[go: up one dir, main page]

WO2025078006A1 - Attribution de sous-bande d'apprentissage sur graphes - Google Patents

Attribution de sous-bande d'apprentissage sur graphes Download PDF

Info

Publication number
WO2025078006A1
WO2025078006A1 PCT/EP2023/078234 EP2023078234W WO2025078006A1 WO 2025078006 A1 WO2025078006 A1 WO 2025078006A1 EP 2023078234 W EP2023078234 W EP 2023078234W WO 2025078006 A1 WO2025078006 A1 WO 2025078006A1
Authority
WO
WIPO (PCT)
Prior art keywords
subnetworks
subnetwork
identifiers
predetermined number
sub
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/EP2023/078234
Other languages
English (en)
Inventor
Daniel ABODE
Renato Barbosa ABREU
Lou SALAUN
Gilberto BERARDINELLI
Ramoni OJEKUNLE ADEOGUN
Thomas Haaning Jacobsen
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.)
Nokia Technologies Oy
Original Assignee
Nokia Technologies Oy
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 Nokia Technologies Oy filed Critical Nokia Technologies Oy
Priority to PCT/EP2023/078234 priority Critical patent/WO2025078006A1/fr
Publication of WO2025078006A1 publication Critical patent/WO2025078006A1/fr
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
    • H04W16/02—Resource partitioning among network components, e.g. reuse partitioning
    • H04W16/10—Dynamic resource partitioning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • Various example embodiments of the present disclosure generally relate to the field of telecommunication and in particular, to methods, devices, apparatuses and computer readable storage medium for graph-learning sub-band allocation, especially for 6th Generation (6G) subnetwork.
  • 6G 6th Generation
  • the first apparatus comprises at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the first apparatus at least to: receive, from a second apparatus, a report of identifiers of a predetermined number of neighboring subnetworks of a subnetwork at which the second apparatus is located, the predetermined number of neighboring subnetworks being associated with a predefined interference metric; generate a subnetworks deployment conflict graph based on the reported identifiers; obtain an inference model associated with sub-band allocation; and transmit, to the second apparatus, at least one of the following: at least one target subband determined based at least on the inference model and assigned to the subnetwork; or the inference model along with the subnetworks deployment conflict graph.
  • a method comprises: receiving, from a second apparatus, a report of identifiers of a predetermined number of neighboring subnetworks of a subnetwork at which the second apparatus is located, the predetermined number of neighboring subnetworks being associated with a predefined interference metric; generating a subnetworks deployment conflict graph based on the reported identifiers; obtaining an inference model associated with sub-band allocation; and transmitting, to the second apparatus, at least one of the following: at least one target sub-band determined based at least on the inference model and assigned to the subnetwork; or the inference model along with the subnetworks deployment conflict graph.
  • a method comprises: transmitting, to a first apparatus, a report of identifiers of a predetermined number of neighboring subnetworks of the subnetwork, the predetermined number of neighboring subnetworks being associated with a predefined interference metric; receiving, from the first apparatus, at least one of the following: at least one target sub-band determined based on the identifiers and assigned to the subnetwork; or an inference model for a sub-band allocation along with a subnetworks deployment conflict graph associated with the identifiers.
  • the first apparatus comprises means for receiving, from a second apparatus, a report of identifiers of a predetermined number of neighboring subnetworks of a subnetwork at which the second apparatus is located, the predetermined number of neighboring subnetworks being associated with a predefined interference metric; means for generating a subnetworks deployment conflict graph based on the reported identifiers; means for obtaining an inference model associated with sub-band allocation; and means for transmitting, to the second apparatus, at least one of the following: at least one target subband determined based at least on the inference model and assigned to the subnetwork; or the inference model along with the subnetworks deployment conflict graph.
  • FIG. 1 illustrates an example communication environment in which example embodiments of the present disclosure can be implemented
  • FIG. 2 illustrates a signaling chart illustrating a process of graph-learning subband allocation according to some example embodiments of the present disclosure
  • FIG. 3 illustrates a signaling chart illustrating a process of graph-learning sub- band allocation according to some example embodiments of the present disclosure
  • FIG. 4 illustrates a signaling chart illustrating a process of graph-learning subband allocation according to some example embodiments of the present disclosure
  • FIG. 5 illustrates an example of graph neural network for sub-band allocation according to some example embodiments of the present disclosure
  • FIG. 8 illustrates a flowchart of a method implemented at a second device according to some example embodiments of the present disclosure
  • FIG. 9 illustrates a simplified block diagram of a device that is suitable for implementing example embodiments of the present disclosure.
  • circuit(s) and or processor(s) such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation.
  • software e.g., firmware
  • circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and/or firmware.
  • circuitry also covers, for example and if applicable to the particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in server, a cellular network device, or other computing or network device.
  • the communications between a terminal device and a network device in the communication network may be performed according to any suitable generation communication protocols, including, but not limited to, the first generation (1G), the second generation (2G), 2.5G, 2.75G, the third generation (3G), the fourth generation (4G), 4.5G, the fifth generation (5G), the sixth generation (6G) communication protocols, and/or any other protocols either currently known or to be developed in the future.
  • suitable generation communication protocols including, but not limited to, the first generation (1G), the second generation (2G), 2.5G, 2.75G, the third generation (3G), the fourth generation (4G), 4.5G, the fifth generation (5G), the sixth generation (6G) communication protocols, and/or any other protocols either currently known or to be developed in the future.
  • Embodiments of the present disclosure may be applied in various communication systems. Given the rapid development in communications, there will of course also be future type communication technologies and systems with which the present disclosure may be embodied. It should not be seen as limiting the scope of the present disclosure to only the aforementioned system.
  • the term “network device” refers to a node in a communication network via which a terminal device accesses the network and receives services therefrom.
  • the network device may refer to a base station (BS) or an access point (AP), for example, a node B (NodeB or NB), an evolved NodeB (eNodeB or eNB), an NR NB (also referred to as a gNB), a Remote Radio Unit (RRU), a radio header (RH), a remote radio head (RRH), a relay, an Integrated Access and Backhaul (IAB) node, a low power node such as a femto, a pico, a non-terrestrial network (NTN) or non-ground network device such as a satellite network device, a low earth orbit (LEO) satellite and a geosynchronous earth orbit (GEO) satellite, an aircraft network device, and so forth, depending on the applied terminology and technology.
  • BS base station
  • AP access point
  • radio access network (RAN) split architecture comprises a Centralized Unit (CU) and a Distributed Unit (DU) at an IAB donor node.
  • An IAB node comprises a Mobile Terminal (IAB-MT) part that behaves like a UE toward the parent node, and a DU part of an IAB node behaves like a base station toward the next-hop IAB node.
  • IAB-MT Mobile Terminal
  • terminal device refers to any end device that may be capable of wireless communication.
  • a terminal device may also be referred to as a communication device, user equipment (UE), a Subscriber Station (SS), a Portable Subscriber Station, a Mobile Station (MS), or an Access Terminal (AT).
  • UE user equipment
  • SS Subscriber Station
  • MS Mobile Station
  • AT Access Terminal
  • the terminal device may include, but not limited to, a mobile phone, a cellular phone, a smart phone, voice over IP (VoIP) phones, wireless local loop phones, a tablet, a wearable terminal device, a personal digital assistant (PDA), portable computers, desktop computer, image capture terminal devices such as digital cameras, gaming terminal devices, music storage and playback appliances, vehicle-mounted wireless terminal devices, wireless endpoints, mobile stations, laptop-embedded equipment (LEE), laptop-mounted equipment (LME), USB dongles, smart devices, wireless customer-premises equipment (CPE), an Internet of Things (loT) device, a watch or other wearable, a head-mounted display (HMD), a vehicle, a drone, a medical device and applications (e.g., remote surgery), an industrial device and applications (e.g., a robot and/or other wireless devices operating in an industrial and/or an automated processing chain contexts), a consumer electronics device, a device operating on commercial and/or industrial wireless networks, and the like.
  • VoIP voice over
  • subnetwork is inter-changeable with “sub-network” and may refer to a network that can be installed in a specific entity that can provide services with extreme performances over a local capillary coverage.
  • intersubnetwork interference used herein may refer to an interference between subnetworks.
  • intra-subnetwork measurement used herein may refer to a measurement within a subnetwork.
  • sub-band used herein may refer to a set of time and frequency domain resources.
  • 6G radio access technology may expect extreme high requirements in terms of latency, reliability and/or throughput and the In-X subnetwork (i.e., subnetwork) may be considered as a promising component of 6G network to meet these extreme performance requirements.
  • the subnetworks may have the following pivotal properties and technical features and the system design for in-X subnetwork shall take the above technical features into account:
  • system design for in-X subnetwork may take the above technical features into account.
  • subnetworks may be seen as a potential evolution of 5G Sidelink, where, on the other hand, many enhancements are needed, for example:
  • in-robot/in-production module subnetworks and in-vehicle subnetworks have extreme performance requirements in both reliability (up to 6 nines or more) and latency (down to the level of lOOus or even below) e.g., for the high demanding periodic deterministic communication services and these use cases may be the most challenging scenarios in 6G system.
  • a solution of graph-learning sub-band allocation is proposed.
  • the proposed solution of the present disclosure may improve the performance of the sub-band allocation and reduce the computational time complexity. Meanwhile, a signaling overhead for the sub-band allocation can be decreased.
  • FIG. 1 illustrates an example communication environment 100 in which example embodiments of the present disclosure can be implemented.
  • the communication environment 100 comprises a plurality of subnetworks, e.g., subnetworks 101, 102, 103, 104, ..., 10N.
  • the subnetwork may also comprise one or more other devices (e.g., UEs).
  • UEs 140-1 and 140-2 are located in the subnetwork 101
  • UEs 141-1 and 141-2 are located in the subnetwork 102
  • UEs 142-1 and 142-2 are located in the subnetwork 103
  • UEs 143-1 and 143-2 are located in the subnetwork 104
  • UEs 14N- 1 and 14N-2 are located in the subnetwork 10N. It is to be understood that the number of UEs may differ between subnetworks.
  • a UE may communicate with other UE(s) and/or AP(s) within the subnetwork at which the UE is located. In some other scenarios, a UE may communicate with other UE(s) and/or AP(s) located at other subnetworks.
  • the UE 141-2 may communicate with the UE 141-1 and the AP 120-1 in the subnetwork 102. The 141-1 may also communicate with the AP 120-2 in the subnetwork 103 or the UE 140-2 in the subnetwork 101.
  • An AP in a subnetwork may act as a centralized AP and other APs in other subnetworks may act as distributed APs.
  • the AP 110 may act as the centralized AP and APs 120-1, 120-2, 120-3 and 120-(N-l) may act as distributed APs. It is to be understood that, depending on capabilities of APs, in some scenarios, some of distributed APs 120-1, 120-2, 120-3 and 120-(N-l) may also act as a centralized AP and the centralized AP 110 may also act as a distributed AP.
  • the centralized AP 110 may also be referred to as a first apparatus, and the distributed APs 120-1, 120-2, 120-3 and 120-(N-l) may also be referred to second apparatus collectively.
  • an AP in the communication environment 100 may be implemented as a terminal device or a network device.
  • the communication environment 100 further comprises a BS 130, which may be considered as a network device in a RAN.
  • the BS 130 may also be referred to as the second apparatus in some scenarios.
  • the BS 130 may communicate with the centralized AP 110 and/or with the distributed APs 120-1, 120-2, 120-3 and 120-(N-l). In some example embodiments, operations described in connection with the centralized AP 110 may be implemented at the BS 130.
  • the communication environment 100 may include any suitable number of subnetworks, APs, network devices and terminal devices.
  • a link from the network device to the terminal device is referred to as a downlink (DL), and a link from the terminal device to the network device is referred to as an uplink (UL).
  • the network device is a transmitting (TX) device (or a transmitter) and the terminal device is a receiving (RX) device (or a receiver).
  • the terminal device is a TX device (or a transmitter) and the network device is a RX device (or a receiver).
  • Communications in the communication environment 100 may be implemented according to any proper communication protocol(s), comprising, but not limited to, cellular communication protocols of the first generation (1G), the second generation (2G), the third generation (3G), the fourth generation (4G), the fifth generation (5G), the sixth generation (6G), and the like, wireless local network communication protocols such as Institute for Electrical and Electronics Engineers (IEEE) 802.11 and the like, and/or any other protocols currently known or to be developed in the future.
  • IEEE Institute for Electrical and Electronics Engineers
  • FIG. 2 shows a signaling chart 200 for a process of graph-learning sub-band allocation according to some example embodiments of the present disclosure.
  • the signaling chart 200 involves a AP 110 in the subnetwork 101 (hereinafter may also be referred to centralized AP 110), AP 120- 1, ..., 120-(N-l) in the subnetworks 102, ..., 10N, respectively (hereinafter may also be referred to distributed APs 120-1, ..., 120-(N-l), respectively or a distributed AP 120 collectively), BS 130 and UEs 140-1, 140-2, ..., 14N-1 and 14N-2.
  • the centralized AP 110 may transmit (202), to all distributed APs 120-1, ..., 120-(N-l) (including devices located at subnetworks at which the distributed APs 120-1, ..., 120-(N-l) are located), a configuration associated with a transmission of reference signals, a measurement on the reference signals and a rule for determining identifiers of one or more neighboring subnetworks to be reported.
  • the configuration may configure the devices in a subnetwork to transmit reference signals.
  • the devices in the subnetwork 10N may also transmit the reference signals to other subnetworks, such as a reference signal transmission (216) between UE 14N-1 and AP 120-1 in the subnetwork 102 and/or a reference signal transmission (218) between UE 14N-2 and AP 120-1 in the subnetwork 102.
  • a reference signal transmission (216) between UE 14N-1 and AP 120-1 in the subnetwork 102
  • a reference signal transmission 218
  • the APs may transmit the reference signals themselves (e.g., with power associated to AP-AP or AP-UE communications, or with power associated to the subnetwork UEs transmissions from UE-UE or UE-AP communication).
  • the configuration may configure may also configure a rule for the distributed AP 120 in a subnetwork to determine, based on the measurement on the signal strength of reference signals associated with its neighboring subnetworks, identifiers of the predetermined number of neighboring subnetworks with a predefined interference metric to be reported.
  • the rule may indicate an interference graph (IG) based rule, which means that the distributed AP 120 may determine and report the predetermined number of neighboring subnetworks (e.g., K-l) associated with the strongest interference. That is, the predetermined number of neighboring subnetworks may have strongest interference for the subnetwork at which the distributed AP 120 is located.
  • IG interference graph
  • the rule may indicate a signal to interference ratio (SIR) Graph (SG) based rule, which means that the distributed AP 120 may determine and report the predetermined number of neighboring subnetworks (e.g., K-l) having lowest SIR.
  • SIR signal to interference ratio
  • SG signal to interference ratio
  • the distributed AP 120-1 in the subnetwork 102 may measure the signal strength of reference signals associated to the subnetwork 10N, i.e., a reference signal transmission (216) between UE 14N-1 in the subnetwork 10N and AP 120-1 and/or a reference signal transmission (218) between UE 14N-2 in the subnetwork 10N and AP 120-1, to determine strongest interferences or to determine SIR.
  • a reference signal transmission 216
  • UE 14N-2 in the subnetwork 10N and AP 120-1
  • the distributed AP 120-1 in the subnetwork 102 may measure the signal strength of reference signals based on e.g., a reference signal transmission (204) between the distributed AP 120-1 and UE 141-1, a reference signal transmission (206) between the distributed AP 120-1 and UE 141-2, for determine SIR.
  • the distributed AP 120-(N-l) in the subnetwork 10N may also measure the signal strength of reference signals associated to the subnetwork 102, i.e., a reference signal transmission (208) between UE 141-1 in the subnetwork 102 and the distributed AP 120-(N-l) and/or a reference signal transmission (210) between UE 141-2 in the subnetwork 102 and the distributed AP 120-(N-l).
  • the predetermined number of neighboring subnetworks may depend on the number of available sub-bands.
  • K is the number of orthogonal subbands in frequency domain, as the simplest implementation.
  • K is the number of orthogonal sub-channels in time, frequency, spatial, and/or code domain.
  • the orthogonal sub-channels are to be allocated to the subnetworks.
  • the value of K may be much lower than N, i.e., K «N, hence, the K sub-bands (or sub-channels) are expected to be reused by multiple subnetworks.
  • the distributed AP 120 may determine the predetermined number of neighboring subnetworks with a predefined interference metric to be reported. For example, the distributed AP 120-1 may determine (220) K-l neighboring subnetworks associated with strongest interference or lowest SIR. Similarly, the distributed AP 120- (N-l) may determine (220) K-l neighboring subnetworks associated with strongest interference or lowest SIR. Each of the distributed AP 120-1 and distributed AP 120-(N- 1) may report (222) identifiers of the determined K-l neighboring subnetworks to the centralized AP 110.
  • the distributed APs 120-1, ..., 120-(N-l) may provide the report periodically, or by a request, or in case the determined neighboring has changed in comparison to previous reports.
  • FIG. 3 shows signaling chart 300 for a process of graph-learning sub-band allocation according to some example embodiments of the present disclosure.
  • the signaling chart 300 involves a AP 110 in the subnetwork 101 (hereinafter may also be referred to centralized AP 110), AP 120-1, ...
  • the nodes represent the intra-subnetwork communication links and edges represent mutual interference between subnetworks.
  • the graph can be unattributed or attributed by the channel information of such links.
  • the training dataset for the GNN includes the training input graph of deployed subnetworks, and corresponding other information needed for the optimization objective which may include channel gain matrix of the deployment if the optimization objective is a function of the SINR.
  • the GNN performs a convolutional operation on the graph to generate the node embedding, which is further processed by a node task function to output the radio resource decision e.g., power and/or channel allocation for the node.
  • the training procedure involves estimating the objective of the radio resource allocation decision as a loss function.
  • the estimated loss function is backpropagated through gradient descent to update the GNN parameters and maximize the objective.
  • the GNN used hereinafter may be referred to as a Gated Graph Neural Network (GGNN) topology which includes at least a loss function (e.g., based on Potts model or any other suitable loss function) to learn from the graph structure. It is to be understood that other suitable type of GNNs may also be used in the solution of the present disclosure.
  • GGNN Gated Graph Neural Network
  • a configuration of graph neural networks topology which may include layers, activation functions, size of embedding, trainable weights initialization, and biases, etc., as well as a configuration of the loss function based on Potts model to learn from the graph structure may be provided (302) from the BS 130 to the centralized AP 110, e.g., via dedicated RRC signaling.
  • the training may be performed by the centralized AP 110 or by an overlaying BS 130.
  • the centralized AP 110 may also offload some part of the training tasks to BS 130, and/or the BS 130 can assist the training, e.g., providing data augmentation based on past acquired data stored in a database. If training is performed in BS 130, the trained model may be finally provided to centralized AP 110 (in case coverage from BS 130 is not always available).
  • the centralized AP 110 may generate (306) a subnetworks deployment conflict graph.
  • ⁇ I> i is the trainable feature transformation weight matrix of the message aggregation function, denotes the previous node embedding of the Neighbor.
  • the update function is implemented using a gated recurrent unit (GRU), where the input is the aggregated message, and the hidden state is the previous node embedding.
  • GRU gated recurrent unit
  • an unsupervised training algorithm which does not depend on any ground truth is employed.
  • the training graphs are generated using the conflict graph model of the subnetworks.
  • a mini -batch of graphs is propagated through the GGNN layers which execute aggregation, update function, and readout function.
  • the O n at training iteration t is updated until training converges.
  • the neighboring subnetwork IDs reported by the APs can be added to the training/testing database. This way, the GGNN trainable parameters (weights/biases) could be continuously adjusted. This may converge to minimize the loss function in long term.
  • the centralized AP 110 may determine (312) the at least one sub-band allocated for the subnetwork 102 and the subnetwork 10N, respectively.
  • the identifiers of neighboring subnetworks reported by the distributed AP 120-1 and the distributed AP 120-(N-l) may be the inputs for the GGNN and the at least one sub-band to be allocated for the subnetwork 102 and the subnetwork 10N are the outputs of the GGNN.
  • the centralized AP 110 may assign (314) the at least one sub-band allocated for the subnetwork 102 and the subnetwork 10N to the distributed AP 120-1 and the distributed AP 120-(N-l), respectively.
  • the process shown in FIG. 3 explains the case where the trained GGNN model is executed for the sub-band allocation in a centralized way.
  • a centralized execution could be preferred if the subnetworks are within the coverage of the centralized AP 110 or a BS 130.
  • the centralized AP 110 or a BS 130 may obtain the K-l neighbor identifiers for all subnetworks, build the conflict graph, execute the trained GGNN model, and signal the sub-band allocation decision to the subnetworks.
  • the determination of the sub-band allocation may also be performed by the distributed AP itself, i.e., the trained GGNN model is executed for the sub-band allocation in a decentralized way, which may be further described in detail with respective to FIG. 4.
  • each subnetwork (or the distributed AP 120) may obtain a copy of the trained GGNN model (e.g., from centralized AP 110, or a BS 130, or forwarded by a distributed AP 120).
  • Each distributed may also receive the subnetworks deployment conflict graph, so that each distributed AP 120 may have a list of its neighboring subnetworks.
  • a configuration for obtaining trained model, exchange embeddings between neighboring APs may be provided (402) from the centralized AP 110 or a BS 130 to each distributed AP 120 in synchronous or asynchronous way.
  • the distributed AP 120 may execute layer I to obtain embedding 6 ⁇ based on the message ⁇ 5 ⁇ -1 received from neighboring subnetworks.
  • layer I may be executed to obtain embedding 6 ⁇ based on the message ⁇ 5 ⁇ -1 received from neighboring subnetworks.
  • such implementation would require L rounds of such message passing and the size of each message depends on the size of the embedding.
  • I 0...L.
  • These operations can be synchronous or asynchronous between the distributed APs.
  • the final node embedding of distributed AP may be used to determine its allocated sub-bands.
  • the distributed AP 120-1 may report (406) its embedding ⁇ (Z) and iteration index I.
  • the distributed AP 120-1 may also receive (408) the distributed AP 120-(N-l)’s embedding ⁇ 5 W (Z) and iteration index I.
  • the distributed AP 120-1 may run one layer of GGNN to obtain (410) an updated node embedding 6- ⁇ (1 + 1).
  • the initial embedding ⁇ (0) may set locally by distributed AP 120- 1 to a constant value of 1.
  • the output of the model is obtained by processing the AP’s final embedding ⁇ (L), which determines the sub-band for the APs (e.g., by a function which compresses the final node embedding after L GGNN layers into a normalized soft vector of size K).
  • the distributed AP 120-1 may determine (412) sub-band allocated for the subnetwork 102 based on the output of the trained GGNN and the assign (414) the determined sub-band for the subnetwork 102.
  • the operations performed in the distributed AP 120-1 may also be implemented in the distributed AP 120-(N-l).
  • the decentralized execution may run in a synchronous way.
  • the BS 130 may start of the execution procedure, otherwise the procedure should start locally by one of the subnetworks, e.g., by the centralized AP 110.
  • the distributed APs can run in an asynchronous way.
  • the distributed AP 120-1 may receive neighboring embeddings with iteration index I' not equal to its own iteration I. If U ⁇ I, the embedding is outdated and simply discarded by distributed AP 120-1. If U > I, the embedding is stored by distributed AP 120-1 to be used later.
  • decentralized execution is that after receiving the model it does not require a connection with the centralized AP 110 (or BS 130), however the procedure has higher signaling overhead compared to centralized execution.
  • the performance of the various benchmarks and the solution proposed in the present disclosure has been compared in terms of the network and per-device spectral efficiency, execution complexity and generalizability.
  • the proposed GGNN based approach assuming two different rules (IG and SG) is compared with a random allocation (RA), Sequential Iterative Sub-band Allocation (SISA) and Centralized Graph Coloring (CGC) method for sub-band allocation.
  • RA random allocation
  • SISA Sequential Iterative Sub-band Allocation
  • CGC Centralized Graph Coloring
  • FIG. 6A and FIG. 6B show the empirical cumulative distribution function (CDF) plot of the realized sum SE and per-device SE respectively for the proposed scheme and the different benchmarks tested with 10000 network realizations.
  • CDF empirical cumulative distribution function
  • the complexity of the proposed solution of the present disclosure can analyzed and compared it with the benchmark algorithms, CGC and SISA in terms of the computational runtime and signaling requirement.
  • the result of the runtime for different numbers of subnetworks, N ⁇ 50, 60, • • • , 200 ⁇ in FIG. 6C is averaged over 10000 realizations.
  • the proposed solution of the present disclosure has a faster runtime, growing at a slower linear rate compared to the SISA and CGC. Hence, it would be more suitable for very dense networks with a large number of subnetworks or APs.
  • the runtime analysis is carried out on a CPU, it is expected that the runtime for the GGNN method would further decrease on a GPU.
  • the GGNN requires less information and therefore incurs fewer signaling resources than SISA.
  • N2 signaling messages are required to be signaled to the central resource management entity from N subnetworks to execute the SISA algorithm, however, the GGNN method only requires N(K-l) signaling messages, where K «N in a large network. This further justifies the suitability of the proposed solution of the present disclosure for large-scale deployment of subnetworks.
  • GGNN models are trained from training graphs constructed based on IG for a given scenario and tested with different scenarios. As shown in Error! Reference source not found., it can be observed that the realized average SE from testing with 10000 snapshots remain relatively the same for a test scenario, regardless of the training scenario. This shows that the trained model can generalize to different numbers of subnetworks, density and channel models. The robustness is because the GGNN model learns based on the graph structure which depends on the graph construction rule and not the distribution of the channel model.
  • FIG. 7 shows a flowchart of an example method 700 implemented at a first apparatus in accordance with some example embodiments of the present disclosure.
  • the method 700 will be described from the perspective of the first apparatus (e.g., the centralized AP 110 or the BS 130) in FIG. 1.
  • the first apparatus generates a subnetworks deployment conflict graph based on the reported identifiers.
  • the first apparatus obtains an inference model associated with subband allocation.
  • the first apparatus transmits to the second apparatus, at least one of the following: at least one of the following: at least one target sub-band determined based at least on the inference model and assigned to the subnetwork; or the inference model along with the subnetworks deployment conflict graph.
  • the method 700 further comprises: determining a configuration associated with a measurement on reference signals from neighboring subnetworks, wherein the configuration indicates at least one interference metric and a rule for the second apparatus to determine the identifiers of the predetermined number of neighboring subnetworks based on the at least one interference metric; and transmitting the configuration to the second apparatus.
  • the rule indicates at least one of: the identifiers of the predetermined number of neighboring subnetworks associated with the strongest interference are to be determined, or the identifiers of the predetermined number of neighboring subnetworks associated with the lowest signal to interference ratio are to be determined.
  • the method 700 further comprises: determining the at least one target sub-band assigned to the subnetwork based on the inference model and the identifier, wherein the identifiers are an input of the inference model and the at least one target sub-band is the output of the inference model.
  • the method 700 further comprises: obtaining, from the at least one second apparatus, information related to respective history identifiers of the predetermined number of subnetworks, neighboring to the respective subnetwork at which the at least one second apparatus is located, with a predefined interference metric; generating the at least one history subnetworks deployment conflict graph based on the respective history identifiers; and training the inference model for the sub-band allocation based on the at least one history subnetworks deployment and a predetermined or configured loss function.
  • the inference model is trained utilizing a loss function based on a Potts model.
  • the first apparatus comprises a centralized access point or a base station
  • the second apparatus comprises a distributed access point or a user equipment located in the subnetwork.
  • FIG. 8 shows a flowchart of an example method 800 implemented at a second apparatus in accordance with some example embodiments of the present disclosure.
  • the method 800 will be described from the perspective of the second apparatus (e.g., the distributed AP 120) in FIG. 1.
  • the second apparatus transmits, to a first apparatus, a report of identifiers of a predetermined number of neighboring subnetworks of the subnetwork, the predetermined number of neighboring subnetworks being associated with a predefined interference metric.
  • the second apparatus receives, from the first apparatus, at least one of the following: at least one target sub-band determined based on the identifiers and assigned to the subnetwork; or an inference model for a sub-band allocation along with a subnetworks deployment conflict graph associated with the identifiers.
  • the rule indicates at least one of the identifiers of the predetermined number of neighboring subnetworks associated with the strongest interference are to be determined, or the identifiers of the predetermined number of neighboring subnetworks associated with the lowest signal to interference ratio are to be determined.
  • the method 800 further comprises: determining at least one sub-band available for the subnetwork, at which the second apparatus is located, based on the inference model for the sub-band allocation and the subnetworks deployment conflict graph.
  • the method 800 further comprises: transmitting, to at least one further second apparatus located in the neighboring subnetworks, a node embedding of the second apparatus; receiving, from the at least one further second apparatus, respective further node embedding of the at least one further second apparatus; updating, by using the inference model and the subnetworks deployment conflict graph, the node embedding of the second apparatus based on the respective further node embedding iteratively; and determining the at least one sub-band available for the subnetwork based on the updated node embedding of the second apparatus.
  • the inference model is trained utilizing a loss function based on a Potts model.
  • a first apparatus capable of performing any of the method 700 may comprise means for performing the respective operations of the method 700.
  • the means may be implemented in any suitable form.
  • the means may be implemented in a circuitry or software module.
  • the first apparatus may be implemented as or included in the centralized AP 110 or the BS 130 in FIG. 1.
  • the first apparatus comprises means for receiving, from a second apparatus, a report of identifiers of a predetermined number of neighboring subnetworks of a subnetwork at which the second apparatus is located, the predetermined number of neighboring subnetworks being associated with a predefined interference metric; means for generating a subnetworks deployment conflict graph based on the reported identifiers; means for obtaining an inference model associated with subband allocation; and means for transmitting, to the second apparatus, at least one of the following: at least one target sub-band determined based at least on the inference model and assigned to the subnetwork; or the inference model along with the subnetworks deployment conflict graph.
  • the first apparatus further comprises: means for determining a configuration associated with a measurement on reference signals from neighboring subnetworks, wherein the configuration indicates at least one interference metric and a rule for the second apparatus to determine the identifiers of the predetermined number of neighboring subnetworks based on the at least one interference metric; and means for transmitting the configuration to the second apparatus.
  • the inference model for the sub-band allocation has been trained at least based on at least one history subnetworks deployment conflict graph, reported history identifiers and a predetermined or configured loss function.
  • the first apparatus further comprises: means for obtaining, from the at least one second apparatus, information related to respective history identifiers of the predetermined number of subnetworks, neighboring to the respective subnetwork at which the at least one second apparatus is located, with a predefined interference metric; means for generating the at least one history subnetworks deployment conflict graph based on the respective history identifiers; and means for training the inference model for the sub-band allocation based on the at least one history subnetworks deployment and a predetermined or configured loss function.
  • the inference model is trained utilizing a loss function based on a Potts model.
  • the predetermined number is associated with the number of available sub-bands.
  • the first apparatus comprises a centralized access point or a base station
  • the second apparatus comprises a distributed access point or a user equipment located in the subnetwork.
  • the first apparatus further comprises means for performing other operations in some example embodiments of the method 700 or the centralized AP 110 or the BS 130 in FIG. 1.
  • the means comprises at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the performance of the first apparatus.
  • a second apparatus capable of performing any of the method 800 may comprise means for performing the respective operations of the method 800.
  • the means may be implemented in any suitable form.
  • the means may be implemented in a circuitry or software module.
  • the second apparatus may be implemented as or included in the distributed AP 120 in FIG. 1.
  • the second apparatus comprises means for transmitting, to a first apparatus, a report of identifiers of a predetermined number of neighboring subnetworks of the subnetwork, the predetermined number of neighboring subnetworks being associated with a predefined interference metric; means for receiving, from the first apparatus, at least one of the following: at least one target sub-band determined based on the identifiers and assigned to the subnetwork; or an inference model for a sub-band allocation along with a subnetworks deployment conflict graph associated with the identifiers.
  • the second apparatus further comprises: means for receiving, from the first apparatus, a configuration associated with a measurement on reference signals from neighboring subnetworks, wherein the configuration indicates at least one interference metric and a rule for the second apparatus to determine the identifiers of the predetermined number of neighboring subnetworks based on the at least one interference metric; and means for measuring, based on the configuration, respective strengths of the reference signals from the neighboring subnetworks; and means for generating the report based on the measured respective strengths and the rule.
  • the rule indicates at least one of: the identifiers of the predetermined number of neighboring subnetworks associated with the strongest interference are to be determined, or the identifiers of the predetermined number of neighboring subnetworks associated with the lowest signal to interference ratio are to be determined.
  • the second apparatus further comprises: means for determining at least one sub-band available for the subnetwork, at which the second apparatus is located, based on the inference model for the sub-band allocation and the subnetworks deployment conflict graph.
  • the second apparatus further comprises: means for transmitting, to at least one further second apparatus located in the neighboring subnetworks, a node embedding of the second apparatus; means for receiving, from the at least one further second apparatus, respective further node embedding of the at least one further second apparatus; means for updating, by using the inference model and the subnetworks deployment conflict graph, the node embedding of the second apparatus based on the respective further node embedding iteratively; and means for determining the at least one sub-band available for the subnetwork based on the updated node embedding of the second apparatus.
  • the inference model comprises a machinelearning model related to a graph neural network.
  • the inference model is trained utilizing a loss function based on a Potts model.
  • the predetermined number is associated with the number of available sub-bands.
  • the first apparatus comprises a centralized access point or a base station
  • the second apparatus comprises a distributed access point or a user equipment located in the subnetwork.
  • the second apparatus further comprises means for performing other operations in some example embodiments of the method 800 or the distributed AP 120 in FIG. 1.
  • the means comprises at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the performance of the second apparatus.
  • FIG. 9 is a simplified block diagram of a device 900 that is suitable for implementing example embodiments of the present disclosure.
  • the device 900 may be provided to implement a communication device, for example, the centralized AP 110 or the BS 130 or the distributed AP 120 as shown in FIG. 1.
  • the device 900 includes one or more processors 910, one or more memories 920 coupled to the processor 910, and one or more communication modules 940 coupled to the processor 910.
  • the communication module 940 is for bidirectional communications.
  • the communication module 940 has one or more communication interfaces to facilitate communication with one or more other modules or devices.
  • the communication interfaces may represent any interface that is necessary for communication with other network elements.
  • the communication module 940 may include at least one antenna.
  • the processor 910 may be of any type suitable to the local technical network and may include one or more of the following: general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs) and processors based on multicore processor architecture, as non-limiting examples.
  • the device 900 may have multiple processors, such as an application specific integrated circuit chip that is slaved in time to a clock which synchronizes the main processor.
  • the memory 920 may include one or more non-volatile memories and one or more volatile memories.
  • the non-volatile memories include, but are not limited to, a Read Only Memory (ROM) 924, an electrically programmable read only memory (EPROM), a flash memory, a hard disk, a compact disc (CD), a digital video disk (DVD), an optical disk, a laser disk, and other magnetic storage and/or optical storage.
  • the volatile memories include, but are not limited to, a random access memory (RAM) 922 and other volatile memories that will not last in the power-down duration.
  • a computer program 930 includes computer executable instructions that are executed by the associated processor 910.
  • the instructions of the program 930 may include instructions for performing operations/acts of some example embodiments of the present disclosure.
  • the program 930 may be stored in the memory, e.g., the ROM 924.
  • the processor 910 may perform any suitable actions and processing by loading the program 930 into the RAM 922.
  • the example embodiments of the present disclosure may be implemented by means of the program 930 so that the device 900 may perform any process of the disclosure as discussed with reference to FIG. 2 to FIG. 8.
  • the example embodiments of the present disclosure may also be implemented by hardware or by a combination of software and hardware.
  • the program 930 may be tangibly contained in a computer readable medium which may be included in the device 900 (such as in the memory 920) or other storage devices that are accessible by the device 900.
  • the device 900 may load the program 930 from the computer readable medium to the RAM 922 for execution.
  • the computer readable medium may include any types of non-transitory storage medium, such as ROM, EPROM, a flash memory, a hard disk, CD, DVD, and the like.
  • non-transitory is a limitation of the medium itself (i.e., tangible, not a signal) as opposed to a limitation on data storage persistency (e.g., RAM vs. ROM).
  • various embodiments of the present disclosure may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. Some aspects may be implemented in hardware, and other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device. Although various aspects of embodiments of the present disclosure are illustrated and described as block diagrams, flowcharts, or using some other pictorial representations, it is to be understood that the block, apparatus, system, technique or method described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.
  • Some example embodiments of the present disclosure also provide at least one computer program product tangibly stored on a computer readable medium, such as a non- transitory computer readable medium.
  • the computer program product includes computerexecutable instructions, such as those included in program modules, being executed in a device on a target physical or virtual processor, to carry out any of the methods as described above.
  • program modules include routines, programs, libraries, objects, classes, components, data structures, or the like that perform particular tasks or implement particular abstract data types.
  • the functionality of the program modules may be combined or split between program modules as desired in various embodiments.
  • Machine-executable instructions for program modules may be executed within a local or distributed device. In a distributed device, program modules may be located in both local and remote storage media.
  • Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages.
  • the program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program code, when executed by the processor or controller, cause the functions/operations specified in the flowcharts and/or block diagrams to be implemented.
  • the program code may execute entirely on a machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
  • the computer program code or related data may be carried by any suitable carrier to enable the device, apparatus or processor to perform various processes and operations as described above.
  • Examples of the carrier include a signal, computer readable medium, and the like.
  • the computer readable medium may be a computer readable signal medium or a computer readable storage medium.
  • a computer readable medium may include but not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the computer readable storage medium would include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

L'invention concerne des procédés, des appareils et un support de stockage lisible par ordinateur pour l'attribution de sous-bandes d'apprentissage sur graphes. Le procédé comprend les étapes suivantes : réception (710), en provenance d'un second appareil, d'un rapport d'identifiants d'un nombre prédéterminé de sous-réseaux voisins d'un sous-réseau au niveau duquel le second appareil est situé, le nombre prédéterminé de sous-réseaux voisins étant associé à une métrique d'interférence prédéfinie; génération (720) d'un graphe de conflit de déploiement de sous-réseaux sur la base des identifiants rapportés; obtention (730) d'un modèle d'inférence associé à une attribution de sous-bande; et transmission (740), au second appareil, d'au moins l'un des éléments suivants : au moins une sous-bande cible déterminée sur la base au moins du modèle d'inférence et attribuée au sous-réseau; ou le modèle d'inférence conjointement avec le graphe de conflit de déploiement de sous-réseaux.
PCT/EP2023/078234 2023-10-11 2023-10-11 Attribution de sous-bande d'apprentissage sur graphes Pending WO2025078006A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/EP2023/078234 WO2025078006A1 (fr) 2023-10-11 2023-10-11 Attribution de sous-bande d'apprentissage sur graphes

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/EP2023/078234 WO2025078006A1 (fr) 2023-10-11 2023-10-11 Attribution de sous-bande d'apprentissage sur graphes

Publications (1)

Publication Number Publication Date
WO2025078006A1 true WO2025078006A1 (fr) 2025-04-17

Family

ID=88412234

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2023/078234 Pending WO2025078006A1 (fr) 2023-10-11 2023-10-11 Attribution de sous-bande d'apprentissage sur graphes

Country Status (1)

Country Link
WO (1) WO2025078006A1 (fr)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120122467A1 (en) * 2010-11-15 2012-05-17 Gunther Auer Method for assigning frequency subbands to a plurality of interfering nodes in a wireless communication network, controller for a wireless communication network and wireless communication network

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120122467A1 (en) * 2010-11-15 2012-05-17 Gunther Auer Method for assigning frequency subbands to a plurality of interfering nodes in a wireless communication network, controller for a wireless communication network and wireless communication network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
ABODE DANIEL ET AL: "Power Control for 6G Industrial Wireless Subnetworks: A Graph Neural Network Approach", 2023 IEEE WIRELESS COMMUNICATIONS AND NETWORKING CONFERENCE (WCNC), IEEE, 26 March 2023 (2023-03-26), pages 1 - 6, XP034340760, DOI: 10.1109/WCNC55385.2023.10118984 *
ADEOGUN RAMONI ET AL: "Learning to Dynamically Allocate Radio Resources in Mobile 6G in-X Subnetworks", 2021 IEEE 32ND ANNUAL INTERNATIONAL SYMPOSIUM ON PERSONAL, INDOOR AND MOBILE RADIO COMMUNICATIONS (PIMRC), IEEE, 13 September 2021 (2021-09-13), pages 959 - 965, XP034004841, DOI: 10.1109/PIMRC50174.2021.9569345 *
WANG ZHAOKANG ET AL: "Empirical analysis of performance bottlenecks in graph neural network training and inference with GPUs", ARXIV,, vol. 446, 13 March 2021 (2021-03-13), pages 165 - 191, XP086597299, DOI: 10.1016/J.NEUCOM.2021.03.015 *

Similar Documents

Publication Publication Date Title
CN114390580B (zh) 波束上报方法、波束信息确定方法及相关设备
Mismar et al. A framework for automated cellular network tuning with reinforcement learning
US20240314613A1 (en) Communication method and apparatus
US12301320B2 (en) Method and system for unsupervised user clustering and power allocation in non-orthogonal multiple access (NOMA)-aided massive multiple input-multiple output (MIMO) networks
WO2019196801A1 (fr) Procédé de transmission de données et appareil et système de communication
US20210409086A1 (en) Ue recommended csi settings
Arvinte et al. Beam management in 5G NR using geolocation side information
US20240137783A1 (en) Signalling support for split ml-assistance between next generation random access networks and user equipment
WO2023093725A1 (fr) Procédé et appareil d'étalonnage
EP2638676A1 (fr) Allocation de ressources dans un système de communication
JP7745105B2 (ja) モデルのためのデータ処理方法、及び装置
US12231371B2 (en) Interference avoidance for cellular networks
US20230217312A1 (en) Adaptive radio access network bit rate scheduling
CN114978268B (zh) 去蜂窝导频长度优化方法、装置、电子设备及存储介质
EP4020912B1 (fr) Procédé et appareil de programmation multi-utilisateur dans un système de communication sans fil
WO2025078006A1 (fr) Attribution de sous-bande d'apprentissage sur graphes
WO2024044896A1 (fr) Procédure de sélection de sous-bande basée sur une pondération du brouillage
US20140044033A1 (en) Apparatus and Method for Communication with a Number of User Equipments Using OFDMA
US20250016065A1 (en) Server and agent for reporting of computational results during an iterative learning process
CN116724517A (zh) 减少干扰和优化参数
KR20230161970A (ko) 향상된 크로스 링크 간섭 측정 및 관리
WO2025054832A1 (fr) Attribution de puissance pour sous-réseau
US20250056519A1 (en) Mechanism for reinforcement learning on beam management
US20250031068A1 (en) Mechanism for functionality based life cycle management
WO2024152170A1 (fr) Attribution de ressources pour sous-réseaux

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: 23789618

Country of ref document: EP

Kind code of ref document: A1