing, or a combination thereof.
  • FIG. 1 illustrates one example of a wireless network
  • the wireless network could include any number of gNBs and any number of UEs in any suitable arrangement.
  • the gNB 101 could communicate directly with any number of UEs and provide those UEs with wireless broadband access to the network 130.
  • each gNB 102-103 could communicate directly with the network 130 and provide UEs with direct wireless broadband access to the network 130.
  • the gNBs 101, 102, and/or 103 could provide access to other or additional external networks, such as external telephone networks or other types of data networks.
  • FIG. 2 illustrates an example base station according to embodiments of the present disclosure.
  • the embodiment of the gNB 102 illustrated in FIG. 2 is for illustration only, and the gNBs 101 and 103 of FIG. 1 could have the same or similar configuration.
  • gNBs come in a wide variety of configurations, and FIG. 2 does not limit the scope of the present disclosure to any particular implementation of a gNB.
  • the gNB 102 includes multiple antennas 200a-200n, multiple radio frequency (RF) transceivers 201a-201n, transmit (TX) processing circuitry 203, and receive (RX) processing circuitry 204.
  • the gNB 102 also includes a controller/processor 205, a memory 206, and a backhaul or network interface (IF) 207.
  • RF radio frequency
  • TX transmit
  • RX receive
  • the gNB 102 also includes a controller/processor 205, a memory 206, and a backhaul or network interface (IF) 207.
  • IF backhaul or network interface
  • the RF transceivers 201a-201n receive, from the antennas 200a-200n, incoming RF signals, such as signals transmitted by UEs in the network 100.
  • the RF transceivers 201a-201n down-convert the incoming RF signals to generate intermediate frequency (IF) or baseband signals.
  • the IF or baseband signals are sent to the RX processing circuitry 204, which generates processed baseband signals by filtering, decoding, and/or digitizing the baseband or IF signals.
  • the RX processing circuitry 204 transmits the processed baseband signals to the controller/processor 205 for further processing.
  • the TX processing circuitry 203 receives analog or digital data (such as voice data, web data, electronic mail, or interactive video game data) from the controller/processor 205.
  • the TX processing circuitry 203 encodes, multiplexes, and/or digitizes the outgoing baseband data to generate processed baseband or IF signals.
  • the RF transceivers 201a-201n receive the outgoing processed baseband or IF signals from the TX processing circuitry 203 and up-converts the baseband or IF signals to RF signals that are transmitted via the antennas 201a-201n.
  • the controller/processor 205 can include one or more processors or other processing devices that control the overall operation of the gNB 102.
  • the controller/processor 205 could control the reception of forward channel signals and the transmission of reverse channel signals by the RF transceivers 201a-201n, the RX processing circuitry 204, and the TX processing circuitry 203 in accordance with well-known principles.
  • the controller/processor 205 could support additional functions as well, such as more advanced wireless communication functions.
  • the controller/processor 205 could support beam forming or directional routing operations in which outgoing signals from multiple antennas 200a-200n are weighted differently to effectively steer the outgoing signals in a desired direction. Any of a wide variety of other functions could be supported in the gNB 102 by the controller/processor 205.
  • the controller/processor 205 is also capable of executing programs and other processes resident in the memory 206, such as an operating system (OS).
  • OS operating system
  • the controller/processor 205 can move data into or out of the memory 206 as required by an executing process.
  • the controller/processor 205 is also coupled to the backhaul or network interface 207.
  • the backhaul or network interface 207 allows the gNB 102 to communicate with other devices or systems over a backhaul connection or over a network.
  • the interface 207 could support communications over any suitable wired or wireless connection(s).
  • the gNB 102 is implemented as part of a cellular communication system (such as one supporting 5G, LTE, or LTE-A)
  • the interface 207 could allow the gNB 102 to communicate with other gNBs over a wired or wireless backhaul connection.
  • the interface 207 could allow the gNB 102 to communicate over a wired or wireless local area network or over a wired or wireless connection to a larger network (such as the Internet).
  • the interface 207 includes any suitable structure supporting communications over a wired or wireless connection, such as an Ethernet or RF transceiver.
  • the memory 206 is coupled to the controller/processor 205.
  • Part of the memory 206 could include a random access memory (RAM), and another part of the memory 206 could include a Flash memory or other read only memory (ROM).
  • RAM random access memory
  • ROM read only memory
  • FIG. 2 illustrates one example of gNB 102
  • the gNB 102 could include any number of each component shown in FIG. 2.
  • an access point could include a number of interfaces 207, and the controller/processor 205 could support routing functions to route data between different network addresses.
  • the gNB 102 while shown as including a single instance of TX processing circuitry 203 and a single instance of RX processing circuitry 204, the gNB 102 could include multiple instances of each (such as one per RF transceiver).
  • various components in FIG. 2 could be combined, further subdivided, or omitted and additional components could be added according to particular needs.
  • FIG. 3 illustrates an example user equipment according to embodiments of the present disclosure.
  • the embodiment of the UE 116 illustrated in FIG. 3 is for illustration only, and the UEs 111-115 and 117-119 of FIG. 1 could have the same or similar configuration.
  • UEs come in a wide variety of configurations, and FIG. 3 does not limit the scope of the present disclosure to any particular implementation of a UE.
  • the UE 116 includes an antenna 301, a radio frequency (RF) transceiver 302, TX processing circuitry 303, a microphone 304, and receive (RX) processing circuitry 305.
  • the UE 116 also includes a speaker 306, a controller or processor 307, an input/output (I/O) interface (IF) 308, a touchscreen display 310, and a memory 311.
  • the memory 311 includes an OS 312 and one or more applications 313.
  • the RF transceiver 302 receives, from the antenna 301, an incoming RF signal transmitted by an gNB of the network 100.
  • the RF transceiver 302 down-converts the incoming RF signal to generate an IF or baseband signal.
  • the IF or baseband signal is sent to the RX processing circuitry 305, which generates a processed baseband signal by filtering, decoding, and/or digitizing the baseband or IF signal.
  • the RX processing circuitry 305 transmits the processed baseband signal to the speaker 306 (such as for voice data) or to the processor 307 for further processing (such as for web browsing data).
  • the TX processing circuitry 303 receives analog or digital voice data from the microphone 304 or other outgoing baseband data (such as web data, e-mail, or interactive video game data) from the processor 307.
  • the TX processing circuitry 303 encodes, multiplexes, and/or digitizes the outgoing baseband data to generate a processed baseband or IF signal.
  • the RF transceiver 302 receives the outgoing processed baseband or IF signal from the TX processing circuitry 303 and up-converts the baseband or IF signal to an RF signal that is transmitted via the antenna 301.
  • the processor 307 can include one or more processors or other processing devices and execute the OS 312 stored in the memory 311 in order to control the overall operation of the UE 116.
  • the processor 307 could control the reception of forward channel signals and the transmission of reverse channel signals by the RF transceiver 302, the RX processing circuitry 305, and the TX processing circuitry 303 in accordance with well-known principles.
  • the processor 307 includes at least one microprocessor or microcontroller.
  • the processor 307 is also capable of executing other processes and programs resident in the memory 311, such as processes for CSI (Channel State Information) reporting on uplink channel.
  • the processor 307 can move data into or out of the memory 311 as required by an executing process.
  • the processor 307 is configured to execute the applications 313 based on the OS 312 or in response to signals received from gNBs or an operator.
  • the processor 307 is also coupled to the I/O interface 309, which provides the UE 116 with the ability to connect to other devices, such as laptop computers and handheld computers.
  • the I/O interface 309 is the communication path between these accessories and the processor 307.
  • the processor 307 is also coupled to the touchscreen display 310.
  • the user of the UE 116 can use the touchscreen display 310 to enter data into the UE 116.
  • the touchscreen display 310 may be a liquid crystal display, light emitting diode display, or other display capable of rendering text and/or at least limited graphics, such as from web sites.
  • the memory 311 is coupled to the processor 307. Part of the memory 311 could include RAM, and another part of the memory 311 could include a Flash memory or other ROM.
  • FIG. 3 illustrates one example of UE 116
  • various changes may be made to FIG. 3.
  • various components in FIG. 3 could be combined, further subdivided, or omitted and additional components could be added according to particular needs.
  • the processor 307 could be divided into multiple processors, such as one or more central processing units (CPUs) and one or more graphics processing units (GPUs).
  • FIG. 3 illustrates the UE 116 configured as a mobile telephone or smartphone, UEs could be configured to operate as other types of mobile or stationary devices.
  • At least some of the functions in the electronic device may be implemented by an AI model.
  • one or more steps of the method executed by an electronic device may be implemented by an AI model.
  • the functions associated with AI may be performed through a non-volatile memory, a volatile memory, and a processor.
  • the processor may be general-purpose processors such as central processing units (CPUs), application processors (APs), or pure graphics processing units such as graphics processing units (GPUs) and visual processing units (VPUs), and/or AI-specific processors such as neural processing units (NPUs).
  • the processor controls the processing of input data according to predefined operating rules or artificial intelligence (AI) models stored in the non-volatile memory and the volatile memory.
  • the predefined operating rules or AI models are provided by training or learning.
  • providing by learning refers to obtaining the predefined operating rules or AI models having a desired characteristic by applying a learning algorithm to a plurality of learning data.
  • This learning may be performed in the electronic device itself in which the AI model according to an embodiment is performed, and/or may be implemented by a separate server/system.
  • the AI model may include a plurality of neural network layers. Each layer has a plurality of weight values. Each layer performs neural network computation by computation between the input data of that layer (e.g., the computation result of the previous layer and/or the input data of the AI model) and the plurality of weight values of the current layer.
  • the learning algorithm is a method of training a predetermined target apparatus (e. g., a robot) by using a plurality of learning data to enable, allow, or control the target apparatus to make a determination or prediction. Examples of the learning algorithm include, but not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.
  • At least one of the steps of the method executed by an electronic device may be implemented by an artificial intelligence model.
  • the processor of the electronic device may preprocess data to convert the data into a form suitable for use as an input to the artificial intelligence model.
  • the artificial intelligence model may be obtained by training.
  • "obtained by training” means that predefined operating rules or artificial intelligence models configured to perform desired features (or purposes) are obtained by training a basic artificial intelligence model with multiple pieces of training data through a training algorithm.
  • the scheme provided in the embodiments of the present disclosure can be applied to distributed MINO (D-MIMO) systems, such as cell-free massive MIMO systems.
  • D-MIMO is widely concerned due to its benefits from good performances of the distributed antenna system and massive MIMO.
  • the accuracy of the obtained channel state information (CSI) of the transmitter is a key factor that affects the performance.
  • the D-MIMO can provide a large throughput gain.
  • C-MIMO centralized MIMO
  • the D-MIMO combines a point-to-point massive MIMO system with a distributed antenna system, and a large number of access points (APs) cooperate to service UEs.
  • APs access points
  • FIG. 4 shows a diagram of an architecture example of a possible cell-free communication system.
  • This system services a plurality of UEs on the same time-frequency resource through a plurality of APs discretely deployed in a serving region, and the APs are connected to a central controller (CC) through forward links.
  • CC central controller
  • Each UE has no cell boundary, so it is called cell-free or decelluarized.
  • Each UE is served by one or more APs.
  • the problems of inter-cell interference and inter-cell handover do not occur in cell-free communication systems.
  • a main challenge for the D-MIMO system is how to determine which access points a UE is served by to achieve the maximum system gain.
  • the selection of the APs will affect a signal to interference plus noise ratio (SINR) of the UE, so an AP selection scheme is very important for cell-free massive MIMO, which nay maximize the efficiency and throughput.
  • SINR signal to interference plus noise ratio
  • the determination of serving access points depends on the CSI between UEs and APs, so it is necessary to acquire accurate CSI.
  • the acquisition of the CSI may be performed in a way based on uplink pilot measurement or downlink pilot measurement.
  • the UE sends a pilot signal
  • the AP measures pilot state information and transmits it to the CC
  • the CC determines the downlink precoding and other information of the serving access point and the corresponding user.
  • the CC also called base station or base station side
  • the CC may configure downlink pilot information for the AP
  • the AP transmits a downlink pilot to the UE according to the configuration
  • the UE performs downlink pilot measurement to acquire CSI
  • the AP reports the CSI to the CC
  • the CC determines the serving access point for the UE based on the reported information.
  • the acquisition of CSI will lead to a huge radio resource overhead.
  • the uplink pilot signal can be used to obtain the CSI to reduce the overhead, this method using channel reciprocity is only suitable for time division duplex (TDD) systems, and many factors (e.g., channel stability, calibration error, etc.) will affect the accuracy of channel estimation.
  • the base station After the base station obtains CSIs of all APs from all UEs, in order to make the best AP selection for the UE, the base station needs to determine an AP associated with the best RSRP for the UE, calculate a channel correlation between the UEs, and coordinate the resource allocation between the APs. Due to the large number of UEs and APs in a MIMO system, computing overhead on the base station side is very huge.
  • the selection of the APs may also adopt a rule-based selection scheme.
  • the UE may make the AP selection according to the CSIs of all APs estimated by it, but this scheme has poor performance due to the absence of CSIs, because the UE cannot obtain CSIs of other UEs and cannot make the best AP selection for itself.
  • the embodiments of the present disclosure provide a new communication scheme. Based on this scheme, access point selection can be realized on the premise of using less radio resource overhead. Based on the scheme provided in the embodiments of the present disclosure, one or more best access points can be selected for the user quickly and accurately.
  • one possible scheme is to maximize performance metrics such as coverage distance and received reference signal power depending on a simple function of the access point selection index.
  • Access points may be ranked based on the performance metrics, and top N access points ranked may be selected according the performance metrics.
  • AI and machine learning can be applied in the CSI feedback and access point selection scheme.
  • the scheme provided in the embodiments of the present disclosure may be implemented based on the AI technology.
  • the scheme provided in the embodiments of the present disclosure can achieve remarkable performance by reducing feedback overhead and/or improving accuracy.
  • the scheme provided in the embodiments of the present disclosure may adopt a collaborative scheme between the base station and the UE, the AI model may be generated by the base station, and the UE uses the AI model.
  • At least one of the problems of reducing the computing complexity of the AI model, ensuring the performance of the UE, and reducing transmission overhead between the UE and the base station may be achieved by using alternative schemes provided in the embodiments of the present disclosure.
  • the scheme corresponding to a node on the other side that directly or indirectly communicates with this node can be obtained.
  • a UE receives information from a node; it can be correspondingly concluded that the node transmits information to the UE.
  • a node transmitting information to a UE may be that this node directly transmits information to the UE or this node transmits information to the UE through other nodes.
  • some term names involved in the embodiments of the present disclosure may adopt the term names that already exist in the communication standards, while some term names may be newly added or defined term names. These newly added or defined term names may also adopt other names in future communication standards, or may be described in other ways (e.g., a paragraph of text description).
  • the names or appellations of various information/messages/parameters/configurations involved in the embodiments of the present disclosure are not unique, and the names or appellations of these information/messages/parameters/configurations can be altered as long as the functions or contents of these information/messages/parameters/configurations or the explanations or descriptions of these information/messages/parameters/configurations can be corresponding or associated.
  • the CC may also be referred to as a central controller, a central processing unit (CPU), a control center, a main control center, a centralized control unit, a central server or other names.
  • CPU central processing unit
  • control center a main control center
  • main control center a main control center
  • centralized control unit a central server or other names.
  • FIG. 5 shows a method executed by a user equipment/user terminal in a wireless communication system according to an embodiment of the present disclosure, wherein the user equipment may be common terminal devices such as mobile phones, computers or wearable devices, or may be other electronic devices similar to or equivalent UEs.
  • the user equipment may be common terminal devices such as mobile phones, computers or wearable devices, or may be other electronic devices similar to or equivalent UEs.
  • the user equipment may communicate with a second node through one or more first nodes.
  • the user equipment, the first node and the second node may be nodes in a cell-free massive MIMO system.
  • the first node is a network node that communicates with the user equipment, and may be an access point, a base station (which may be construed as a base station with simplified functions), a super-surface antenna or the like.
  • the second node is used to process the information of all first nodes, and communicates with the first node through a forward link and communicates a core network through a backward link.
  • the combination of the first node and the second node may be interpreted as a base station, and all first nodes are connected to the second node to provide services for users and eliminate the cell boundary of the conventional cellular network.
  • the second node communicates with a core network and may provide configurations for the first node and the user equipment.
  • the first node and the second node take an AP and a CC as examples, respectively.
  • the method executed by a UE in a wireless communication system provided in the embodiment of the present disclosure may include the following steps S510 to S530.
  • step S510 first information is received, the first information including model information related to at least one first AI model used for access point selection.
  • step S520 second information related to an access point used for the UE is determined based on UE related information through a second AI model, the second AI model being one of the at least one first AI model, the UE related information including CSI.
  • step S530 the second information is reported.
  • the determining, through a second AI model and based on UE related information, second information related to an access point used for the UE comprises:
  • the second information determining, through the second AI model and based on the UE related information and information related to a UE distribution situation in a space associated with the second AI model, the second information.
  • the information related to the UE distribution situation may also be referred to as UE distribution information, spatial distribution information, UE distribution coding, distribution index, and the like.
  • Each AI model may be associated with its own spatial range (such as a cube), and the UE distribution situation in each spatial range is likely to be different. Since the CSI of the UE is related to the position of the UE, the UE distribution situation in a certain spatial range can reflect channel state situations in the spatial range.
  • the UE is using the second information determined through the second AI model, by using the UE distribution information associated with the used second AI model as an input of the model, the UE can obtain more accurate AP selection results based on its own CSI and the UE distribution information that can reflect channel state information in the spatial range.
  • the embodiments of the present disclosure do not limit the way in which the UE obtains the UE distribution information associated with each AI model.
  • the UE may receive the UE distribution information of each AI model broadcast by the CC.
  • the UE distribution information associated with each first AI model may also be included in the first information.
  • the UE may not need to report the CSI, but may, based on the UE related information (including the CSI), use the AI model to obtain and report the second information related to the access point of the UE.
  • the resource overhead can be effectively reduced based on the scheme provided in the embodiments of the present disclosure.
  • the UE may obtain the second information for reporting through the second AI model, based on the UE related information and the UE distribution information related to the selected second AI model.
  • the UE distribution information may reflect the distribution situation of UEs in a cube associated with the model (such as position information of the UEs in the cube), and may represent CSI information of other UEs in the cube. Therefore, the performance of AP selection can be better ensured by obtaining the second information based on the UE related information and the UE distribution information associated with the second AI model.
  • this method may further include: receiving configuration information related to CSI measurement; and, receiving at least one reference signal based on the configuration information.
  • the CSI is obtained based on the received reference signal.
  • the configuration information is configuration information used for CSI measurement.
  • the configuration information includes a reference signal resource configuration.
  • the CC configures, for the UE, one or more reference signal resources (e.g., channel state information-reference signal (CSI-RS) resources) used for CSI measurement through the configuration information.
  • the UE may receive, based on the configuration information, the reference signal transmitted through the AP by the CC.
  • the UE performs CSI measurement based on the received reference signal to obtain the CSI.
  • CSI-RS channel state information-reference signal
  • the CC may configure, for the UE, the reference signal resources related to each AP, so that the UE can estimate the channel between each AP and the UE based on the configuration to obtain the CSI corresponding to each AP.
  • the association between the AP and the reference signal resource may be an explicit association or an implicit association.
  • the reference signal resource configuration may include the identifier of each AP and the reference signal resource associated with the identifier of each AP; or, the reference signal resource configuration includes a plurality of reference signal resources and the indicator of each reference signal resource. The CC knows that each reference signal resource is associated with which AP, and the indicator of the reference signal resource may be used as the implicit identifier of the AP.
  • the UE may be known or unknown for the UE that which reference signal resource is associated with which AP.
  • the UE may only needs to receive a reference signal for measurement according to the configuration information; and, in the process of determining the second information by using an AI model, in accordance with the appointed rule or the rule configured by the CC, the UE may process the UE related information according to the rule and then input it into the model.
  • the UE may also know the identifier of the AP.
  • the UE may process, according to the rule, the UE related information into data that satisfies the input data requirements of the model, and then input it into the model.
  • the at least one first AI model may be one or more of a plurality of AI models pre-trained on the network node side (e.g., CC).
  • the first AI model is a mode related to access point selection.
  • the second information used for determining the serving access point of the UE may be predicted based on the UE related information.
  • the information may include implicit access point indication information or explicit access point indication information, for example, the identifiers of one or more APs.
  • the implicit indication information may be a code or a value. This code is the indication information or mapping value of the access point, and different codes may correspond to different AP or AP combinations.
  • the explicit access point indication information may be the identifiers of one or more APs.
  • the UE related information used for determining the second information may optionally include information related to the UE's traffic and/or environment information related to the UE.
  • the information related to the UE's traffic may include, but not limited to, the traffic type, traffic amount or other information of the UE, and the environment information related to the UE may include, but not limited to, geographic position related information of the UE (also referred to as UE geographic position similarity information) and/or channel state related information of the UE (channel similarity information of the UE).
  • the geographic position related information of the UE may include but is not limited to at least one of a position of the UE, the movement speed of the UE, the environment (e.g., indoor environment or outdoor environment) where the UE is located, and so on.
  • the channel status related information of the UE may include but is not limited to at least one of signal receive path information of the UE, the RSRP, and the channel state information of the UE.
  • the signal receive path information of the UE includes whether the signal receive path of the UE is a line-of-sight path or a non-line-of-sight path.
  • the UE related information may also be referred to as the feature data of the UE.
  • the second information may be determined based on the feature data of the UE.
  • different first AI models may be trained by using the training data with different UE feature data. The way of acquiring the corresponding training data will not be limited in the embodiment of the present disclosure.
  • the CC may receive the feature data of the UE reported by each UE, and classify these feature data. Each class corresponds to one AI model. One first AI model corresponding to each class may be obtained by training the corresponding AI model based on the training data of each class.
  • the first information may include model information related to one first AI model.
  • the UE may use this model to determine the second information related to an access point for the UE (for example, the above information for determining the serving access point of the UE), and the second information is reported to the CC.
  • the CC may determine the at least one corresponding AP as the serving access point of the UE based on the second information, and the CC transmits information for the UE through at least one of the at least one AP.
  • the CC may also select which APs as the serving access point of the UE with reference to only the second information reported by the UE. It depends on the CC. For example, the CC may determine the serving access point of each UE with reference to the second information received from a plurality of UEs.
  • the model information of any first AI model includes the model parameters of this model, for example, the weight matrix and offset of the model.
  • the first information includes the model information related to at least one first AI model used for access point selection, and the first information may also be described as including at least one first AI model used for access point selection (or the first AI model related to access point selection).
  • the number of the at least one first AI model is at least two.
  • the first information further includes information related to selection of each first AI model.
  • the UE may determine, based on the information related to model selection in the first information, to use which first AI model as the second AI model.
  • the information specifically included in the information related to selection of the first AI model will not be uniquely limited in the embodiment of the present disclosure.
  • the information related to selection of each first AI model may include at least one of the following:
  • different first AI models may be associated with or correspond to different spatial positions (also referred to as cubes).
  • the spatial position may be a spatial range.
  • the UE may select, according its position, the first AI model corresponding to the spatial position of its position as the second AI model.
  • the at least one first AI model may be at least some of AI models in a first model set (also referred to as first model library, first model group, etc.), and the first model set includes an AI model associated with each of a plurality of cubes of a serving region.
  • a first model set also referred to as first model library, first model group, etc.
  • the first model set includes an AI model associated with each of a plurality of cubes of a serving region.
  • the boundaries of the serving region may be preset, and the serving region may be divided into the plurality of cubes, and adjacent cubes may have overlap or no overlap.
  • the embodiments of the present disclosure do not uniquely limit the way in which the serving region is divided.
  • the serving region may be divided into the plurality of cubes according to a size of the serving region and a pre-configured space size, or the serving region may be divided into a set number of cubes.
  • the position, scope, and so on of the serving region may be predetermined, and a CC and a plurality of APs may be deployed in the serving region based on actual requirements.
  • the serving region may be determined based on already deployed APs. For example, a region (such as a minimum bounding cube) that can cover these APs may be determined according to these already deployed positions, may be used as the serving region, or may be extended outward to a certain extent, and the extended region may be used as the serving region.
  • one serving region is a coverage range where all APs connected under a CC can perform radio wave radiation.
  • the plurality of cubes of the serving region may be obtained by dividing the serving region based on the UE related information of a plurality of sample UEs in the serving region.
  • the UE related information comprises at least one of the following:
  • the plurality of cubes of the serving region may be determined based on the geographic position information of each sample UE belonging to each UE class, wherein said each UE class is obtained by clustering the plurality of sample UEs based on the UE related information of the plurality of sample UEs.
  • the UEs of each cube after division may have similar geographic positions and similar channel state information.
  • an AI model associated with the cube may be trained based on feature data of each sample UE associated with the cube.
  • the base station side may transmit relevant information of part or all of the trained AI models (the at least one first AI model) to the UEs in the serving region, and the UEs may select a more suitable second AI model from the received first AI models based on the UE related information.
  • the UEs may select the first AI model with a higher similarity according to the similarity between relevant information of the cube associated with the first AI model (such as geographic position information of the cube, UE channel state information of the cube, etc.) and the UE related information.
  • relevant information of the cube associated with the first AI model such as geographic position information of the cube, UE channel state information of the cube, etc.
  • multiple classes may be obtained by clustering based on the UE related information of the plurality of sample UEs.
  • Each class includes at least one sample UE, and the sample UEs in the same class have similar features.
  • a corresponding cube may be determined based on the geographic position information of each sample UE in the class.
  • a space that can cover the geographic position information of each UE in the class (such as a minimum bounding cube) may be used as the corresponding cube of the class.
  • the plurality of cubes of the serving region may be obtained by extending the first cube associated with each UE class, where the first cube associated with each UE class may be determined based on the geographic position information of each sample UE belonging to each UE class.
  • the determined first cube may be extended. By extending, each cube can contain more APs, and the blank region in the serving region is filled.
  • the extension of the first cube associated with each UE class is based on at least one of the following:
  • item 1 geographic position information of each AP whose signal strength meets the requirement of an edge UE in the first cube; or
  • item 2 a blank region in the serving region.
  • an edge UE in one cube refers to a UE located at an edge of the first cube, for example, a UE whose distance from any boundary of the first cube is less than a preset distance.
  • an AP in the serving region with better service quality for the edge UE may be determined, for example, an AP with higher RSRP for the edge UE.
  • an AP with good service quality for the edge UE may be selected among the APs that are located outside the first cube. By extending the initial first cube based on positions of these APs, these APs may also be included in the extended cube.
  • the cube may be extended to the blank region so as to fill the blank region.
  • the first boundary of the first cube is extended to the second boundary of the serving region.
  • At least one of the two first cubes is extended to the blank region to fill the blank region.
  • one or both of the two cubes may be extended to the blank region respectively, so as to fill the blank region between the two cubes, where the extended two cubes intersect or partially overlap.
  • the two cubes may be extended to the blank region respectively, and both of the extended two cubes cover the blank region.
  • the extended two cubes are the smallest bounding cubes that can cover the blank region.
  • the UE density may refer to the number of UEs in a unit area of the cube or the number of UEs in the cube.
  • that at least one original cube of the two first cubes is extended to the blank region may comprises:
  • extension speed associated with one of the boundaries is related to at least one of the following:
  • the density of APs in a unit space refers to the density of APs in a unit area of the space.
  • the density of the sample UEs may be the number of sample UEs in the cube or the number of UEs in a unit area of the cube (such as a ratio of the number of UEs in the and a size of the space).
  • the unit space may refer to a space formed with the boundary being a starting boundary of the unit space and extended outward with a unit length.
  • the unit length may be a specified length or a length determined according to boundary lengths of all the first cubes of the serving region (or the cubes extended after using the item 1). For example, the length of the shortest one of all the boundaries of the first cube may be used as the unit length.
  • the extension speed of a boundary may be positively correlated with a ratio of the density of APs in the unit space extending to the blank region with the boundary being a starting point and the density of the sample UEs in the first cube to which the boundary belongs.
  • FIG. 7g there is a blank region between a boundary i of the first cube a and a boundary j of the first cube b, and a side i is associated with two unit spaces.
  • There are 9 UEs in the cube a there are 5 APs in the first unit space and there are 2 APs in the second unit space, so the extension speed of the first unit space is 5/9, and the extension speed of the second unit space is 2/9.
  • different first AI models may have different effective times and/or expiration times, and the UE may select the first AI model in the effective stage as the second AI model.
  • the UE may report the environment information related to the UE to the CC, for example, the position of the UE, the movement speed of the UE or the like.
  • the CC may determine to transmit which model to the UE according to the environment information related to the UE. For example, the CC may predict the current position of the UE according to the position reported last time by the UE and the speed of the UE, and transmits the model information of at least one first AI model with the spatial position matched with the current position of the UE to the UE.
  • the second AI model may be determined by the UE from the at least one first AI model based on at least one of the following:
  • position information of the UE information related to a spatial position associated with each first AI model; channel state information of the UE; UE channel state information of the space associated with each first AI model; information related to signal strength of an access point associated with each first AI model; or information related to the UE's capabilities.
  • the UE may select one or more AI models from the plurality of first AI models received as the second model, and for a UE with weaker capability, the UE may select one first model as the second model.
  • the UE may use one or more first AI models whose associated spatial position (such as the geographic centroid of the cube) is closest to the position information of the UE as the second AI model.
  • the UE may obtain corresponding second information respectively by each second AI model based on the UE related information, and the UE may report part or all of the second information corresponding to the second AI model.
  • the channel state information of the UE refers to the channel state information obtained by the current UE through measurement.
  • the UE channel state information of the space associated with the first AI model represents the channel state information of the space associated with the first AI model, which may be obtained by statistics of the channel state information of each AP located in the space.
  • a CC may collect the channel state information of each AP reported by each UE in the serving space.
  • the CC may use the mean of the channel state information of each AP in the cube reported by each UE as the UE channel state information of the cube.
  • the UE When the UE selects the second AI model from the received first AI models, it may select at least one first AI model whose spatial UE channel state information is highly similar to the channel state information of the current UE as the second AI model, based on a similarity between the channel state information of the UE and the UE channel state information associated with the first AI model. For example, the UE may select at least one first AI model as candidate models based on its position information, and select the second AI model from the candidate models based on the similarity of channel state information.
  • the CC may explicitly or implicitly inform the UE of signal strength information of the AP associated with each first AI model, and the UE may select one or more first AI models with stronger signal strength as the second AI model based on the signal strength information of the AP associated with each first AI model.
  • the UE may select a plurality of first AI models whose associated spatial position is closest to the UE as the candidate models based on its position information, and then select at least one candidate model with stronger signal strength from the candidate models as the second AI model, according to the signal strength of the AP associated with each candidate model.
  • the CC may obtain the signal strength information of the AP based on statistical information. For example, the CC may calculate the mean of RSRP corresponding to each AP at the spatial position associated with the first AI model reported by each UE, and use this mean as the signal strength of the AP associated with the first AI model.
  • the methods provided in the embodiments of the present disclosure may also comprise:
  • each fourth AI model receiving seventh information associated with each fourth AI model, the seventh information being related to a UE distribution situation in a space associated with the fourth AI model, wherein said each fourth AI model comprises the at least one first AI model.
  • the CC may periodically broadcast UE distribution information associated with each AI model (for example, AI models in a first model set) to UEs in a serving region, or broadcast the updated UE distribution information associated with each AI model to the UEs in the serving region when the UE distribution information associated with the AI models changes greatly.
  • the UE distribution information associated with each first AI model may also be included in the above first information.
  • the CC may make the determination according to a distribution situation of each UE in the serving region.
  • the UE may input the information related to the UE (such as CSI) and the UE distribution information associated with the second AI model into the second model to obtain the second information and report it.
  • information related to the UE such as CSI
  • the UE distribution information associated with the second AI model into the second model to obtain the second information and report it.
  • the second information reported to the CC by the UE may include at least one of the following:
  • the first value being used for indicating the access point used for the UE
  • the first value may be the information output by the second AI model based on the UE related information, i.e., the output information of the second AI model.
  • Different output information may be associated with different APs.
  • the meaning of the first value may be known or unknown by the UE.
  • the UE may report the output information of the model to the UE according to the protocol appointment, and the meaning of the output information may be unknown for the UE.
  • the output of the second AI model may also be the identifier information of the AP or the probability value of each AP, and the UE may select and report the identifier information of a certain number of APs with larger probability values, or report the identifier information of APs with a probability value greater than or equal to a certain threshold.
  • the output of the second AI model may be the first value, and the UE may determine at least one AP corresponding to this first value according to the first value and report the identifier information of these APs.
  • the identifier information of the access point used for the UE is obtained based on the first value and a first mapping relationship, the first mapping relationship includes at least one second value and the identifier information of the access point associated with each second value, and the first value is one of the at least one second value.
  • the first mapping relationship may be configured for the UE by the CC.
  • the second information reported by the UE may include the output information of the model, and may also include the information obtained based on the output information of the model, for example, the first value or the identifier information of the AP determined according to the first value.
  • the specific form of the identifier information of the AP will not be limited, and may be the information which can uniquely identify the AP such as the index or serial number of the AP, or may be the identifier information of the AP obtained based on other information of the AP.
  • the identifier information may be the identifier obtained based on the antenna index and sub-band index of the AP.
  • the identifier information of the AP may also be represented by using the indicator (e.g., CSI-RS resource indicator) of the reference signal resource.
  • the UE reporting the second information may include:
  • the UE may perform the reporting operation when the second information determined currently is different from the second information reported last time by the UE.
  • the second information may adopt a periodic reporting mechanism, and the UE performs the operation of determining the second information based on the UE related information by using the second AI model according to the period configured by the CC or the predetermined period. If the second information determined this time is the same as the second information determined last time, the UE may not perform the reporting action, thereby reducing the consumption of communication resources, saving the battery level of the UE or the like.
  • the method provided by the present disclosure may further include: receiving third information related to reporting; and
  • the reporting the second information includes: reporting the second information based on the third information.
  • the third information configures how the UE can report the second information.
  • the third information includes information related to at least one of the following:
  • a reporting period a reporting period
  • a reporting trigger event a reporting trigger event
  • the third information may also be referred to as reporting configuration information.
  • the reporting configuration information may be configured together with the reference signal resource configuration used for CSI measurement.
  • the UE receives the configuration information related to CSI measurement, and the configuration information may include the reference signal resource configuration and the reporting configuration.
  • the UE may perform CSI measurement based on the reference signal resource configuration and may perform reporting based on the reporting configuration.
  • the CC may inform a reporting period to the UE through the third information, and the UE performs periodic reporting according to this period.
  • the CC may also inform a reporting trigger event through the third information, and the reporting trigger event may include one or more events. When any of the events occurs, the UE performs reporting once.
  • the reporting trigger event may also include the appointed event, and the reporting trigger event may also be described as a reporting trigger condition. When the event occurs or the reporting trigger condition is satisfied, the UE needs to perform reporting.
  • reporting trigger event may include an event related to at least one of the following:
  • the change of the second AI model the mobility of the UE; the traffic information of the UE; the RSRP measured by the UE; the change of the CSI; the change of the second information; a change of the first AI model; or a change of the UE distribution situation in the spatial position information associated with the first AI model.
  • the name of each event will not be limited in the embodiment of the present disclosure, and the event may also be explained in the form of a paragraph of text.
  • the "change" in the above items may also be referred to as "update”.
  • the change may mean the difference.
  • the change of the second AI model means that the second AI model used currently is different from the second AI model used last time
  • the change of the second information may mean that at least one item in the second information is different from the second information reported last time.
  • the change may also mean that the amount or degree of change reaches a certain degree.
  • the change of the CSI may mean that at least one item of information in the CSI is different from the corresponding information in the CSI measured last time, or the CSI of at least one AP is different from the CSI measured last time, or the CSI of no less than a first number of APs is different from the CSI measured last time, or the amount of change of the CSI of at least one AP is greater than a threshold.
  • the change of the first AI model indicates that the at least one AI model received by the UE is changed.
  • the UE needs to select the second AI model to be used from the least one AI model received recently, and obtain the second information through the second AI model.
  • the change of the UE distribution situation in the spatial position associated with the AI model refers to a change of the UE distribution situation associated with the at least one AI model received by the UE, such as a change of a UE distribution index associated with an AI model (the distribution index reflects the UE distribution situation associated with the AI model (such as UE distribution density), and may also be used to identify other UE's channel state information in the spatial position associated with the AI model).
  • the UE may determine the second information based on the UE related information and the updated UE distribution situation.
  • the event related to the mobility of the UE, the event related to the traffic information of the UE and the event related to the RSRP measured by the UE may also mean that the amount of change of these information of the UE exceeds the corresponding threshold or the parameter value of these information satisfies a certain condition.
  • the event related to the mobility of the UE includes at least one of the following:
  • the amount of change of the speed of the UE is greater than or equal to a first threshold; the speed of the UE is greater than or equal to a second threshold; and, the speed of the UE is less than or equal to a third threshold.
  • the event related to the traffic information of the UE includes at least one of the following:
  • the traffic type of the UE is changed; the amount of change of the traffic amount of the UE is greater than or equal to a fourth threshold; the traffic amount of the UE is greater than or equal to a fifth threshold; and, the traffic amount of the UE is less than or equal to a sixth threshold.
  • the event related to the PSRP measured by the UE includes at least one of the following:
  • the amount of change of the RSRP measured by the UE is greater than or equal to a seventh threshold; the RSRP measured by the UE is greater than or equal to an eighth threshold; and, the RSRP measured by the UE is less than or equal to a ninth threshold.
  • the specific values of the above thresholds may be appointed in advance or may be configured for the UE by the CC.
  • the amount of change may mean the amount of change of the current parameter amount (e.g., speed, received signal strength and traffic amount) of the UE relative to the parameter amount of the UE at the time of last reporting.
  • the RSRP measured by the UE may be used as a representation of the signal strength of the reference signal received or measured by the UE.
  • the RSRP measured by the UE may include the RSRP of at least one reference signal measured by the UE (for example, the RSRP of at least one reference signal with RSRP ranked top measured by the UE), or the RSRP of reference signals associated with one or some specified APs.
  • the specified APs mean which APs can be informed to the UE by the CC.
  • this configuration may include the identifier of each reference signal reference, and this configuration may further include specified reference signal resources.
  • These specified reference signal resources may be used an implicit implementation of the above specified APs, because the reference signal resources are associated with APs.
  • the UE may determine whether to trigger reporting based on the RSRP of these specified reference signals.
  • reporting may be triggered by the RSRP satisfying the condition (for example, the amount of change of the RSRP is greater than or equal to the seventh threshold) in the RSRP of each reference signal, or reporting may be triggered when the amount of change of the RSRP of the reference signal associated with the specified AP or the amount of change of the RSRP of the specified reference signal satisfies the condition.
  • the condition for example, the amount of change of the RSRP is greater than or equal to the seventh threshold
  • the relationship among the first threshold, the second threshold and the third threshold will not be uniquely limited.
  • the second threshold may be greater than or equal to the third threshold.
  • reporting may be triggered when the speed of the UE is changed from a higher speed to less than or equal to the third threshold, or reporting may be triggered when the speed of the UE is changed from a lower speed to greater than or equal to the second threshold.
  • the second threshold is greater than the third threshold, reporting is triggered when the speed of the UE is greater than or equal to the second threshold or less than or equal to the third threshold, and the UE may not perform reporting when the speed of the UE is between the second threshold and the third threshold.
  • the relationship among the fourth threshold, the fifth threshold and the sixth threshold will not be uniquely limited in the embodiment of the present disclosure, and the relationship among the seventh threshold, the eighth threshold and the ninth threshold will not be uniquely limited in the embodiment of the present disclosure.
  • the fifth threshold may be greater than or equal to the sixth threshold.
  • the eighth threshold may be greater than or equal to the ninth threshold.
  • the system may support at least two reporting modes, and the UE may determine to adopt which reporting mode according to its own information and/or the configuration of the CC.
  • the system may also support one appointed reporting mode, for example, the first reporting mode or the second reporting mode described hereinafter.
  • the two reporting modes may be implemented separately or in combination.
  • the UE may determine to adopt the first reporting mode or the second reporting mode according to the first condition.
  • the UE may determine, by using the second AI model and based on the UE related information, the second information related to the access point used for the UE, and then report the second information.
  • the scheme that the UE determines the second information through the second AI model and then reports the second information may be called the first reporting mode.
  • the UE adopts the first reporting mode.
  • the first condition may include at least one of the following:
  • the speed of the UE is greater than or equal to the first threshold; and, fourth information has been received, the fourth information being used for indicating to report the second information related to the access point used for the UE.
  • the first reporting mode when the UE is a UE that is in a non-static state or moving at a high speed, the first reporting mode may be adopted. Or, when the UE has received the indication information (fourth information) indicative of adopting the first reporting mode transmitted by the CC, the first reporting mode is adopted.
  • the fourth information may include an indicator.
  • the value of the indicator is the first value
  • the UE adopts the first reporting mode; and, when the value of the indicator is the second value, other reporting modes (e.g., the second reporting mode described hereinafter) are adopted.
  • this method further includes: transmitting the environment information related to the UE (e.g., the speed of the UE).
  • the fourth information is in response to the environment information, or the fourth information is based on the environment information.
  • the UE may report its environment information to the CC periodically or according to the instruction of the CC, and the CC may determine, according to the environment information of the UE, to make the UE adopt which reporting mode. For example, the UE may transmit reporting mode request information which includes the environment information related to the UE; and, the UE receives reporting mode determination information which includes an indicator related to the reporting mode, and the UE may determine to adopt which reporting mode according to this indicator.
  • this method further includes: in a case where the first condition is not satisfied, reporting fifth information, the fifth information including identifier information of at least one access point and information related to CSI of the at least one access point.
  • the reporting mode for the fifth information may be called the second reporting mode.
  • the UE may report the related information of the CSI of some or all of the Aps in the serving region, wherein the related information of the CSI may be the CSI or the compressed CSI, for example, the information obtained after performing data dimension reduction on the CSI.
  • the at least one access point may be APs satisfying a predetermined condition.
  • This predetermined condition will not be limited in the embodiment of the present disclosure, and may be the appointed condition or may be configured to the UE by the CC.
  • the UE may select and report the CSI of one or more APs with a good channel condition. It is possible to determine, based on the channel information (e.g., channel matrix) obtained by the UE performing measurement based on the reference signal, whether the channel condition is good or not.
  • the channel information e.g., channel matrix
  • the at least one access point may include at least one of the following:
  • an access point whose associated channel matrix has a feature value greater than or equal to a threshold.
  • the feature value of the channel matrix may be obtained by matrix decomposition.
  • the feature value of the channel matrix may be obtained by performing a singular value decomposition (SVD) operation on the channel matrix.
  • SVD singular value decomposition
  • a larger feature value indicates a better channel condition.
  • the feature value may also be replaced with a singular value.
  • a plurality of feature values/singular values may be obtained by performing an SVD operation on the channel matrix of the channel between any AP and the UE.
  • the maximum feature value/singular value may be used as the feature value of this channel matrix, or the mean value of a plurality of feature values ranked top is used as the feature value, or the like.
  • N access points ranked top may be determined as the at least one access point according to the rule of ascending order, or APs associated with the feature values greater than the threshold may be determined as the at least one access point.
  • the method provided by the present disclosure may further include: for each access point in the at least one access point, fusing the identifier information of this access point with the information related to CSI of this access point to obtain sixth information; and
  • the fifth information includes the encoded result of each access point in the at least one access point.
  • the information related to the CSI of APs that is reported through the fifth information by the UE may be the CSI of some or all APs, or the compressed information of the CSI.
  • Selecting APs based on the predetermined condition is also a data compression process.
  • the compression process may also be to encode the information related to the CSI of APs by using an AI model, so that the amount of data is reduced by encoding.
  • the information related to the CSI of each AP may be subjected to self-attention encoding by an encoder, thereby realizing further compression of information.
  • the CC may obtain the information related to the CSI of the at least one AP by decoding.
  • the encoding may be encoding the identifier information and CSI of the AP.
  • the fusing the identifier information of this access point with the CSI of this access point includes:
  • the way of encoding the identifier information of the access point will not be limited in the embodiment of the present disclosure.
  • the identifier information of the access point is processed by using a specific algorithm (e.g., the appointed function) to obtain the encoded result of the identifier information, the encoded result of the identifier information is then fused with the information related to the CSI to obtain the sixth information, and the sixth information of each AP is encoded based on the self-attention mechanism to obtain the compressed information.
  • the identifier information of the AP may be represented by the position information of the AP.
  • the position information of the AP may be based on the antenna index and sub-band index of the AP.
  • unique indexes may be allocated to each antenna and each sub-band based on the number of antennas and the number of sub-bands of each AP.
  • a unique position identifier may be obtained based on the indexes of all antennas and the indexes of all sub-bands, and the position identifier may be encoded to obtain the position encoded result.
  • the communication system will be described by taking a cell-free communication network environment as an example.
  • This embodiment provides a scheme of performing cube division and determining a cube-model based on UE features.
  • the space where cellular communication is deployed may be divided into a number of cubes, and an AI model may be trained for each cube.
  • an AI model may be trained for each cube.
  • it may be implemented by a suitable AI model based on the environment where the UE is located and the measurement result obtained by performing reference signal measurement by the UE.
  • FIGS. 6a and 6b show flowcharts of the scheme provided in this embodiment.
  • the shown user is a UE
  • the base station side may be a combination of an access point and a central control unit.
  • the AI model may be generated on the base station side, and the base station side provides the generated AI model to the UE.
  • this scheme may include some or all of the following steps.
  • step 1 cube division is determined based on the multimodal data quantify saturation, and AI models are trained, respectively. This step corresponds to the small cube division step and the model generation step in FIG. 6b.
  • the propagation environment is complex, and the UE shows diversified features, including movement speed, being in the indoor or outdoor environment, altitude, traffic type, traffic amount, distribution density of the UE, distribution density of the AP, line-of-sight, non-line-of-sight or the like.
  • Quantitative analysis of these multimodal data can help the network side to aggregate the UE features in different regions, so that the communication resource scheduling is more region-specific. Different regions have different feature aggregation degrees. For example, UEs in a certain region generally have a low speed and a high traffic amount, and most of them are video services which are in a scattering environment. Therefore, for UE features in different regions, the matched training data may be selected for AI model training. This specific training method can improve the training efficiency of the AI model, and improve the accuracy of the output result of the model.
  • the serving region may be divided into several smaller regions (cubes), and the UEs with close position and high channel similarity have more similar features, which may make the AI model simpler to further reduce the complexity of the AI model, and may enable the AI model to have higher generalization ability.
  • the division of the cubes may be based on UE geographic position information (which may also be referred to as geographic position similarity information) and/or UE channel state information (which may also be referred to as channel similarity information), wherein the UE geographic position information may include but is not limited to a position of the UE, information about whether the UE is indoor/outdoor, etc., and the UE channel state information may include, but is not limited to such information as LOS/NLOS, RSRP, channel correlation, and the like.
  • UE geographic position information may include but is not limited to a position of the UE, information about whether the UE is indoor/outdoor, etc.
  • the UE channel state information may include, but is not limited to such information as LOS/NLOS, RSRP, channel correlation, and the like.
  • the UE channel state information and the UE geographic information may be used for clustering in the serving region.
  • the UEs having similar LOS/NLOS, RSRP, and channel correlation mean that they are in similar channel conditions, and similar geographic information means that the UEs are under a proximal scenario, and they may select the same AP subset. For example, UEs in an indoor building will be clustered into a cubic, and outdoor UEs may be considered as another cubic, because the indoor UEs may not be able to connect to the outdoor APs due to a link blockage, and UEs that are far away from each other will likely not connect to the same AP.
  • the UE channel state information and the UE geographic information of the sample UEs may be quantified and then used as clustering samples.
  • a clustering algorithm may be used to cluster these quantified samples into multiple classes.
  • an interval corresponding to each class in the serving region may be obtained based on the positions of all the UEs belonging to each class, to realize the division of the serving region.
  • the cubes may be further extended based on the RSRP and UE/AP density information after the initial division of the serving region based on the UE geographic position information and/or the UE channel information.
  • An optional scheme for the division and extension of the cubes will be described hereinafter in conjunction with specific examples.
  • the step 1 may include some or all of the following steps 1-1 to 1-3.
  • the multimodal data may be reported to the CC by each UE, or may be obtained by data simulation or testing.
  • step 1-1 the multimodal data quantify saturation is determined.
  • a central control unit may collect the UE related information reported by a plurality of UEs.
  • the collected information may include multimodal environment information of the UEs, and quantify it to obtain the multimodal data quantify saturation (MDQS).
  • MDQS multimodal data quantify saturation
  • Table 1a shows an example of the MDQS obtained by quantifying the multimodal environment information of UEs.
  • the illustrative description is given by taking the multi-environment data of four UEs as an example.
  • the multimodal environment data of a large number of UEs may be acquired/collected for quantification, so that the more accurate result of cube division is obtained by clustering a large number of quantified results of the environment data.
  • the multimodal environment information may include at least one of the geographic position similarity information and the channel similarity information.
  • the geographic position similarity information includes position information, indoor information and outdoor information.
  • the position information may be coordinate information of a three-dimensional coordinate system, or may be longitude, latitude and altitude information.
  • the channel similarity information includes at least one of line-of-sight path/non-line-of-sight path information, RSRP, and information on the channel correlation between the UEs. The higher the channel correlation between the UEs is, the higher the channel similarity between the UEs is.
  • the channel correlation between the UEs which corresponds to one UE may be calculated by the CC based on the sample data.
  • the CC may calculate the channel correlation between the UEs based on the CSI reported by each UE in the serving region.
  • the CC may calculate channel correlation 1 between the UE1 and UE2, and channel correlation 2 between the UE1 and UE3, and may take the mean of the channel correlation 1 and the channel correlation 2 as the channel correlation corresponding to the UE1.
  • it may also calculate the channel correlation between one or more UEs that are close to the UE1 and the UE1.
  • the specific quantification mode and the granularity of quantification for each piece of data in different multimodal environment data will not be limited in the embodiment of the present disclosure, and Table 1a and Table 1b only shows an optional example.
  • the traffic amount may be classified into two classes by using one threshold, or the traffic amount may be subdivided into more classes by using a plurality of thresholds.
  • the traffic amount may be classified into four classes, i.e., high, medium, low and very low, by using three thresholds.
  • the quantification operation may be executed by a UE or an AP, in addition to the CC.
  • the CC may provide a quantification rule to each UE through APs, and the UEs may quantify the multimodal environment data according to this rule and then provide the quantified multimodal data to the CC through APs.
  • UEs may transmit the multimodal environment data to APs, and the APs quantify the multimodal environment data according to the quantification rule obtained from the CC and then transmit it to the CC.
  • step 1-2 cube division is determined.
  • the central control unit may cluster the multimodal data by using a clustering algorithm of machine learning to obtain several different clustering results, and different clustering results correspond to different cubes.
  • the specific clustering algorithm will not be limited in the embodiment of the present disclosure.
  • the UE samples may be clustered based on the geographic position similarity information and the channel similarity information in the table 2.
  • a K-means clustering algorithm may be selected to divide the UE samples into several classes.
  • the UE samples are divided into four classes: indoor UE class 0, outdoor UE class 1, outdoor UE class 2 and indoor UE class 3.
  • the serving region is divided based on the geographic position similarity information and the channel similarity information, and each serving region is trained by AI models independently, which may reduce the complexity of AI models, the complexity of UE processing calculation, and overhead of transferring the models.
  • an initial boundary of the cube is determined for each UE class based on the UE position information.
  • the cube boundary may be determined using a regular shape, such as a rectangle or circular shape, and so on.
  • FIG. 7b shows an optional example of using the rectangular shape to cover the UE samples of each UE class with the smallest rectangular area, based on the position information of the UEs divided into each UE class, to obtain the initial boundary of the cube.
  • a rectangular coordinate system may be established for the serving space.
  • a position center may be calculated based on the position information of all the UE samples in the class.
  • the positions of the farthest UEs of the horizontal axis and the vertical axis may be determined respectively, so as to determine a side length and boundary of the rectangle.
  • the initial boundary of the cube is a boundary of the smallest bounding rectangle that may include the positions of all the UEs in that class.
  • the cube adopts the rectangular shape, so that the information of the cube may be stored and transmitted with less bit overhead, but only the side length information and the center position information are needed.
  • the regular rectangular shape is convenient for the extension and adjustment of subsequent cubes.
  • other irregular graphics may also be used.
  • the cube is a 3D space to consider some serving regions of tall buildings.
  • a vertical view of the cube is used in the embodiments of the present disclosure to introduce a division operation of the cube.
  • the cube may be extended based on RSRP of an edge UE in the cube. Because there may be some APs in an outer layer of the initial cube, these APs have better RSRP performance. Therefore, in order to prevent UEs in the cube from being not capable of selecting these APs with high RSRP, it is necessary to extend the cube to include these APs with high RSRP.
  • a UE whose distance from the boundary of the cube is less than a certain threshold may be determined as an edge UE of the cube.
  • an AP that meets a certain RSRP threshold (for example, greater than the RSRP threshold) may be screened out from each AP based on RSRP of each AP to form an AP set that meets a high RSRP requirement, and an AP that is in the outer layer of the cube within this set is determined.
  • the RSRPs of all APs for the UE may be calculated, and the APs whose RSRP meets a certain condition may be screened out from all the APs to form an AP set corresponding to the UE.
  • the RSRPs of some APs (such as APs around the UE in the vicinity of the UE) for the UE may also be calculated.
  • the APs whose RSRP meets a certain condition are screened out from said some APs to form an AP set corresponding to the UE.
  • the APs that meet the condition may be but not is limited to APs whose corresponding RSRPs are greater than a certain threshold, or one or more APs ranked top are selected by the RSRPs in descending order.
  • the rectangle in the figure represents 1 initial cube
  • UE1 is a spatial edge UE in the cube.
  • these several APs are APs that meet the high RSRP requirement of the UE1, the APs located outside the cube among these several APs may be used as the AP set corresponding to the UE1.
  • a corresponding AP set may be obtained in a same way for other edge UEs in the initial cube, as shown in a legend on the leftmost side in FIG. 7c.
  • the initial cube After determining the APs of the outer layer of the cube that meets the high RSRP requirement, the initial cube is extended.
  • a linear distance between the outer layer APs and the position center (geographic centroid) of the cube on the horizontal and longitudinal axes is calculated respectively, and the maximum horizontal and longitudinal axis distances are selected to extend the initial cube based on the maximum distances, as shown in FIG. 7d.
  • the horizontal distance between the center of the cube and each AP and the vertical distance between the center position and each AP may be calculated respectively to determine the maximum horizontal and longitudinal distances as a distance between the center position of the extended cube and two sides of the extended cube, the initial cube is extended, and the extended cube is obtained.
  • the position coordinates of each edge may be determined based on the position center (geographic centroid) and side length of the cube, and then according to a position relationship of the sides or a relationship between the sides and the boundary, it is possible to determine whether there is a corresponding blank region on each side.
  • each cube (rectangle) in the figure represents an extended cube.
  • a cube i as an example, assuming that the coordinate of the center position of the cube i is ( , ), and the side length is ( , ), where denotes a horizontal side length, and denotes a longitudinal side length.
  • the positions of the four sides of the cube i may be denoted as follows:
  • the blank region in the serving region may be determined according to a position relationship between the sides of different cubs or a position relationship between the sides of the cube and the boundary of the serving region. As shown in FIG. 7e, there is a blank region between the lower boundary of the cube i and the upper boundary of the cube j, and there is also a blank region between the left side of the cube i and the boundary of the serving region.
  • the AP density will affect the service quality of UEs in the cube due to the difference of AP distribution in a gap region. But for two sides of the two cubes stretched in the same space, a filling ratio of each side needs to be determined.
  • the ratio of each extended space is calculated based on the density of APs and the density of UEs.
  • the density of UEs refers to the density of UEs in the cube.
  • the number of UEs in the cube may be used as the density of UEs.
  • the density of APs is the density of APs outside the cube.
  • the density of APs is related to the sides of the blank region adjacent to the cube. For a side of the blank region adjacent to the cube, the density of APs may be understood as the number of APs in a unit space or unit area located outside this side in the cube.
  • principles for filling the blank region may comprise: more APs may be included in the filled cube to serve the UEs, and if too many UEs are included in one cube, resource contention will increase.
  • a density ratio of the APs and the UEs is calculated, where denotes the density of APs and denotes the density of UEs, and the S is called an extension speed.
  • the length of the unit space be , which denotes the length of side extension of the cube.
  • the value of the length of the unit space may also be different for different application requirements or environments.
  • a side on the right of this cube is a side adjacent to the blank region.
  • An orange rectangle region and a purple rectangle region as shown in the example of FIG. 7f are two unit spaces, and the side lengths of two opposite sides of the rectangle regions are the lengths of the unit spaces , and the side lengths of other two opposite sides are the side lengths of sides adjacent to the blank region.
  • the extension ratio of the side i may be calculated by ( )/L
  • the extension ratio of the side j may be calculated by ( )/L
  • the extension ratio of the side j is ( + )/L
  • FIG. 7g there is a blank region between the cube a and the cube b, and sides i and j are sides adjacent to the blank region (which may be called sides of the blank region or blank sides), where .
  • the side i may first extend outward by one unit space, and the length of one side of a region where the side j extends outward is .
  • the extension speed corresponding to the side i changes from to and the extension speed corresponding to the side j is still without change, and the extension length of the remaining blank region to be filled is , as shown in Fig. 7g (c).
  • the extension process terminates when any of the following conditions is satisfied:
  • the lower boundary of the cube c in which the side n is extended may be extended outward, and the upper boundary of the cube a may be extended upward.
  • the side j By extending, the side j will be located in the extended cube c, as shown in FIG. 7h, and the above condition 1) is satisfied, at this time the extension of boundary j may be stopped.
  • the side m by extending the side m outward, a part of the extended side m may be located in the extended cube a, and a part of the extended side m intersects with the extended side k. If the above condition a is satisfied, the extension of the side m may be stopped.
  • FIG. 7i shows an example diagram of cube division.
  • the server space provided for the UE is divided into four cubes, i.e., cube 1 to cube 4.
  • the spatial position information of each cube is shown by one rectangle in FIG. 7i, and different cubes correspond to multimodal environment data with different features. It is to be noted that the division of cubes may only be a logical division, and the spatial positions of different cubes may not be or may be overlapped.
  • the spatial position of the cube is the spatial range occupied by the cube, or referred to as the coverage range of the cube.
  • an AI model corresponding to each cube may be obtained by training based on the training data corresponding to each cube.
  • step 1-3 the training data of each cube is determined, and AI model training is performed based on the training data of each cube to obtain an AI model corresponding to each cube.
  • the above at least one first AI model is at least one of the plurality of AI models corresponding to the cubes.
  • the training data set corresponding to each cube may be selected from candidate data sets based on the clustered features of the current cube. For example, if the features of the current cube show the features of low speed, line of sight, most video services and high traffic amount, the selected training data set should also satisfy these features.
  • the way of acquiring the training data set will not be limited in the embodiment of the present disclosure.
  • the candidate data sets may come from actual measurement data, or from simulation data, or from a public data set. After the candidate data sets are obtained, for each cube, according to the features of this cube, the data conforming to the features may be selected from the candidate data sets to serve as the training data set corresponding to this cube.
  • AI model training may be performed by using the training data set of this cube to obtain the AI model corresponding to this cube.
  • the trained AI model may also be referred to as a cube-model.
  • the specific model structure of the AI model will not be limited in the embodiment of the present disclosure.
  • the CC may train the AI model based on the respective training data set of each cube to obtain the respective cube-model of each cube. It is to be noted that, in practical applications, the training of the model may also be executed by other electronic devices; and, after each trained AI model is obtained, each AI model is provided to the CC. Optionally, after the model is trained, the CC may also continuously collect the multimodal environment reported by the UE. When the amount of the collected data reaches a certain amount, the CC or other electronic devices may update each AI model based on these data.
  • the input of the cube-model may be the feature data (UE related information) of the UE, and the output may be the output value corresponding to the feature data.
  • This output value may be the identifier of the AP, or a value that has a correspondence with at least one AP.
  • the output of the cube-model may be a code (which may be called cube code), and different codes may be mapped to different AP selection strategies.
  • the mapping relationship may be predefined and maintained by the CC.
  • each piece of training data in the training data set may include the feature data and tag of one UE, wherein the feature data of the UE may include the CSI based on each reference signal in a plurality of reference signals, and may optionally include one or more of the multimodal environment data of the UE.
  • the multimodal environment data may be unquantified or quantified data.
  • the tag of the training data represents the real code corresponding to the feature data of the UE in the training data.
  • the input to the cube-model also includes a UE distribution information number (also referred to as distribution number, distribution index, cube distribution information, cube number, cube distribution index, or other name), where the number is used to identify the cubes, each of which corresponds to its own unique number.
  • the number is used to identify one AI model.
  • the distribution index of each cube may be determined by a CC (also referred to as base station side or network side), and the meaning of the distribution index may be known or unknown to the UE.
  • the UE may only know which AI model correspond to which number.
  • the UE distribution number may be obtained based on the UE distribution information.
  • the UE distribution information number for each cube may be defined as follows.
  • each cube may be divided into multiple small squares, and the UE distribution density under each square is marked.
  • each small square may be an area of 1 square meter, assuming that there are K UEs (sample UEs) in the cube, the cube is divided into M squares, and ki UEs exist in the ith square, then the UE density of the ith square is marked as ki/K.
  • a UE density distribution set ⁇ k1/K, k2/K, ..., KM/K ⁇ of the cube is obtained.
  • Different UE density distribution sets are assigned different numbers, which are called distribution numbers.
  • the UE distribution sets may have multiple different cases. All possible UE distributions may be divided into Q classes. When a correlation between multiple UE distribution sets is greater than a certain threshold, they may be divided into the same class. Naturally, UE distributions of the same density are also divided into the same class.
  • the Q classes are numbered respectively to obtain Q distribution numbers. As shown in FIG. 8b, two different types of UE distribution information may be numbered as 1 and 2 respectively, and the same distribution number represents a similar UE distribution set.
  • one distribution number may correspond to one or more cubes, where the value of Q is less than or equal to the number of cubes.
  • the base station may use the distribution numbers of different distributions as inputs of AI model training.
  • the AI model may learn a mapping relationship between the distribution numbers and actual distribution situations through training, and then transmit the current distribution numbers and the trained AI model to the UE, so that the UE may implicitly obtain a global interference situation, improve the accuracy of AP selection results, and avoid the leakage of privacy information such as UE positions.
  • the base station may generate the distribution numbers (distribution indexes) based on an actual UE distribution situation, and broadcast/multicast the distribution numbers to the UE within the coverage of the cube (such as the scope of the serving region).
  • the base station may periodically update the distribution numbers based on the UE distribution situation and broadcast the latest distribution numbers to the UE.
  • the base station may also update the distribution numbers when the distribution situation changes greatly, map different UE distribution information to different distribution numbers, and re-broadcast the updated distribution numbers.
  • the AI model may be trained online or offline.
  • training data sets of the cube may be built based on the corresponding CSI of each AP in the cube and the UE distribution in the cube.
  • the CC also referred to as the base station side or base station
  • the CC may obtain data sets for model training from a plurality of UEs within the range the serving region, for example, receiving full report data from each UE, which may include but not be limited to a CSI report, and then dividing the full report data into each cube according to the features of each cube, to get the training data for training the AI model respectively corresponding to each cube.
  • complete report data obtained by the CC is stored in a complete database.
  • the CSI report corresponding to each AP reported by each UE may be used to build training sets of each cube based on the complete report data and the data of each cube (such as respective index of the cube).
  • the ellipses in FIG. 8c denote information about other cubes that are not shown.
  • the input of the model may include the CSI of all APs in the cube and the distribution index of the cube, and the output of the model is the selection result of APs in the cube.
  • the feature data of the UE in each piece of training data of this cube may be input into the model, or the feature data of the UE in each piece of training data of this cube and the distribution numbers corresponding to this cube are input into the model, and the code corresponding to each piece of training data may be predicted by the model (It may represent an AP selection result, and different codes correspond to different AP selection results).
  • a training loss may be calculated based on the predicted code and tag corresponding to each piece of training data, the model parameter is adjusted based on the training loss, and the model is continuously trained based on the training data until an AI model satisfying a training end condition is obtained (for example, the loss function converges or the number of training iterations reaches the preset number, or the like).
  • the output of the AI model may be a probability vector, and each probability in this vector is the probability that the code corresponding to the training data is each code.
  • the tag of the training data may be a real probability vector. The probability that there is one code in this vector is 0, while the probabilities of other codes are 0.
  • the training loss of this training data may be obtained by calculating the difference between the probability vector output by the model and the real probability vector.
  • the feature data of one UE may be input into the model to obtain one probability vector; the code corresponding to the maximum probability in the probability vector is used as an output code of the model; and, an AP corresponding to this UE may be determined based on the mapping relationship between this output code and APs.
  • the AI model may be obtained by the scheme provided in the step 1 in the embodiment of the present disclosure, or may be obtained in other ways.
  • One AI model may correspond to one cube or may correspond to the UE feature, e.g., the position of the UE, the speed of the UE, the UE being located indoor or outdoor or the like.
  • it may be unknown for the UE how each AI model is trained and whether each AI model is associated with one cube. Of course, the UE may also know them.
  • the step 1 is a preparation. How to obtain a plurality of AI models and the specific information related to each AI model will not be limited in the embodiment of the present disclosure. The way of obtaining, by the CC, the data used for cube division in the step 1 will not be limited in the present disclosure.
  • the plurality of UEs here are general UEs.
  • the UE in the following step 2 may be or may not be one of the plurality of UEs.
  • the plurality of UEs may also be UEs specifically configured for the collection of the multimodal environment data.
  • step 2 a cube-model is selected and a serving AP is determined.
  • the CC may transmit a suitable cube-model to the UE through an AP.
  • the UE generates a code according to the measured feature result (corresponding to the feature data of the UE) and the cube-model to match the dynamic and diverse scenario, wherein the code generated by the model may also be referred to as the cube-model-generated code.
  • the CC may determine a serving AP for the UE according to the cub-model-generated code.
  • the step 2 may include the following steps 2-1 and 2-4.
  • step 2-1 a cube-model is selected.
  • the CC may select or determine cube-model information to be transmitted to each UE.
  • the cube-model information transmitted to the UE by the CC may be one AI model or one group of AI models and related information. This step corresponds to the selection of one or more candidate cubes for UEs in the cube on the base station side in FIG. 6b, and the model information of each AI model corresponding to these cubes (a list of models in the following) may be transmitted to the UE.
  • the UE may select one or more AI models from the received models as the AI models to be used. Based on the UE related information, the selected models may be adopted to obtain the AP selection result.
  • the CC may transmit a cube-model list.
  • This list may include multiple pieces of cube information and corresponding cube-models.
  • the cube information may include spatial position information or spatial range information covered by the cube associated with each model.
  • the UE may detect which cube its position belongs to and use the corresponding cube-model.
  • the CC may transmit one cube-model list including a plurality of cube-models and the timing information of different cube-models, e.g., effective time and/or expiration time, etc.
  • the UE may trigger the corresponding timer, and use different cube-models according to different timing information.
  • the timing information may be the effective time.
  • an AI model currently in the effective stage is selected according to the current time.
  • the effective information may include the effective time and the effective duration. The UE may determine, according to the effective time and the effective duration, whether one AI model is available or failed.
  • the CC may transmit an associated cube-model list, which may include information about the cubes that the UE may re-select and/or adjacent cubes to which the UE may switch.
  • the associated cube-model list may include cube-model information (for example, at least one cube-model), cube geographic position information associated with each model, and optionally, the list may also include all UE channel state information obtained by the base station (UE channel state information associated with the model). For example, an average of the channel state information of all sample UEs associated with each model in the list (the channel state information of one UE may include channel attribute values of multiple dimensions.
  • the CC may average the attribute values of each UE in the obtained cube in this dimension, and the average of each dimension is taken as the UE channel state information of the cube) may be referred to as a sample centroid of the channel state information.
  • the specific contents of the cube geographic position information associated with one model are not uniquely limited in the embodiments of the present disclosure, as long as the information of cube positions can be obtained theoretically, such as, for example, information on the central position (geographic position centroid) and side length of the cube, or the position of each vertex of the cube, or the central position of the cube and the vertex position of at least one orientation.
  • the distribution index corresponding to each cube-model in the associated cube list may also be transmitted to the UE.
  • the overhead may be effectively reduced by mapping the UE distribution information to the distribution indexes.
  • the CC may determine the associated cube-model list of each cube based on switching and re-selection between the cubes, determine the list for UEs in each cube, and determine an initial associated cube based on the cube where the UEs are located.
  • the current cube, as well as the corresponding cube-model information, and the geographic position information of the cube may be first used as an initial associated cube-model list. Whether to add other cube information to the initial associated cube list is determined based on one or more of the following conditions: when a UE of the current cube is at an edge of the cube, adjacent other cube information is added to the associated cube-model list. In the example shown in FIG.
  • the initial associated cube-model list of cube1 is the cube1. Since UE2 in the cube1 is at an edge of the cube and adjacent to cube2, then, the relevant information of the cube2 is added to the associated cube-model list of the cube1, including cube-model information, cube geospatial position information and side length information of the cube2.
  • the model information associated with other adjacent cubes is added to the associated cube-model list.
  • the model learning associated with other cubes within a certain range around the center of the current cube may be added to the list, and the certain range may be a preset range or a range determined by other ways.
  • the other surrounding adjacent cubes may be determined based on the speed of the UE.
  • the cubes in a circular coverage area with the center of the current cube as the center of a circle and the radius being the speed multiplied by the duration of a period is added to the list, where the duration of the period may be a preset value, such as 1 second, or may be a set multiple of the period of broadcast signal SSB (Synchronization Signal and PBCH block), and the UE's speed may be a UE speed estimated by the base station side.
  • SSB Synchronization Signal and PBCH block
  • the cubes added to the list of the cube1 are cube2, cube3, cube4, cube5, cube6.
  • one or more cubes within the preset range with better corresponding channel intensity may be used as an associated cube, and relevant information of AI models corresponding to these cubes may be added to the list.
  • the above UE's speed may adopt the same speed, such as the average speed or the maximum speed of all UEs in the cube.
  • the list After determining the associated cube-model list of the cube, the list will be transmitted to all UEs in the coverage range of the cube.
  • a corresponding circular coverage area may also be determined for each UE based on its respective speed to obtain a list respectively corresponding to each UE.
  • the list corresponding to each UE may be transmitted to the corresponding UE, or a union of list corresponding to each UE in the cube may be used as the associated cube-model list associated with the cube.
  • the CC may transmit the list to each UE in the cube.
  • This scheme may enable the UE to select an appropriate AI model for AP selection according to its actual situation, thereby ensuring the system performance.
  • the CC may predict the position of the UE after a period of time (denoted by time N), and transmits, to the UE, the cube-model corresponding to the cube to which this position belongs.
  • the UE uses this cube-model.
  • the time N may be a time from the time when the CC transmits the cube-model or pilot to the time when the UE starts measurement. Since the pilot transmission period is large, the delay needs to be considered for users who are moving at a high speed.
  • the cube-model and the pilot signal may or may not be transmitted simultaneously.
  • the time N may also refer to a period of time from the time when the CC transmits the configuration information used for channel state information measurement to the UE to the time when the CC prepares to transmit the model to the UE, or a period of time from the time when the CC transmits the model to the UE last time to the time when the CC prepares to transmit the model to the UE this time.
  • the CC may transmit all models and the cube information corresponding to each model to the UE, and the UE may autonomously determine to use which model according to its position, or the CC may instruct or trigger the UE to use which model.
  • the CC may dynamically schedule the UE and transmit the effectiveness information of each model to the UE, and the UE may perform model selection according to the latest effectiveness information received from the CC. If there is a plurality of effective models currently, the UE may perform further screening according to its position.
  • the UE may determine which AI model to use on the terminal side from the associated cube-model list, based on at least one of the at least one AI model received, the distribution index associated with the model, the UE's own geographic position information, the measured channel state information (such as RSRP, etc.), and the UE processing capability information, such as the step of selecting a cube on the UE side as shown in FIG. 6b.
  • the UE may select a plurality of cube models, the number of cube models selected finally may be based on the computing capability of the UE.
  • UE may input a UE distribution index associated with the selected cube model and the CSI obtained by UE measurement into the selected AI model to obtain the AP selection result.
  • the UE may sort the received models in descending order based on a similarity between the UE channel state information (sample centroid of channel state information) of the cube associated with each model received from the CC and the channel state information of the UE, and select one or more models with the similarity ranked top as the AI models to be used by the UE (the second AI model).
  • the Euclidean distance between the channel state information of the current UE and the sample centroid of the cube associated with each AI model may be calculated. The Euclidean distance represents the similarity. The smaller the distance is, the greater the similarity is.
  • the received respective AI models may be sorted according to the distances from small to large, to obtain a list of AI models sorted according to the similarity of the channel state information.
  • the UE is located at overlap of two cubes.
  • the associated cube-mode list transmitted by the CC to the UE includes relevant information of AI models corresponding to the two cubes cube0 and the cube1, including the AI model, the position information of centroid of the channel state information of the cube associated with the AI model, the distribution index associated with the AI model, and the like.
  • the UE may calculate a distance dists0 between the UE and the cube0 (the distance may be a distance between the UE's position and the cube0's geographic centroid (central position), or a distance between the UE's channel state information and the centroid of the cube0's channel state information), and a distance dists1 between the channel state information measured by the UE and the centroid of the cube1. If dists0 > dists1, it means that the channel state information similarity between the UE and the cube0 is less than that between the UE and the cube1.
  • the AI models associated with the cube1 are ordered before the AI models associated with the cube0, and the UE will preferentially select the AI models associated with the cube1 when selecting the AI models to be used.
  • the channel attribute values of each dimension of the current UE may be taken as a vector
  • the UE channel state information of the cube associated with an AI model may be taken as a vector, to calculate the Euclidean distance between the two vectors.
  • the UE may compare its own geographic position information with the cube position information associated with each AI model contained in the list received from the CC side (such as the geographic position centroid and side length of the cube), to determine in which cube it is located and determine whether it is at an edge of the cube.
  • to determine whether the cube position is at an edge position may comprise: the UE calculates a distance between its own position and an edge of the cube. When the distance is lower than a certain threshold (the distance between the UE and any side of the cube is less than the distance threshold), the UE may be considered to be at the edge position of the cube, otherwise (the distance between the UE and each side of the cube is not less than the distance threshold), the UE may be considered to be in a center area of the cube.
  • the UE is located at an edge position of the cube.
  • the UE is in a central area of cube1. If the UE is located in the center area of the cube, the AI model corresponding to the cube covering the UE position is used for AP selection of the UE.
  • the model 1 corresponding to the cube1 may be used as a final model for use by the UE.
  • the UE determines which model in the associated cube list to use based on at least one of the UE position, channel state information of the UE, and the UE capability information.
  • the UE may determine a cube where the UE is located and an adjacent cube according to the geographic position information of the cube and its own real-time position information.
  • the AI model of the cube where the UE is located is an available candidate model.
  • the UE calculates its similarity with the adjacent cube, and if the similarity exceeds a certain threshold, the AI model corresponding to the adjacent cube is then added to the candidate models that the UE may use.
  • the UE calculates its similarity with the adjacent cube, which may be represented by calculating a distance between the UE channel state information and the position centroid (geographic position centroid) of the adjacent cube, or by calculating a distance between the channel state information of the UE and the sample centroid of channel state information of the adjacent cube. The smaller the distance is, the greater the similarity is.
  • the models corresponding to the adjacent cube whose similarity meets a condition are added to the candidate models.
  • the models corresponding to the adjacent cube whose similarity is greater than a threshold may be added to the candidate models, or the models corresponding to one or more similarities ranked top in the order from the largest to the smallest may be selected to be added to the candidate models.
  • the RSRPs corresponding to these adjacent cubes may be further compared to determine the models finally selected into the candidates for use.
  • the models of adjacent cubes corresponding to RSRPs that meet the conditions (such as RSRPs greater than a certain threshold or one or more RSRPs ranked top in the order of RSRPs from highest to lowest) may be added to the candidate models.
  • the UE may select one or more models from the candidate models to generate the AP selection result based on its processing capability.
  • the RSRPs corresponding to the adjacent cubes may be compared first, and the models of adjacent cubes whose corresponding RSRP meets a certain condition may be added to the candidate models.
  • the similarity between the UE and the plurality of adjacent cubes may be calculated again, and further screening may be conducted according to the corresponding similarity of each cube.
  • the UE is an edge UE, and adjacent cubes of the cube2 includes cube1 and cube3.
  • the distance 1 (such as Euclidean distance) between the channel state information of the UE and the centroid of the channel state information of the cube1, and the distance 3 between the channel state information of the UE and the centroid of the channel state information of the cube2 may be calculated respectively.
  • the UE is a UE with low processing capability, only one model (AI model of the cube where the UE is located) may be selected as the final model (the second model), and if the UE is a UE with high processing capability, a plurality of models may be selected as finally used models.
  • AI models of the cube1 and the cube3 may be added to the candidate models of the UE. If the distance 3 is less than the distance 1, the similarity between the UE and the cube3 is greater than that between the UE and the cube1, and the models of the cube where the UE is located and the models of the cube3 may be selected as the models finally used by the UE. If the distance 3 equals the distance 1, the models of the cube whose corresponding RSRP is higher may be used as the models used by the UE further by comparing the size of RSRP corresponding to the cube1 and the size of RSRP corresponding to the cube3. It is assumed that the RSRP corresponding to the cube1 is greater than that corresponding to the cube3. The model of the cube where the UE is located and the model of the cube1 may be used as the models finally used by the UE.
  • the RSRP corresponding to the cube may be transmitted by the CC to the UE, and the RSRP may be obtained by the CC based on RSRPs of all or part of the sample UEs corresponding to each AP in the cube (for example, the RSRP of each AP in the cube obtained by part or all of the sample UEs by measurement).
  • the CC may calculate the mean of the RSRPs reported by the UE to obtain the RSRP corresponding to the cube. Taking the cube1 in FIG. 8i as an example, the RSRP corresponding to the cube may be denoted as .
  • the "equal” or “same” provided in the embodiments of the present disclosure may mean absolutely equal or substantially equal. Taking the two distances as an example, if the two distances are equal, it may mean that the search between the two distances is less than a threshold.
  • the UE may determine the finally used AI models based on one or more of the information in the associated cube list received, the UE's own position information, the channel state information, and the UE capability information.
  • the UE may first compare similarities between a plurality of overlapping cubes, and if the similarity exceeds a certain threshold, the AI models of the corresponding cubes are added to the candidate models available to the UE.
  • the similarity with the adjacent cube calculated by the UE may be expressed by calculating a distance between the UE geographic position information and the geographic position centroid of the adjacent cube, or by calculating a distance between the UE channel state information and the centroid of the channel state information sample of the adjacent cube.
  • the RSRPs may be further compared to determine the models finally selected into the candidates for use.
  • the UE selects one or more models from the candidate models based on its own processing capability to generate the AP selection result.
  • a UE is located in an overlapping space of a plurality of cubes, and the distances between the UE and each cube may be calculated respectively, such as distance 1, distance 2 and distance 3 as shown in FIG. 8j.
  • the AI model of the corresponding cube with the smallest distance may be selected as the model used by the UE.
  • the RSRP measured by the UE may be used to determine whether there is a new model to be added to the candidate models.
  • the UE may obtain RSRPs of all measurable APs through measurement, and may compare the measured RSRPs with a certain threshold (such as a preset threshold). When the RSRPs meet the threshold requirement (such as greater than the certain threshold), the models of the cube where the AP corresponding to the RSRP is located are added to the candidate models.
  • a certain threshold such as a preset threshold
  • the UE is in a mobile state (such as high-speed mobile state)
  • the position of the UE is varied, and the selected AP may not be an optimal AP, so it is necessary to switch to a new cube and AP frequently.
  • the UE may construct a range (which may be referred to as a RSRP feature space) that satisfies the RSRP requirement (which may be referred to as the minimum RSRP requirement for the UE) based on the RSRP of the measurable AP.
  • the UE estimates its position and velocity vector within a certain period of time (and a preset duration) based on its own speed and moving direction, and estimates its moving trajectory within this certain period of time based on its position and velocity vector, where the moving trajectory is within the range of meeting the RSPR requirement.
  • the models corresponding to the cube through which the moving trajectory passes may be added to the candidate models, to calculate a channel state information similarity between the UE and each candidate model.
  • the UE selects one or more models from the candidate models based on its own processing capability to generate the AP selection result.
  • the UE in the mobile state selects APs whose RSRP (greater than the RSRP threshold) meets the requirement from all APs in which it can measure and obtain the RSRP, and obtains a range based on the positions of these APs that meet the requirement.
  • the closed range connected by the positions of these APs that meet the requirement may be taken as the range that meets the requirement of RSRP.
  • Each cube through which the UE's moving trajectory passes among all the cubes within the range may be taken as candidate cubes, and the AI models corresponding to these candidate cubes may be taken as the candidate models. In the example shown in FIG. 8k.
  • the cube within the range of meeting the requirement of RSRP include cubes 1 to 5, but the UE's moving trajectory does not pass through the cube2, so the candidate cubes are the cubes 1 and 3 to 5, and the models of these 4 cubes may be added to the candidate models of the UE.
  • the UE may select part of the 4 models, if the UE's capability is low, the UE may select one of the models as the models to be used, for example, to select one or more models based on the length of a line segment where the UE's moving trajectory falls on each cube. For example, the model corresponding to the cube5 will be used as the finally used AI model.
  • the above scheme may be used to transmit an associated cube list to the UE. All UEs in the cube may receive the associated cube list that is broadcast/multicast, where the list may include information about the AI models associated with the possible cubes to which the UE may move, switch, or load balance. The UE may select the AI models to be used according to one or more of the AI model, the distribution index, the UE position, the UE channel state information, and the UE capability information corresponding to the received cubes.
  • the embodiments of the present disclosure propose simplified UE distribution information (distribution index) to represent CSI information of other UEs in the cube, which may be used for the AI model. Since the detailed CSI information of all UEs will bring about heavy overhead, the CSI of the received channel depends on the position of the UE, the position of the UE may be used to indicate the CSI state, but the direct use of the position of the UE will bring privacy issues.
  • the UE distribution information (distribution number/index) provided in the embodiments of the present disclosure is used to indicate the UE position information, which may effectively reduce the overhead and in turn avoid privacy issues.
  • the distribution indexes provide an easier way to indicate the position information, and the distribution indexes may be used to represent information between the UEs, which will guarantee the performance of AP selection.
  • the scheme of dividing each cube into multiple small squares as shown in FIG. 8b above is adopted.
  • the UE distribution information only includes the number of UEs in the network. Different UE distributions are marked with different distribution indexes, and the corresponding distribution index of each cube may be used for the input of the AI model of the corresponding cube.
  • at least one AI model and the distribution index (the first information) of each model may be transmitted by the base station to the UE, and the UE may make the AP selection based on these information.
  • the input of the second AI model in addition to the CSI information measured by the UE, may also be used by the UE to utilize the distribution number of the AI model received as an input for the generation of AP selection result.
  • the UE may input the CSI of each AP measured and obtained by it and the corresponding distribution index of the second AI model into the second AI model.
  • the second AI model the AP selection is realized and the index or code of the selected AP for the UE is obtained.
  • the UE may use the AI model to calculate the service AP of the UE based on the received real-time channel state information of each AP in the cube associated with the AI model and the distribution index associated with the AI model. If the UE uses a plurality of AI models, the AP selection results of the plurality of models may be processed. The AP selection results obtained by each AI model may be combined and reported, or part of the AP selection results may be reported. For example, the AP selection results obtained by the AI models in the cube with good channel status received by the UE may be reported.
  • the UE may directly report the output of the AI model when reporting the AP selection results.
  • the output of the model may be a code (such as an AP index), and the code may be directly reported, and the CC may know which AP or APs the UE wants to select based on the code.
  • the UE may map the AP selection results to an AP index/bitmap based on a mapping relationship of the AP index/bitmap corresponding to the cube, and report a mapping result.
  • the mapping relationship may be a mapping relationship between the AP index/bitmap inside the cube and the model output of the cube, but not a global relationship, which may reduce the transmission load.
  • the mapping relationship corresponding to each cube may be transmitted by the CC to the UE, or may be agreed by the UE and the CC.
  • the table 2 below shows an example of an AP/bitmap mapping relationship in the cube, where column 1 in the table represents cube indexes (index of the cube associated with the AI model used).
  • the UE may look up the table to determine a bitmap corresponding to the corresponding AP index, and report the determined bitmap to the CC.
  • the output of the model is a bitmap.
  • the UE may determine the AP index corresponding to the bitmap based on this table and report it.
  • the UE uses the AI model corresponding to a cube index 1, and the output of the model is an AP index 2.
  • the UE may report the AP index 2 or the bitmap 010 to the CC, and the CC may know that the AP corresponding to the information reported by the UE is AP2 based on this mapping relationship.
  • the cube and the distribution situation of the UE may change.
  • the above cube-model selection and transmission process may be triggered when a certain condition is satisfied. For example, when the CC predicts, based on the historical position and speed information reported by the UE and through periodic prediction or monitoring (for example, the CC may use a support vector machine (SVR)), that the UE will move or move to the position of another cube, the cube-model selection and transmission process is executed. The CC may select the corresponding cube-model according to the cube where the UE is located or will be located, and then transmit it to the UE.
  • SVR support vector machine
  • the position information of the UE may be periodically predicted or monitored by the CC or may be periodically reported by the UE, or the latest position information may be reported to the CC by the UE when the change of the position information satisfies a certain condition.
  • the CC may also not execute the model selection and transmission process.
  • step 2-2 the cube-model-generated code is updated and reported.
  • the trigger mechanism for code reporting will not be uniquely limited in the embodiment of the present disclosure.
  • a periodic reporting mode may be adopted, or the reporting is triggered by the CC through a signaling.
  • the process of updating and reporting the cube-model-generated code by the UE may be triggered by a "period + event" double-trigger mechanism to adapt to the change of the user state.
  • the UE updates the selection of APs based on the configured period, the feature data of the UE, a change of the channel condition, AI model related information received by the UE, a distribution index of the UE, cube-model-generated code generated by the cube-model, and so on, and reports it to the CC.
  • the CC may configure a reporting period for the UE, and the UE performs periodic reporting according to this period.
  • the CC may also configure a reporting trigger event for the UE, and the UE performs reporting when the reporting trigger event occurs.
  • the reporting trigger event may also be appointed in advance.
  • FIG. 9b shows an optional update scheme for AP selection based on a dual trigger mechanism of periods and events according to an embodiment of the present disclosure.
  • the example shows four possible cases in which the UE may re-perform the AP selection when the corresponding conditions are satisfied.
  • the condition 1 is the arrival of a new period.
  • the UE may perform an AP selection operation (inputting the obtained CSI and the distribution index associated with the AI model selected for use into the corresponding AI model to obtain a model output) and update the AP selection result.
  • the condition 2 is that the UE receives a new distribution index, and the UE distribution index associated with each model changed, indicating that a UE distribution situation in each cube changed.
  • the UE may make the AP selection based on the newly obtained distribution index and update the AP selection result.
  • the condition 3 is that the RSRP of the cube associated with the selected AI model measured by the UE (such as the mean of the RSRP of each AP in the cube) is lower than a threshold, and the UE may re-select the AI models to be used, make the AP selection based on the re-selected AI model, and update the AP selection result.
  • the condition 4 is that the UE receives relevant information about the new cube model (such as a new associated cube-model list). The UE may select the AI models to be used from new cube models, make the AP selection based on the newly selected AI models, and update the AP selection result.
  • the AI models to be used needs to be re-selected and the AP selection is performed based on the newly selected AI models.
  • Periodic trigger and event trigger are described below in conjunction with an example.
  • the reporting period is T (e.g., 1 minute), and the process will be triggered at the beginning of each period T to timely match the changing user state.
  • the reporting may be triggered by an event.
  • the UE performs reporting.
  • the reporting process may be triggered when the feature data of the UE is changed significantly, so that the UE can be served by a suitable AP timely.
  • the feature data may include, but not limited to:
  • the used cube-model for example, the reporting being triggered after the used model is updated
  • the mobility of the UE for example, the reporting being triggered after the UE changes from the static state to the moving state;
  • the reporting being triggered by the sharp increase of the traffic amount (for example, the amount of change of the traffic amount exceeds a certain threshold);
  • the reporting being triggered when the RSRP measured by the UE is less than a certain threshold
  • channel information for example, the reporting being triggered when the change of one or more CSI satisfies a certain condition.
  • step 2-3 the UE updates and reports the cube-model-generated code.
  • the UE generated the cube-model-generated code (an implementation of the first value) based on the perceived feature information (for example, the above UE related information) and/or UE distribution index associated with the model by using the cube-model (the second AI model), and reports the cube-model-generated code to the CC.
  • this cube-model-generated code is different from the cube-model-generated code reported last time, this cube-model-generated code is reported to the CC.
  • the input of the second AI model may include at least one of the CSI (the measurement information of the pilot shown in the figure) obtained by the UE performing CSI measurement based on each pilot (reference signal) transmitted by the CC, the traffic information of the UE, the speed of the UE, the position of the UE and other information, and may further include the UE distribution index associated with the AI model, wherein the measurement information of the pilot may include, but not limited to, channel state information, received signal power (e.g., the RRSP of the received pilot) or the like, and the traffic information of the UE may include, but not limited to, the traffic type and/or traffic amount or the like.
  • each piece of input data may be subjected to data format conversion according to a certain rule and then input into the model, and calculated by the model to output a code.
  • step 2-4 the CC determines a serving AP.
  • the CC may determine a serving AP for the UE according to the cub-model-generated code reported by the UE.
  • the CC may directly select an AP having a corresponding predefined code-AP mapping relationship with the cube-model-generated code to serve as the serving AP of the UE.
  • the CC may finally determine the serving AP of the UE with reference to the cube-model-generated code reported by the UE and in combination with the density of UEs in the cube, the overall traffic amount and other factors. If the density of UEs and the overall traffic amount are large, less serving APs may be allocated for each UE.
  • the UE when reporting to the CC, may report the code generated using the AI model, or may report the identifier of at least one AP corresponding to this code to the CC.
  • the CC may preconfigure the mapping relationship between the code (the second value) and APs to each UE.
  • the UE may determine the identifiers of one or more APs corresponding to this code, and may transmit the determined identifiers of the APs to the CC.
  • the CC reported by the UE may be interpreted as that the UE expects the CC to transmit information to it through these APs.
  • the CC may determine, according to the APs reported by the UE or with reference to the APs reported by the UE, to use which APs to provide services for the UE.
  • the output of the AI model corresponding to each cube may also be the identifiers of APs.
  • the UE may process the feature data of the UE according to this rule and then output it into the selected AI model, so as to obtain the identifiers of APs through this AI model.
  • the output of the model may be an N-dimension probability vector. Each position in this vector corresponds to one AP.
  • the UE may report the identifiers of one or a preset number of positions with the highest probability or positions with probabilities greater than a threshold in this vector to the CC.
  • the code output by the AI model may be one of a plurality of candidate codes. If it is assumed that the access points deployed in the serving region of the cell-free system includes an access point 1 to an access point 5 and the served UEs include a user 1, a user 2 and a user 3, as shown in FIG. 11, the plurality of candidate codes include a code 1, code 2 and a code 3. Table 3 shows an example of the mapping relationship between codes and APs index.
  • the CC may determine one or more of the access point 1, the access point 2 and the access point 3 as the serving access point of this user; and, if it is assumed that the code reported by the user 2 is the code 2, the CC may determine the access point 2 and the access point 4 as the serving access points of this user.
  • This embodiment provides an AP selection scheme based on CSI compression feedback.
  • the AP selection scheme based on CSI compression feedback can be applied to users who are static/moving at a low speed, and suitable for scenarios with slow environment change and small channel information change. Since the channel environment is not changed much, the UE may feed back the CSI by using a long period configuration, so that the overhead of communication resources can be reduced.
  • the CC may determine whether to adopt the CSI compression feedback scheme according to the speed feature of the UE. For a static or low-speed (e.g., a UE with a speed less than or equal to the first threshold), it is determined to adopt the CSI compression feedback scheme.
  • a medium/high-speed UE e.g., a UE with a speed equal to or greater than the first threshold
  • the AP selection scheme in Embodiment 1 or other schemes may be adopted.
  • This embodiment provides a CSI compression scheme based on joint AP division (JADNet). This scheme can be applied to a D-MIMO system architecture.
  • the UE For any AP, the UE performs CSI measurement based on the reference signal.
  • the obtained CSI may be represented by a matrix.
  • This matrix may be referred to as a CSI matrix or channel matrix, which is the result of channel estimation between the AP and the UE.
  • the serving region of the D-MIMO system includes APs, K UEs are connected to the APs, and each AP has antennas and sub-bands.
  • the UE performs measurement based on the reference signal, and the estimated channel matrix between this AP and the UE may be represented by .
  • the combination of the channel matrices between all APs and UEs may be represented by , where .
  • the feature subset construction is proposed.
  • the CSI matrices of APs with good channel conditions can be aggregated, and different feature subsets correspond to different AP sets.
  • This embodiment of the present disclosure further provides an AP granularity based position encoding scheme to identify the position of the selected AP.
  • This embodiment of the present disclosure further provides a new AI model which applies the self-attention mechanism of the local AP.
  • This model uses the weak correlation characteristic between different APs in the D-MIMO to further reduce the complexity of the model.
  • the CSI compression feedback scheme provided in the embodiment of the present disclosure can effectively reduce the computing complexity and have good recovery accuracy.
  • the AI model which can realize CSI compression in this embodiment includes an encoder part and a decoder part, wherein the encoder part may be called a local AP block self-attention based (LABS) encoder, also referred to as an LABS encoder, and the corresponding decoder part may be called an LABS decoder.
  • LCS local AP block self-attention based
  • the feature subset construction may not be implemented, and the positions and CSI of all APs may be encoded, compressed, and fed back to the CC.
  • FIG. 12 shows a flowchart of an optional implementation of the scheme provided in this embodiment. As shown in FIG. 12, this scheme may include the following steps.
  • step 1 the CC determines whether to adopt the CSI compression scheme.
  • This step is an optional step.
  • the UE may report the information related to environment information to the CC (the environment information reporting shown in the figure).
  • this information may include the speed of the UE, and the CC may determine, based on the speed of the UE, whether the UE adopts the CSI compression feedback mode.
  • the CC determines that the UE adopts the CSI compression feedback mode, for example, optionally, in a case where the speed of the UE is less than or equal to a threshold.
  • the CC transmits indication information of using this mode to the UE, for example, the channel state information compression feedback acknowledgement shown in FIG. 12.
  • the UE may perform the CSI compress feedback scheme, i.e., the scheme provided in this embodiment.
  • the UE may perform the feedback scheme provided in the above Embodiment 1 of the present disclosure.
  • the UE may feed the code generated using the AI model based on the UE related information back to the CC.
  • step 2 the UE performs the CSI compression processing on the measured CSI by using an encoder, and reports the compressed CSI to the CC.
  • step 3 the CC decompresses the received compressed CSI to obtain the CSI of each AP, and the CC may determine the serving AP of the UE based on the CSI of APs.
  • the UE side may compress the CSI information based on the feature subset and AP position encoding, use a low-complexity LABS encoder to process the compressed data, and feed the compressed information to the central controller/central processor on the network side.
  • the central controller/central processor decodes the CSI by using an LABS decoder, and obtains CSI decoded information and corresponding AP positions through AP position decoding.
  • the network side obtains the CSI of some APs (for example, the channel information of APs, e.g., channel gain matrices, also referred to as channel matrices)
  • the CSI of all APs may be obtained through the CSI channel relationship among APs.
  • the CC may input the decoded CSI of some APs into the trained AP model, and this model may predict the CSI of all APs based on the CSI of some APs.
  • the central controller/central processor may determine the channel state from each AP to the UE and then select APs serving the UE according to the channel state, wherein there may be one or more selected APs.
  • FIG. 13 shows a principle diagram of a CSI compression mode based on an AI model according to an embodiment of the present disclosure.
  • the CSI compression and decompression process will be described below with reference FIG. 13.
  • the CSI compression process executed by the UE may include the following steps 1 to 3, and the CSI decompression process executed by the CC may include the following steps 4 to 6. It is to be noted that, in actual implementations, some steps are optional.
  • the step 1 is optional, and the CSI of all APs may be encoded and compressed by the steps 2 and 3.
  • the CC may obtain the CSI of all APs by decompression. The steps will be described below, respectively.
  • step 1 a feature subset of CSI channel information is constructed.
  • the UE side may construct a feature subset of CSI information, i.e., selecting the CSI of some APs or the compressed information of the CSI of some APs, for example, the channel matrices of APs or the feature matrices of the channel matrices.
  • the compression of the CSI information can be realized.
  • the CSI matrices of some APs include most information of the channel matrices in the D-MIMO.
  • the dimension of data is reduced by extracting the CSI matrices of some APs with good channels.
  • the feature subset is constructed based on the features of the channel matrices.
  • the feature subset construction process may be described as follows.
  • the UE performs singular value decomposition (SVD) on the received estimated CSI channel information of each AP, so as to realize the compression of the CSI channel information through the SVD operation.
  • SVD singular value decomposition
  • the estimated channel information of any AP obtained based on the reference signal by this UE may be represented as .
  • the feature value of the AP may be obtained.
  • the decompression expression of may be represented as:
  • matrices i.e., a left singular matrix and a right singular matrix, respectively, and or may be used as the feature matrix of after SVD.
  • a singular value or feature value is a singular value matrix.
  • the feature value of the AP with the maximum feature value may be extracted from the decompressed .
  • a higher feature value means that there is a good channel condition between the AP and the UE, so that the corresponding AP and its CSI matrix may be selected into the feature subset.
  • the APs with feature values greater than a threshold are selected as APs based by the feature subset construction.
  • the threshold may be configured to the UE by the CC, or may be appointed.
  • the UE may determine the threshold based on the maximum value in the feature values of all APs.
  • APs with feature values greater than are selected to report CSIs after SVD. All or some of the APs may be selected based on the threshold, and the CSI or CSI compressed information of these selected APs may be reported. When some of the APs are selected, the CSI of some of the APs is reported.
  • the set formed by the CSI channel matrices of the selected APs or the feature information (e.g., or ) of the channel matrices after SVD is used as the feature subset.
  • any AP has Nt antennas and Ns sub-bands, and the total number antennas in the M APs is .
  • the channel information of all sub-bands and all antennas of all selected APs are gathered.
  • the output of the feature subset construction step may be the channel matrices of the M APs, which may be represented as , where includes the CSI of all selected APs, e.g., channel matrices.
  • the matrix may also be called the CSI matrix of the selected APs.
  • step 2 AP granularity position encoding is performed.
  • the CC on the network side may decode the CSI or CSI compressed information of the known APs reported by the UE, so as to obtain the channel information of APs that do not report CSI according to the CSI of the known APs.
  • the AI model should know the identifiers of the selected APs, thereby helping the decoder to recover the CSI of these APs.
  • the UE when reporting the CSI or CSI compressed information of the APs, the UE needs to report the information of the corresponding APs in the feature subset, so as to enable the CC to know that the UE reports the CSI of which APs.
  • the UE side may encode the position of each selected AP.
  • the input of the position encoding is the identifier information of all APs in the feature subset, which may also be referred to as the position information, for example, the antenna global index value p of the APs.
  • the feature subset includes the channel information of M APs, any AP has Nt antennas and Ns sub-bands, and the total number of antennas in the M APs is .
  • the channel information of any AP may be represented as , and the -dimension information corresponds to one AP.
  • the position based encoding formula may be represented as:
  • a trigonometric function value can be obtained by substituting any antenna index and sub-band index into the above formula.
  • Each AP can obtain a group of values of the sin function and the cos function, and this group of values is the encoded results.
  • the encoded position information can be obtained by the above encoding method. This position information may be represented as one position encoding matrix .
  • the position encoding matrix of any AP is .
  • the antenna position of the selected AP may be input into the AP granularity based position encoding part to obtain the position encoding result of each AP, and the overall position encoding matrix may be obtained by integrating the position encoding matrices of all the selected APs.
  • the overall position encoding matrix is merged (e.g., added) with the feature subset to obtain .
  • the position coding result of each AP in the two matrices is merged with the CSI matrix.
  • The is used as the input of the LABS encoder.
  • step 3 LABS based CSI compression encoding is performed.
  • the D-MIMO system aims to serve a UE through a plurality of APs that are far away, so that the correlation between the channels of different APs is very weak.
  • the embodiment of the present disclosure provides a scheme which can further reduce the computing complexity based on a self-attention block based AI model.
  • the AI model used for CSI compression encoding may also be called an LABS model.
  • This model includes a compression module and a decompression module, wherein the compression module includes an LABS encoder, and the decompression module includes an LABS decoder.
  • the compression module may be deployed on the UE side, while the decompression module may be deployed on the CC side.
  • the input of the compression module may be the identifier information and CSI of the AP, for example, corresponding to each selected AP.
  • the compression module encodes to obtain the compressed information, the UE may transmit the compressed information to the CC through the AP, and the CC decompresses the received compressed information to obtain the recovered CSI matrix to obtain the CSI of each selected AP.
  • the AI model may be trained based on the training sample set by machine learning.
  • the way of acquiring/constructing samples in the training sample set will not be limited in the embodiment of the present disclosure.
  • each sample may include one sample .
  • decoding may be performed based on the compressed information by the decoder to obtain the recovered .
  • the training loss of this sample is obtained by using the difference (e.g., distance, 1 minus similarity) between and .
  • Each iteration capability may use a batch of samples, and the training loss of one iteration is obtained by calculating the sum or mean value of the training losses of multiple samples in this batch of samples.
  • the model may be continuously trained after the model parameter of the AI model is adjusted based on this training, until the AI model satisfying the condition is obtained.
  • a verification data set and a test data set may also be constructed.
  • the model is verified and tested by using the two data sets. If the performance index of the model cannot satisfy the set condition, the model may be continuously trained until the AI model satisfying the condition is obtained. Subsequently, this model may be applied in the CSI compression in practical application scenarios. For example, the compression part of the model is deployed in the UE, while the decompression part of the model is deployed in the CC.
  • the serving region of the D-MIMO system including APs, K UEs being connected to the APs and each AP having antennas and sub-bands as examples.
  • reference signals may be received from total antennas, and a CSI matrix may be calculated for sub-bands .
  • the channel estimation result (i.e., the CSI matrix) of any AP may be represented as
  • the CSI matrices of all APs may be represented as .
  • the corresponding to each of the K UEs may be obtained by simulation or by collecting data from the UE (the may be used as in the training stage).
  • the position encoding result of each AP in the APs may be obtained, the position encoding results of these APs are gathered to obtain the overall position encoding matrix in the training stage, and the corresponding to one UE may be combined with to obtain a training sample.
  • Each sample is input into the AI model to obtain the decoded matrix, and the training loss is obtained by calculating the difference between the input data and the output data of the model.
  • the input of the model may also not include , and the training loss may be obtained by calculating the difference (e.g., distance) between the input and the decoded by the model.
  • the encoder and the decoder of the AI model provided in the embodiment of the present disclosure may be an encoder-decoder model based on the Transformer structure. This model is designed and trained to minimize the distance between the and the CSI matrix decompressed by the model, so that the difference between the output of the model and the becomes smaller.
  • the LABS encoder can further reduce the complexity of CSI information processing through the local AP self-attention mechanism.
  • each AP corresponding to obtained by the above steps 1 and 2 may be used as a local AP, and each piece of AP information (e.g., the information obtained after merging the encoded position information and the channel information, the dimension of the information being ) in is used as an information block.
  • Each AP may be used as a local AP.
  • the channel information when each local AP transmits the measurement information is not overlapped with the channel information of other APs.
  • the UE receives the reference signal of the local AP and obtains the corresponding channel state information (e.g., channel matrix), the information represents the channel transmission characteristic of the local AP.
  • the channel state information e.g., channel matrix
  • the CSI channel matrix may be divided into non-overlapped channel information block sets by using the local AP as a standard.
  • the number of channel information blocks is equal to the number N of APs selected based on the feature subset construction, and the size of each information block is equal to the number of antennas of the AP.
  • the CSI channel matrix of each local AP is one channel information block, and self-attention calculation is performed on each channel information block.
  • the CSI channel matrix in each channel information block (this matrix integrates the channel matrix with the position encoding information of the AP) is used as the input of the self-attention module and processed by the self-attention module.
  • the processing dimension of the self-attention module can be effectively reduced, and the complexity of operation can be reduced.
  • the output information of all self-attention modules may be integrated into one overall channel information matrix for the subsequent conventional encoding operation, and the UE transmits the output of the LABS encoder to the CC on the network side.
  • FIG. 16 shows a principle diagram of the calculation of the LABS encoder.
  • This schematic diagram takes four information blocks as an example.
  • Each information block is the fused information of the CSI and the position encoding result of one AP, i.e., the data of one AP in .
  • the calculated results of the information blocks are merged.
  • the merged information may be processed by a full connector to obtain the CSI decompression result.
  • the full connector may include a residual layer (a residual connection shown in the figure), a feedforward layer and a residual layer which are cascaded.
  • the calculation amount can be effectively reduced.
  • step 4 LABS based decoding is performed.
  • the CC Upon receiving the channel encoding matrix information transmitted by the UE, the CC performs a conventional operation to obtain the integrated self-attention encoding channel information.
  • the self-attention decoding may be as follows: the decoded channel information is split into a plurality of blocks according to an information integration rule in the encoder, the corresponding channel information blocks are recovered by the self-attention mechanism, and all the channel information blocks are integrated to recover the CSI matrix .
  • step 5 AP position based decoding and CSI channel prediction are performed.
  • the CC obtains the output by the LABS decoder, then obtains the antenna index corresponding to the CSI information through a linear position decoder, and obtains the AP information corresponding to the CSI information according to the global index encoding rule.
  • step 6 the serving AP of the UE is selected based on the CSI.
  • the CC may determine an AP serving the UE based on the encoded channel information of each AP; or, the CC may predict the CSI channel information of all APs based on the decoded channel information of some APs or obtain the channel information of all APs according to the known channel information relationship among all APs, and determine an AP serving the UE according to the complete channel information.
  • the CC may determine, according to the CSI channel information reported (the reporting may be periodic reporting, or reporting triggered based on the event, or reporting when the UE receives the reporting instruction from the CC) by the UE, whether the channel state between each AP and the UE is changed sharply, and may reselect an AP serving the UE according to the CSI channel information.
  • the UE needs to report a large number of CSI, so that the processing complexity is high and the effectiveness of AP selection is affected. Therefore, this method may be deployed in a static/low-speed user scenario under the central controller/central processor-AP-UE architecture.
  • a CSI compression entity may be deployed on the UE side, while a CSI decompression entity may be deployed on the CC side.
  • the AP selection method based on CSI compression feedback LABS encoding/decoding may also be deployed in a static/low-speed user scenario under the gNB-UE architecture.
  • a CSI compression entity may be deployed on the UE side, while a CSI decompression entity may be deployed on the base station side.
  • a plurality of downlink beams of the base station may be used as a plurality of APs, and the feature subset may be the CSI or the feature information of CSI of a plurality of beams with good channel conditions in the plurality of downlink beams.
  • the base station may decode the compressed CSI fed back by the UE to obtain the CSI of the plurality of beams and may select the transmitting beam of the base station for the UE from these beams based on the CSI; or, the base station may predict the CSI of all beams according to the CSI of these beams, and then select the transmitting beams from all beams according to the CSI of all beams.
  • the scheme provided in the embodiment of the present disclosure may be based on one or more of the features of different users, the features of different regions, the channel features of different APs and other information, so that the accuracy and effectiveness of access point selection are improved in the case of less time-frequency resource overhead.
  • the present disclosure provides two optional solutions based on the features of UEs in the serving region of the communication system.
  • the first solution is as follows: the whole serving region is divided based on the user environment information, AI models are separately trained for each divided region on the network side, the trained model information is transmitted to the UE side, and the UE generates and feeds back AP selection information, thereby solving the problem of heavy overhead caused by CSI reporting.
  • the second solution is as follows: the CSI is compressed by the designed CSI compression module and fed back to the CC, and the CC decompresses the CSI information and determines the AP selection result.
  • the two solutions may be implemented separately or in combination.
  • different strategies may be selected based on the speed of the UE.
  • the second solution may be adopted. This is because the UE may use a long period configuration to feed back the CSI when the channel environment is not changed much, thereby reducing the overhead of communication resources.
  • the serving AP is determined by the first solution.
  • the effective effects of the CSI compression feedback scheme provided in the optional embodiments of the present disclosure may include at least one of the following.
  • the feature subset construction proposes the construction of the feature subset based on the channel condition between the AP and the UE.
  • the feature value of SVD may be used to indicate whether the feature of the channel condition is good.
  • An AP with a larger feature value has a better channel condition. Therefore, in the order from largest to smallest feature values, the feature values ranked top are selected as the feature subset, and the CSI corresponding to the selected APs may be used together as the input of the encoder.
  • the AP granularity-based position encoding is designed to generalize the network of different APs selected through feature subset construction, which can help the decoder to recover the CSI of the related APs and avoid confusing the identities of the APs.
  • the LABS based encoder-decoder divides the CSI of all APs (e.g., all APs selected based on the feature subset construction) into non-overlapped channel information blocks.
  • the size of each block is equal to the size of the channel matrix of the AP. Based on this AI model, the complexity of storage and AI models can be reduced.
  • the optional scheme in the above Embodiment 1 of the present disclosure can perform a multimodal data quantification operation on the UE feedback information, the network side measurement and other information, and can be used to cluster the service features in the serving region and divide the serving region into a plurality of different cubes.
  • the network side can determine different AI models for different cubes by using training data with different features, and then transmit them to the UE.
  • the CC transmits a cube-model list, and the UE may select a suitable AI model from the list according to the model related information received from the CC; or, the CC may transmit a cube-model to the UE, and the UE uses this cube-model.
  • the UE triggers the process of updating and reporting the cube-model-generated code through a "period + event" double-trigger mechanism.
  • the UE can use the local measurement information (e.g., the CSI measurement information obtained based on the received reference signal by the UE) as an input, then input it into the AI model to generate a code, and feed the code back to the CC. This scheme does not require the UE to feed back the CSI.
  • the optional scheme in the above Embodiment 2 of the present disclosure can be used directly; or, the CC or AP determines, according to the UE features (e.g., the speed feature of the UE), whether to use the CSI compression feedback scheme, and feed acknowledgement information back to the UE to inform the UE whether to use the CSI compression feedback scheme.
  • the UE features e.g., the speed feature of the UE
  • the UE can be instructed to use the scheme in Embodiment 1; while for a static or low-speed UE, the UE can be instructed to use the scheme in Embodiment 2.
  • a new CSI compression/decompression module is designed, which can greatly reduce complexity while ensuring accuracy in comparison to the conventional scheme.
  • the data dimension can be reduced by feature subset construction, the AP position information can be protected from missing in combination with the AP position encoding, and the complexity can be further reduced by using the self-attention mechanism of AP division.
  • the schemes provided in the embodiments of the present disclosure can be applied to, but not limited to, cell-free communication systems.
  • the APs are configured with a large number of antennas, and the serving region has a wide range.
  • the environment is complicated, and UEs in different regions have differentiated feature information.
  • the conventional CSI feedback scheme is adopted, a heavy communication overhead will be caused, and the system performance will be affected.
  • the feedback overhead can be reduced by using the scheme in the above Embodiment 1 of the present disclosure.
  • Embodiment 2 For a low-speed user, Embodiment 2 is adopted; while for a high-speed user, Embodiment 1 is adopted. Based on the schemes provided in the embodiments of the present disclosure, the accuracy of AP selection can be ensured, and the feedback overhead can be reduced.
  • the scheme provided in Embodiment 2 may also be adopted to ensure the CSI feedback accuracy and effectively reduce the feedback overhead.
  • the UE may transmit 1-bit indication information to the UE. If the value of this one bit is 1, the UE adopts the scheme in Embodiment 1; and, if the value of this one bit is 0, the UE adopts the scheme in Embodiment 2.
  • an embodiment of the present disclosure provides a method executed by a node in a wireless communication system, including steps of:
  • first information including model information related to at least one first artificial intelligence (AI) model used for access point selection; and
  • AI artificial intelligence
  • the second information being determined by a second AI model based on UE related information, the second AI model being one of the at least one first AI model, the UE related information including CSI.
  • the second information is determined by the second AI model, based on the UE related information and the information related to the UE distribution situation in the space associated with the second AI model.
  • the number of the at least one first AI model is at least two, and the first information further includes information related to selection of each first AI model.
  • the information related to selection of each first AI model includes at least one of the following:
  • the second information includes at least one of the following:
  • the first value being used for indicating the access point used for the UE
  • the identifier information of the access point used for the UE is obtained based on the first value and a first mapping relationship, the first mapping relationship includes at least one second value and the identifier information of the access point associated with each second value, and the first value is one of the at least one second value.
  • the second information is reported in a case where the first value is different from the first value reported last time, and/or the second information is reported in a case where the identifier information is different from the identifier information reported last time.
  • the UE related information further includes at least one of the following:
  • the method further includes: transmitting third information related to reporting; wherein the second information being reported based on the third information.
  • the third information includes information related to at least one of the following:
  • a reporting period a reporting period
  • a reporting trigger event a reporting trigger event
  • the reporting trigger event includes an event related to at least one of the following:
  • a change of the second AI model a mobility of the UE; traffic information of the UE; an RSRP measured by the UE; a change of the CSI; a change of the second information; a change of the first AI model; or a change of the UE distribution situation in the space associated with the first AI model.
  • the event related to the mobility of the UE includes at least one of the following:
  • an amount of change of the speed of the UE is greater than or equal to a first threshold; the speed of the UE is greater than or equal to a second threshold; and, the speed of the UE is less than or equal to a third threshold;
  • the event related to the traffic information of the UE includes at least one of the following:
  • the traffic type of the UE is changed; the amount of change of the traffic amount of the UE is greater than or equal to a fourth threshold; the traffic amount of the UE is greater than or equal to a fifth threshold; and, the traffic amount of the UE is less than or equal to a sixth threshold;
  • the event related to the RSRP measured by the UE includes at least one of the following:
  • the amount of change of the RSRP measured by the UE is greater than or equal to a seventh threshold; the RSRP measured by the UE is greater than or equal to an eighth threshold; and, the RSRP measured by the UE is less than or equal to a ninth threshold.
  • the second information is reported in a case where a first condition is satisfied, wherein the first condition includes at least one of the following:
  • the speed of the UE is greater than or equal to the first threshold
  • the UE has received fourth information, the fourth information being used for indicating to report the second information related to the access point used for the UE.
  • the method further includes: receiving fifth information, wherein the fifth information is reported in a case where the first condition is not satisfied, and the fifth information includes identifier information of at least one access point and information related to CSI of the at least one access point.
  • the at least one access point includes at least one of the following:
  • an access point whose associated channel matrix has a feature value greater than or equal to a threshold.
  • the fifth information includes an encoded result of each access point in the at least one access point
  • the encoded result of each access point is obtained by encoding sixth information of this access point by using a third AI model based on a self-attention mechanism
  • the sixth information is obtained by fusing the identifier information of this access point with the information related to CSI of this access point.
  • the sixth information is obtained by fusing the encoded identifier information of this access point with the information related to CSI of this access point.
  • the second AI model is determined from the at least one first AI model based on at least one of the following:
  • the method further comprises:
  • each fourth AI model transmitting seventh information associated with each fourth AI model, the seventh information being related to a UE distribution situation in a space associated with the fourth AI model, wherein said each fourth AI model comprises the at least one first AI model.
  • the at least one first AI model is at least some of AI models in a first model set, and the first model set comprises an AI model associated with each of a plurality of cubes of a serving region.
  • the plurality of cubes of the serving region are obtained by dividing the serving region.
  • the plurality of cubes of the serving region are obtained by dividing the serving region based on the UE related information of a plurality of sample UEs in the serving region.
  • the UE related information comprises at least one of the following:
  • the plurality of cubes of the serving region are determined based on the geographic position information of each sample UE belonging to each UE class,
  • each UE class is obtained by clustering the plurality of sample UEs based on the UE related information of the plurality of sample UEs.
  • the plurality of cubes of the serving region are obtained by extending a first cube associated with each UE class,
  • the first cube associated with said each UE class is determined based on the geographic position information of each sample UE belonging to each UE class.
  • the extending a first cube associated with each UE class is based on at least one of the following:
  • the first boundary of the first cube is extended to the second boundary of the serving region.
  • At least one of the two first cubes is extended to the blank region.
  • At least one of the two first cubes is extended to the blank region comprises:
  • extension speed associated with one of the boundaries is related to at least one of the following:
  • the extension speed associated with one of the boundaries is related to a ratio, the ratio being a ratio of the density of APs in the unit space extending to the blank region with the boundary being a starting point and the density of the sample UEs in the first cube to which the boundary belongs.
  • this node may be a CC.
  • the CC may transmit first information to a UE through an AP, and receive second information from the UE.
  • the scheme provided in the embodiments of the present disclosure may divide the serving region into smaller regions (cubes) with similar features, and the smaller regions with similar features may make the scale of the AI model smaller.
  • the UE with close position and high channel similarity has more similar features, which can make the AI model simpler and further reduce the complexity of the AI model, thus realizing high generalization ability.
  • the cube-based AI model may reduce the computing complexity on the base station side and the UE side. In the examples shown in FIGS.
  • a space where a plurality of UEs that have close positions, are in the same environment (such as indoor or outdoor), and have similar channel similarities (such as similar signal receive path information (such as LOS or NLOS), similar RSRP information, and similar channel correlation information) may be divided into one cube.
  • the AI model may be trained separately based on a training set corresponding to each cube (related data of the sample UE), and the AI model corresponding to each cube may be obtained, such as the AI model for the cube1 and the AI model for the cube2 as shown in FIG. 17b.
  • FIG. 18 shows a flowchart of a communication method according to an embodiment of the present disclosure.
  • the communication method may comprise the following steps:
  • Step A The UE collects UE related information.
  • Step B The UE reports the UE related information to a base station side.
  • the UE reports the UE related information may be triggered by the base station.
  • the UE reports the information when receiving a report instruction transmitted by the base station.
  • the UE related information may include but is not limited to UE signal receive path information (such as LOS/NLOS), RSRP information of the UE, and the like.
  • Step C The base station side may determine a cube range based on the UE related information reported by the UE.
  • the base station side may divide the serving region into cubes based on the collected relevant information of a large number of UEs (sample UEs) in the serving region.
  • the UE may calculate UE channel state information (for example, a channel correlation between the UEs) and estimate UE geographic position information based on the collected UE related information, while the CC may calculate UE/AP density information of an initially divided cube based on the collected UE related information.
  • the base station side may initially divide the cube based on the UE channel state information and the UE geographic position information, extend the initially divided cube based on the UE density and AP density, and fill a blank region between the extended cubes to obtain each divided cube.
  • the CC may periodically determine each cube.
  • Step D Train the cube models based on CSIs of all APs in the cubes and the UE distribution information.
  • the UE distribution information of each cube may be determined, such as a distribution index.
  • an AI model of the cube may be trained and obtained based on the CSIs of all APs in the cubes and the UE distribution information of the cubes.
  • the problems of high computing complexity and high AI model complexity existed in performing the AP selection may be solved.
  • Step E Determine an associated cube-model list based on switching and re-selection of the cubes.
  • Step F Determine the distribution index of the cube based on the UE distribution information in the cube.
  • Step G The base station side transmits the associated cube-model list to the UE side.
  • Step H The base station side broadcasts the UE distribution index.
  • Step I The UE selects one or more cube models to be used from the received cube models based on the received associated cube-model list, a position of the UE, the UE channel state information, and the UE capability information.
  • the UE may make model reselection based on a periodic trigger mechanism and/or an event trigger mechanism.
  • Step J The UE obtains the AP selection result through the selected AI model based on the UE distribution index associated with the selected AI model and the UE channel state information.
  • Step K The UE reports information about the selected APs (second information) to the CC.
  • the UE selects an appropriate AI model and reports the information related to the AP selection obtained based on the appropriate AI model to the CC to reduce uplink transmission overhead.
  • the schemes provided in the embodiments of the present disclosure may greatly reduce AP selection overhead without affecting the performance.
  • An embodiment of the present application further provides an electronic device, including at least one transceiver and at least one processor coupled to the at least one transceiver.
  • the at least one processor is configured to implement the method provided in any one of optional embodiments of the present application.
  • FIG. 19 shows a schematic structure diagram of an electronic device 4000 to which the solution of the embodiment of the present application is applied.
  • the electronic device 4000 shown in FIG. 17 may include a processor 4001 and a memory 4003.
  • the processor 4001 is connected to the memory 4003, for example, through a bus 4002.
  • the electronic device 4000 may further include a transceiver 4004. It should be noted that, in practical applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute any limitations to the embodiments of the present application.
  • the processor 4001 may be a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), or a field programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logical blocks, modules and circuits described in connection with the present disclosure.
  • the processor 4001 may also be a combination for realizing computing functions, for example, a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.
  • the bus 4002 may include a path to transfer information between the components described above.
  • the bus 4002 may be a peripheral component interconnect (PCI) bus, or an extended industry standard architecture (EISA) bus, etc.
  • the bus 4002 may be an address bus, a data bus, a control bus, etc.
  • the bus is represented by only one thick line in FIG. 19. However, it does not mean that there is only one bus or one type of buses.
  • the memory 4003 may be, but not limited to, read only memories (ROMs) or other types of static storage devices that can store static information and instructions, random access memories (RAMs) or other types of dynamic storage devices that can store information and instructions, may be electrically erasable programmable read only memories (EEPROMs), compact disc read only memories (CD-ROMs) or other optical disk storages, optical disc storages (including compact discs, laser discs, discs, digital versatile discs, blue-ray discs, etc.), magnetic storage media or other magnetic storage devices, or any other media that can carry or store desired program codes in the form of instructions or data structures and that can be accessed by computers.
  • ROMs read only memories
  • RAMs random access memories
  • EEPROMs electrically erasable programmable read only memories
  • CD-ROMs compact disc read only memories
  • optical disc storages including compact discs, laser discs, discs, digital versatile discs, blue-ray discs, etc.
  • magnetic storage media or other magnetic storage devices or any other media that can carry or
  • the memory 4003 is used to store application program codes for executing the solutions of the present application, and is controlled by the processor 4001.
  • the processor 4001 is used to execute the application program codes stored in the memory 4003 to implement the solution provided in any method embodiment described above.
  • Embodiments of the present disclosure provide a computer-readable storage medium having a computer program stored on the computer-readable storage medium, the computer program, when executed by a processor, implements the steps and corresponding contents of the foregoing method embodiments.
  • Embodiments of the present disclosure also provide a computer program product including a computer program, the computer program when executed by a processor realizing the steps and corresponding contents of the preceding method embodiments.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Computer Security & Cryptography (AREA)
  • Quality & Reliability (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

La divulgation concerne un système de communication 5G ou 6G destiné à prendre en charge un débit supérieur d'émission de données. La présente divulgation a trait à des systèmes de communication 5G ou 6G destinés à prendre en charge un débit de données plus élevé que des systèmes de communication 4G, tels qu'un système d'évolution à long terme (LTE). Des modes de réalisation de la présente divulgation concernent un procédé exécuté par un équipement utilisateur dans un système de communication sans fil, un procédé exécuté par un nœud, un équipement utilisateur et un nœud. Le procédé exécuté par un équipement utilisateur comprend les étapes consistant en : la réception de premières informations, les premières informations comprenant des informations de modèle relatives à au moins un premier modèle d'intelligence artificielle (IA) utilisé pour une sélection de point d'accès ; la détermination, par l'intermédiaire d'un second modèle d'IA et sur la base d'informations relatives à l'UE, de secondes informations relatives à un point d'accès utilisé pour l'UE ; et le rapport des secondes informations, le second modèle d'IA étant l'un du ou des premiers modèles d'IA, et les informations relatives à l'UE comprenant des informations d'état de canal (CSI). Sur la base des procédés fournis dans les modes de réalisation de la présente divulgation, les exigences de communication peuvent être mieux satisfaites.
PCT/KR2025/005719 2024-04-28 2025-04-28 Procédé exécuté par un équipement utilisateur dans un système de communication sans fil, procédé exécuté par un nœud, équipement utilisateur et nœud Pending WO2025230256A1 (fr)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN202410528468 2024-04-28
CN202410528468.1 2024-04-28
CN202411504186.4A CN120857171A (zh) 2024-04-28 2024-10-25 由无线通信系统中的用户设备执行的方法、由节点执行的方法、用户设备及节点
CN202411504186.4 2024-10-25

Publications (1)

Publication Number Publication Date
WO2025230256A1 true WO2025230256A1 (fr) 2025-11-06

Family

ID=97423950

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2025/005719 Pending WO2025230256A1 (fr) 2024-04-28 2025-04-28 Procédé exécuté par un équipement utilisateur dans un système de communication sans fil, procédé exécuté par un nœud, équipement utilisateur et nœud

Country Status (2)

Country Link
CN (1) CN120857171A (fr)
WO (1) WO2025230256A1 (fr)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100311435A1 (en) * 2009-06-08 2010-12-09 Infineon Technologies Ag Base station selecting devices and methods for establishing communication connections for radio communication terminal devices
US20190306736A1 (en) * 2017-04-26 2019-10-03 Verizon Patent And Licensing Inc. System and method for access point selection and scoring based on machine learning
WO2021126907A1 (fr) * 2019-12-16 2021-06-24 Qualcomm Incorporated Configuration de réseau neuronal pour assistance aux systèmes de communications sans fil
US20220394586A1 (en) * 2021-05-27 2022-12-08 GM Global Technology Operations LLC Predictive wi-fi data offloading systems and methods
US20230004864A1 (en) * 2019-10-28 2023-01-05 Google Llc End-to-End Machine-Learning for Wireless Networks

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100311435A1 (en) * 2009-06-08 2010-12-09 Infineon Technologies Ag Base station selecting devices and methods for establishing communication connections for radio communication terminal devices
US20190306736A1 (en) * 2017-04-26 2019-10-03 Verizon Patent And Licensing Inc. System and method for access point selection and scoring based on machine learning
US20230004864A1 (en) * 2019-10-28 2023-01-05 Google Llc End-to-End Machine-Learning for Wireless Networks
WO2021126907A1 (fr) * 2019-12-16 2021-06-24 Qualcomm Incorporated Configuration de réseau neuronal pour assistance aux systèmes de communications sans fil
US20220394586A1 (en) * 2021-05-27 2022-12-08 GM Global Technology Operations LLC Predictive wi-fi data offloading systems and methods

Also Published As

Publication number Publication date
CN120857171A (zh) 2025-10-28

Similar Documents

Publication Publication Date Title
WO2021215847A1 (fr) Architecture fonctionnelle et interface pour contrôleur intelligent ran en temps non réel
WO2022145953A1 (fr) Procédé et appareil de gestion de faisceau prédictif
WO2023287086A1 (fr) Procédé et dispositif d'émission ou de réception d'informations de faisceau dans un système de communication sans fil
WO2020091544A1 (fr) Procédé de rapport d'informations d'état de canal dans un système de communication sans fil, et dispositif associé
WO2022177323A1 (fr) Procédé et appareil pour mesurer et signaler des paramètres de communication dans un système de communication sans fil
WO2023068727A1 (fr) Procédé et dispositif pour installation automatique de ris
WO2023234741A1 (fr) Procédé et appareil de prise en charge d'apprentissage machine ou de rétroaction de csi assistée par intelligence artificielle
WO2023068686A1 (fr) Procédé et dispositif d'émission/réception de signal sans fil dans un système de communication sans fil
WO2023219388A1 (fr) Procédé et appareil d'indication et de mise à jour d'état de tci par signalisation dynamique
WO2023090762A1 (fr) Procédé et dispositif d'émission/réception d'un signal sans fil dans un système de communication sans fil
WO2023055111A1 (fr) Procédé et dispositif d'émission/réception d'un signal sans fil dans un système de communication sans fil
WO2023080658A1 (fr) Procédé et dispositif de transmission et de réception d'un canal physique dans un système de communication sans fil
WO2024014905A1 (fr) Procédé et appareil de rapport de faisceau dans un système de communication sans fil
WO2023136554A1 (fr) Procédé et dispositif d'émission/réception d'un signal sans fil dans un système de communication sans fil
WO2025230256A1 (fr) Procédé exécuté par un équipement utilisateur dans un système de communication sans fil, procédé exécuté par un nœud, équipement utilisateur et nœud
WO2023249223A1 (fr) Procédé exécuté par une station de base, station de base et support de stockage lisible par ordinateur
WO2025075223A1 (fr) Procédé de fonctionnement d'un modèle d'ia/ml dans un système de communication sans fil, et appareil associé
WO2024071853A1 (fr) Procédé et dispositif d'indication de faisceau dans système de communication sans fil
WO2023080636A1 (fr) Procédé et appareil pour réaliser une transmission/réception en liaison montante ou en liaison descendante dans un système de communication sans fil
WO2023163474A1 (fr) Procédé et dispositif de transmission ou de réception d'informations d'état de canal basées sur la quantification dans un système de communication sans fil
WO2025230245A1 (fr) Procédé et appareil pour une construction de cartographie radio dans un système de communication sans fil
WO2025216488A1 (fr) Procédé et dispositif de gestion de faisceaux dans un système de communication sans fil
WO2024172472A1 (fr) Procédé exécuté par un terminal, dispositif électronique et support de stockage
WO2025121837A1 (fr) Procédé et dispositif de réception et d'émission d'informations
WO2025159258A1 (fr) Procédé exécuté par un premier nœud, procédé exécuté par un second nœud et nœuds correspondants dans un système de communication sans fil

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 25797694

Country of ref document: EP

Kind code of ref document: A1