WO2025230585A1

WO2025230585A1 - Methods, apparatuses and systems for handover decision making

Info

Publication number: WO2025230585A1
Application number: PCT/US2025/012094
Authority: WO
Inventors: Omar SLEEM; Amit Kalhan; Henry Chang
Original assignee: Kyocera Corp
Current assignee: Kyocera Corp
Priority date: 2024-05-01
Filing date: 2025-01-17
Publication date: 2025-11-06
Anticipated expiration: 2026-11-01
Also published as: WO2025230584A1

Abstract

Methods, apparatuses and systems for neighboring cell's reference signal received power reporting-based machine learning for handover decision making. In one embodiment, a method includes: transmitting, at a wireless communication device, a first signal to a first plurality of wireless communication nodes, wherein the first plurality of wireless communication nodes includes a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal includes at least one sounding reference signal (SRS); receiving, at the wireless communication device, a second signal from the first wireless communication node, wherein the second signal is determined based on the first signal, and the second signal includes: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

Description

METHODS, APPARATUSES AND SYSTEMS FOR HANDOVER DECISION MAKING

TECHNICAL FIELD

[0001] The disclosure relates generally to wireless communications and, more particularly, to methods, apparatuses and systems for neighboring cell’s reference signal received power reporting-based machine learning for handover decision making.

BACKGROUND

[0002] Following the standardization of the initial 5GNew Radio (NR) Release 15 in 2018, the evolution of 5GNR has progressed swiftly, laying the groundwork for the global commercial development of 5G technology. The utilization of millimeter Wave (mmWave) spectrum has been pivotal in delivering performance enhancements due to its provision of large bandwidth. However, mmWave signals are susceptible to significant free space path loss and other atmospheric perturbations.

[0003] Consequently, both the 5G Node Base Station (gNB) and the User Equipment (UE) must establish highly directional links to maintain acceptable communication quality. This necessitates the process of aligning transmit beams at the gNB with receive beams at the UE, particularly in Downlink (DL) transmission scenarios, which is termed as Beam Management (BM). As transmit and receive beams are selected from finite-sized codebooks, identifying the optimal beam pair primarily relies on an exhaustive search process involving sweeping through all beams in the codebook. However, this approach incurs considerable training overhead. Therefore, there is a need to streamline the beam selection process and reduce the above training overhead for the beam search in a wireless communication network. SUMMARY

[0004] The exemplary embodiments disclosed herein are directed to solving the issues relating to one or more of the problems presented in the prior art, as well as providing additional features that will become readily apparent by reference to the following detailed description when taken in conjunction with the accompany drawings. In accordance with various embodiments, exemplary systems, methods, devices and computer program products are disclosed herein. It is understood, however, that these embodiments are presented by way of example and not limitation, and it will be apparent to those of ordinary skill in the art who read the present disclosure that various modifications to the disclosed embodiments can be made while remaining within the scope of the present disclosure.

[0005] In some embodiments, a method includes: transmitting, at a wireless communication device, a first signal to a first plurality of wireless communication nodes, wherein the first plurality of wireless communication nodes includes a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal includes at least one sounding reference signal (SRS); receiving, at the wireless communication device, a second signal from the first wireless communication node, wherein the second signal is determined based on the first signal, and the second signal includes: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

[0006] In some embodiments, the second signal is determined from a plurality of Layer 1 Reference Signal Received Power (Ll-RSRP) values using a machine learning (ML) model, wherein each of the plurality of Ll-RSRP values is measured from the first signal at a respective one of the first plurality of wireless communication nodes. [0007] In some embodiments, the ML model is trained using reported CQI from the wireless communication device, wherein the reported CQI is used as a ground truth for training the ML model. In some embodiments, transmissions of the first signal are triggered by radio resource management (RRM) measurement reports transmitted from the wireless communication device to the first wireless communication node, wherein the RRM measurement reports include RRM measurements, wherein the RRM measurements include at least one of: a reference signal received power (RSRP) value; a reference signal received quality (RSRQ) value; and a signal -to-interference-plus-noise ratio (SINR) value.

[0008] In some embodiments, the first signal is transmitted in a time slot during which each of a plurality of wireless communication devices transmits a respective SRS to the first plurality of wireless communication nodes. In some embodiments, the first signal is transmitted based on an SRS resource allocation, wherein the SRS resource allocation is determined based on assistance information of the wireless communication device, wherein the assistance information includes at least one of: a trajectory of the wireless communication device; a speed of the wireless communication device; a movement direction of the wireless communication device; and a location of the wireless communication device.

[0009] In some embodiments, the indication used to indicate the neighboring cell to which the wireless communication device should hand over is determined based on each of a plurality of handover probabilities associated with a respective one of the first plurality of wireless communication nodes, wherein the plurality of handover probabilities is determined by the ML model.

BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Various exemplary embodiments of the present disclosure are described in detail below with reference to the following figures. The drawings are provided for purposes of illustration only and merely depict exemplary embodiments of the present disclosure to facilitate the reader's understanding of the present disclosure. Therefore, the drawings should not be considered limiting of the breadth, scope, or applicability of the present disclosure. It should be noted that for clarity and ease of illustration these drawings are not necessarily drawn to scale.

[0011] FIG. 1 A illustrates an exemplary wireless communication network, in accordance with some embodiments of the present disclosure.

[0012] FIG. IB illustrates a block diagram of an exemplary wireless communication system, in accordance with some embodiments of the present disclosure.

[0013] FIG. 2 A illustrates a signaling diagram between a BS and a UE for spatial domain beam prediction with a machine learning model implemented on the BS side, in accordance with some embodiments.

[0014] FIG. 2B illustrates another signaling diagram between a BS and a UE for spatial domain beam prediction with a machine learning model implemented on the UE side, in accordance with some embodiments.

[0015] FIG. 2C illustrates yet another signaling diagram between a BS and a UE for temporal domain beam prediction with a machine learning model implemented on the BS side, in accordance with some embodiments.

[0016] FIG. 2D illustrates still another signaling diagram between a BS and a UE for temporal domain beam prediction with a machine model implemented on the UE side, in accordance with some embodiments. [0017] FIG. 3 illustrates an exemplary design framework for computing CQI based on

RSRP reporting from neighboring cells, in accordance with some embodiments of the present disclosure.

[0018] FIG. 4 illustrates another exemplary design framework for computing handover probabilities based on RSRP reporting from neighboring cells, in accordance with some embodiments of the present disclosure.

[0019] FIG. 5 illustrates a signaling diagram for SRS resource determination, in accordance with some embodiments of the present disclosure.

[0020] FIG. 6 illustrates an exemplary time-frequency resource allocation diagram for SRS transmission, in accordance with some embodiments of the present disclosure.

[0021] FIG. 7 illustrates a deep neural network (DNN) model used to implement a machine learning model, in accordance with some embodiments.

[0022] FIG. 8 illustrates an example method for CQI determination and handover decision, in accordance with some embodiments.

DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

[0023] Various exemplary embodiments of the present disclosure are described below with reference to the accompanying figures to enable a person of ordinary skill in the art to make and use the present disclosure. As would be apparent to those of ordinary skill in the art, after reading the present disclosure, various changes or modifications to the examples described herein can be made without departing from the scope of the present disclosure. Thus, the present disclosure is not limited to the exemplary embodiments and applications described and illustrated herein. Additionally, the specific order and/or hierarchy of steps in the methods disclosed herein are merely exemplary approaches. Based upon design preferences, the specific order or hierarchy of steps of the disclosed methods or processes can be re-arranged while remaining within the scope of the present disclosure. Thus, those of ordinary skill in the art will understand that the methods and techniques disclosed herein present various steps or acts in a sample order, and the present disclosure is not limited to the specific order or hierarchy presented unless expressly stated otherwise.

[0024] Figure 1A illustrates an exemplary wireless communication network 100, in accordance with some embodiments of the present disclosure. In a wireless communication system, a network side communication node or a base station (BS) 102 can be a node B, an E-UTRA Node B (also known as Evolved Node B, eNodeB or eNB), a New Generation eNB (ng-eNB), a gNodeB (also known as gNB) in new radio (NR) technology, a pico station, a femto station, or the like. A terminal side communication device or a user equipment (UE) 104 can be a long range communication system like a mobile phone, a smart phone, a personal digital assistant (PDA), tablet, laptop computer, or a short range communication system such as, for example a wearable device, a vehicle with a vehicular communication system and the like. A network communication node and a terminal side communication device are represented by a BS 102 and a UE 104, respectively, and in all the embodiments in this disclosure hereafter, and are generally referred to as “communication nodes” and “communication device” herein. Such communication nodes and communication devices may be capable of wireless and/or wired communications, in accordance with various embodiments of the invention. It is noted that all the embodiments are merely preferred examples, and are not intended to limit the present disclosure. Accordingly, it is understood that the system may include any desired combination of communication nodes and communication devices, while remaining within the scope of the present disclosure. [0025] Referring to Figure 1 A, the wireless communication network 100 includes a first

BS 102-1, a second BS 102-2, a first UE 104-1, a second UE 104-2, a third UE 104-3, and a fourth UE 104-4. In some embodiments, the first BS 102-1 and the second BS 102-2 comprise a first plurality of antennas 106-1 to 106-n and a second plurality of antennas 116-1 to 116-n’, respectively. The first plurality of antennas 106-1 to 106-n may communicate with a plurality of UEs 104 to form a first multiple-input multiple-output (MIMO) system, and the second plurality of antennas 116-1 to 116-n’ may communicate with the plurality of UEs 104 to form a second MIMO system.

[0026] In some embodiments, a plurality of UEs 104 may form direct communication (e.g., uplink) channels 103-1, 103-2, 103-3, and 103-4 with the first BS 102-1 and the second BS 102-2. In some embodiments, the plurality of UEs 104 may also form direct communication (e.g., downlink) channels 105-1, 105-2, 105-3, and 105-4 with the first BS 102-1 and the second BS 102-2. The direct communication channels between the plurality of UEs 104 and a distributed unit of the BS 102 can be through interfaces such as an Uu interface, which is also known as E-UTRAN air interface. In some embodiments, the UE 104 comprises a plurality of transceivers which enables the UE 104 to support multi connectivity so as to receive data simultaneously from the first BS 102-1 and the second BS 102-2. The first BS 102-1 and the second BS 102-2 each is connected to a core network (CN) 108 on a user plane (UP) through an external interface 107, e.g., an lu interface, an NG-U interface, or an Sl-U interface. In some embodiments, the CN 108 is one of the following: an Evolved Packet Core (EPC) and a 5G Core Network (5GC). In some embodiments, the CN 108 further comprises at least one of the following: Access and Mobility Management Function (AMF), User Plane Function (UPF), and System Management Function (SMF). [0027] A direct communication channel 111 between the first BS 102-1 and the second

BS 102-2 is through an Xn interface, in accordance with some embodiments. In some embodiments, a BS (e.g., a gNB) is split into a Distributed Unit (DU) and a Central Unit (CU) on the UP, between which the direct communication is through a Fl-U interface. In some embodiments, a CU of the second BS 102-2 can be further split into a Control Plane (CP) and a User Plane (UP), between which the direct communication is through an El interface. Hereinafter, in the present disclosure, an Xx interface is used to describe one of the following interfaces, the NG interface, the SI interface, the X2 interface, the Xn interface, the Fl interface, and the El interface. When an Xx interface is established between two nodes, the two nodes can transmit control signaling on the CP and/or data on the UP.

[0028] Figure IB illustrates a block diagram of an exemplary wireless communication system 150, in accordance with some embodiments of the present disclosure. The system 150 may include components and elements configured to support known or conventional operating features that need not be described in detail herein. In some embodiments, the system 150 can be used to transmit and receive data symbols in a wireless communication environment such as the wireless communication network 100 of Figure 1 A, as described above.

[0029] The system 150 generally includes a first BS 102-1, a second BS 102-2, and a UE 104, collectively referred to as BS 102 and UE 104 below for ease of discussion. The first BS 102-1 and the second BS 102-2 each comprises a BS transceiver module 152, a BS antenna array 154, a BS memory module 156, a BS processor module 158, and a network interface 160. In the illustrated embodiment, each module of the BS 102 is coupled and interconnected with one another as necessary via a data communication bus 180. The UE 104 comprises a UE transceiver module 162, a UE antenna 164, a UE memory module 166, a UE processor module 168, and an I/O interface 169. In the illustrated embodiment, each module of the UE 104 is coupled and interconnected with one another as necessary via a date communication bus 190. The BS 102 communicates with the UE 104 via a communication channel 192, which can be any wireless channel or other medium known in the art suitable for transmission of data as described herein.

[0030] As would be understood by persons of ordinary skill in the art, the system 150 may further include any number of modules other than the modules shown in Figure IB. Those skilled in the art will understand that the various illustrative blocks, modules, circuits, and processing logic described in connection with the embodiments disclosed herein may be implemented in hardware, computer-readable software, firmware, or any practical combination thereof. To clearly illustrate this interchangeability and compatibility of hardware, firmware, and software, various illustrative components, blocks, modules, circuits, and steps are described generally in terms of their functionality. Whether such functionality is implemented as hardware, firmware, or software depends upon the particular application and design constraints imposed on the overall system. Those familiar with the concepts described herein may implement such functionality in a suitable manner for each particular application, but such implementation decisions should not be interpreted as limiting the scope of the present invention.

[0031] A wireless transmission from a transmitting antenna of the UE 104 to a receiving antenna of the BS 102 is known as an uplink (UL) transmission, and a wireless transmission from a transmitting antenna of the BS 102 to a receiving antenna of the UE 104 is known as a downlink (DL) transmission. In accordance with some embodiments, the UE transceiver 162 may be referred to herein as an “uplink” transceiver 162 that includes a radio frequency (RF) transmitter and receiver circuitry that is each coupled to the UE antenna 164. A duplex switch (not shown) may alternatively couple the uplink transmitter or receiver to the uplink antenna in time duplex fashion. Similarly, in accordance with some embodiments, the BS transceiver 152 may be referred to herein as a “downlink” transceiver 152 that includes RF transmitter and receiver circuitry that are each coupled to the antenna array 154. A downlink duplex switch may alternatively couple the downlink transmitter or receiver to the downlink antenna array 154 in time duplex fashion. The operations of the two transceivers 152 and 162 are coordinated in time such that the uplink receiver is coupled to the uplink UE antenna 164 for reception of transmissions over the wireless communication channel 192 at the same time that the downlink transmitter is coupled to the downlink antenna array 154. Preferably, there is close synchronization timing with only a minimal guard time between changes in duplex direction. The UE transceiver 162 communicates through the UE antenna 164 with the BS 102 via the wireless communication channel 192. The BS transceiver 152 communications through the BS antenna 154 of a BS (e.g., the first BS 102-1) with the other BS (e.g., the second BS 102-2) via a wireless communication channel 196. The wireless communication channel 196 can be any wireless channel or other medium known in the art suitable for direct communication between BSs.

[0032] The UE transceiver 162 and the BS transceiver 152 are configured to communicate via the wireless data communication channel 192, and cooperate with a suitably configured RF antenna arrangement 154/164 that can support a particular wireless communication protocol and modulation scheme. In some exemplary embodiments, the UE transceiver 162 and the BS transceiver 152 are configured to support industry standards such as the Long Term Evolution (LTE) and emerging 5G standards (e.g., NR), and the like. It is understood, however, that the invention is not necessarily limited in application to a particular standard and associated protocols. Rather, the UE transceiver 162 and the BS transceiver 152 may be configured to support alternate, or additional, wireless data communication protocols, including future standards or variations thereof.

[0033] The processor modules 158 and 168 may be implemented, or realized, with a general purpose processor, a content addressable memory, a digital signal processor, an application specific integrated circuit, a field programmable gate array, any suitable programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof, designed to perform the functions described herein. In this manner, a processor module may be realized as a microprocessor, a controller, a microcontroller, a state machine, or the like. A processor module may also be implemented as a combination of computing devices, e.g., a combination of a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a digital signal processor core, or any other such configuration.

[0034] Furthermore, the steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in firmware, in a software module executed by processor modules 158 and 168, respectively, or in any practical combination thereof. The memory modules 156 and 166 may be realized as RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. In this regard, the memory modules 156 and 166 may be coupled to the processor modules 158 and 168, respectively, such that the processors modules 158 and 168 can read information from, and write information to, memory modules 156 and 166, respectively. The memory modules 156 and 166 may also be integrated into their respective processor modules 158 and 168. In some embodiments, the memory modules 156 and 166 may each include a cache memory for storing temporary variables or other intermediate information during execution of instructions to be executed by processor modules 158 and 168, respectively.

The memory modules 156 and 166 may also each include non-volatile memory for storing instructions to be executed by the processor modules 158 and 168, respectively.

[0035] The network interface 160 generally represents the hardware, software, firmware, processing logic, and/or other components of the base station 102 that enable bi-directional communication between BS transceiver 152 and other network components and communication nodes configured to communication with the BS 102. For example, network interface 160 may be configured to support internet or WiMAX traffic. In a typical deployment, without limitation, network interface 160 provides an 802.3 Ethernet interface such that BS transceiver 152 can communicate with a conventional Ethernet based computer network. In this manner, the network interface 160 may include a physical interface for connection to the computer network (e.g., Mobile Switching Center (MSC)). The terms “configured for” or “configured to” as used herein with respect to a specified operation or function refers to a device, component, circuit, structure, machine, signal, etc. that is physically constructed, programmed, formatted and/or arranged to perform the specified operation or function. The network interface 160 could allow the BS 102 to communicate with other BSs or a CN over a wired or wireless connection.

[0036] Referring again to Figure 1 A, as mentioned above, the BS 102 repeatedly broadcasts system information associated with the BS 102 to one or more UEs 104 so as to allow the UEs 104 to access the network within the cells where the BS 102 is located, and in general, to operate properly within the cell. Plural information such as, for example, downlink and uplink cell bandwidths, downlink and uplink configuration, cell information, configuration for random access, etc., can be included in the system information. Typically, the BS 102 broadcasts a first signal carrying some major system information, for example, configuration of the cell where the BS 102 is located through a Physical Broadcast Channel (PBCH). For purposes of clarity of illustration, such a broadcasted first signal is herein referred to as “first broadcast signal.” It is noted that the BS 102 may subsequently broadcast one or more signals carrying some other system information through respective channels (e.g., a Physical Downlink Shared Channel (PDSCH)).

[0037] Referring again to Figure IB, in some embodiments, the major system information carried by the first broadcast signal may be transmitted by the BS 102 in a symbol format via the communication channel 192 (e.g., a PBCH). In accordance with some embodiments, an original form of the major system information may be presented as one or more sequences of digital bits and the one or more sequences of digital bits may be processed through plural steps (e.g., coding, scrambling, modulation, mapping steps, etc.), all of which can be processed by the BS processor module 158, to become the first broadcast signal. Similarly, when the UE 104 receives the first broadcast signal (in the symbol format) using the UE transceiver 162, in accordance with some embodiments, the UE processor module 168 may perform plural steps (de-mapping, demodulation, decoding steps, etc.) to estimate the major system information such as, for example, bit locations, bit numbers, etc., of the bits of the major system information. The UE processor module 168 is also coupled to the I/O interface 169, which provides the UE 104 with the ability to connect to other devices such as computers. The I/O interface 169 is the communication path between these accessories and the UE processor module 168.

[0038] Referring again to Figure 1 A, during the transmission of signals between the BS 102 and the UE 104, the established wireless transmission channels between the BS 102 and the UE 104 may introduce various impairments and distortions to the transmitted signals due to factors such as fading, interference, and noise. Channel estimation can be performed to estimate the characteristics of the communication channel between the BS 102 and the UE

104 to optimize wireless communication system performance and to improve the reliability of communication. A conventional way to perform channel estimation is to use channel reciprocity for MIMO precoding in the downlink by estimating the UL channel based on the symmetry properties between the UL and DL channels. That is, the UE 104 can periodically transmit pilot signals or Sounding Reference Signals (SRSs) during specific time slots allocated for UL channel sounding, and the corresponding BS 102 can measure the received SRSs to estimate the UL channel characteristics such as channel gains and phases. In case of a time-division duplexing (TDD) transmission, there is channel reciprocity between the UL and DL channels. This means that the UL and DL channel responses are related, allowing information obtained from UL measurements to be used for DL transmission. For example, the BS 102 can perform DL MIMO precoding based on the extracted UL channel state information (CSI) from the received pilot signals or SRSs. Once the DL MIMO precoding matrix is determined, the BS 102 can use it to precode the DL data transmission, which helps in mitigating the effects of channel fading and interference and improving the quality of the received signal at the UEs.

[0039] In some embodiments, a massive MIMO system is used with a number of antennas at the BS 102 to enhance data throughput and spectrum efficiency. However, the implementation of MIMO systems necessitates accurate CSI acquisition at the BS and UE transmitters, which can be achieved through codebook-based feedback in Frequency Division Duplexing (FDD) networks or reciprocity-based sounding in Time Division Duplexing (TDD) networks. Despite these methods, the exhaustive beam search required for selecting optimal transmit-receive (Tx-Rx) beam pairs in 5GNew Radio (NR) technology results in significant signaling overhead and delays, thus necessitating more efficient approaches. [0040] A significant milestone in the progression of 5G NR technology is the introduction of 5G- Advanced, initially delineated in the 3rd Generation Partnership Project (3GPP) release 18. Within the framework of 5G- Advanced, one prominent aspect involves the integration of Artificial Intelligence (Al) leveraging Machine Learning (ML) techniques to provide solutions across various use cases, including enhancements in Channel State Information (CSI), Beam Management (BM), and positioning accuracy. In contrast to conventional methods, Al-based approaches harness ML techniques, particularly Neural Networks (NNs), to extract features from training data. Consequently, AI/ML-based algorithms can be effectively employed in BM to mitigate overhead and enhance beam prediction accuracy, marking a departure from traditional approaches.

[0041] In line with the 3GPP design for 5G-Advanced, two representative BM sub-use cases are considered, namely Spatial Domain Beam Prediction (BM Case-1) and Temporal Domain Beam Prediction (BM Case-2). In the Spatial Domain Beam Prediction (BM Case-1) sub-use case, the prediction of the spatial domain DL beam is conducted by providing an element of the prediction Set c/Z, relying on measurements provided by the measurement Set B. Within this context, 3GPP has defined two alternatives for the relationship between sets c/Z and B

[0042] Alternative 1 : Set B being a subset of Set c/Z. In this scenario, both sets c/Z and B comprise narrow beams. For example, the prediction Set c/Z and the measurement Set B could be the beams in Channel State Information-Reference Signal (CSLRS) resources. In one embodiment, Set c/Z comprises 100 narrow beams covering the entire spatial area around the UE, and Set B is a smaller subset of 10 beams from Set c/Z. Then the UE may measure signal quality such as Layer 1 Reference Signal Received Power (Ll-RSRP) on each beam in Set B, then based on these measurements, an ML model may be configured to predict which of the

100 beams in Set Jl (including those not directly measured) would be optimal for the UE.

[0043] Alternative 2: Set B being different from Set c/Z. Alternatively, Set B could differ from Set A while there is no overlap between Set B and Set A. In this case, Set c/Z may comprise Narrow Beams (NBs) accompanying CSI-RS resources while Set B may comprise Wide Beams (WBs) on Synchronization Signal Block (SSB) resources. In one embodiment, Set c/Z comprises 100 narrow DL beams for precise communication, while Set B comprises fewer, wider beams (e.g., 10 wide beams from the SSB that provide coarse location data). The UE may then measure the signal quality (e.g. Ll-RSRP) on each of the wider beams in Set B. Next, the ML model may use the measurements from these 10 wide beams in Set B to predict the best narrow beam(s) in Set c/Z.

[0044] In the Temporal Domain Beam Prediction (BM Case-2) sub-use case, the temporal prediction of the DL beam for N future time instances involves predicting beams from Set c/Z based on inputs from the measurement Set B. This scenario parallels BM Case-1, where Set B could either be a subset of Set c/Z or different from Set c/Z. Furthermore, in BM Case-2, an additional option exists where Set B could be identical to Set c/Z but obtained at a different time (e.g. an earlier time) from Set c/Z. In one embodiment, a BS communicates with a moving UE (such as a vehicle) in a dynamic environment. The BS can then use an ML model to predict which beams in Set c/Z will be optimal over the next few time instances (i.e., N future time instances) based on current or past measurements from Set B. In another embodiment, Set c/Z comprises 100 narrow beams covering the entire area, while Set B comprises 10 beams selected from within Set c/Z that cover the region around the UE’s current position. At the current time (time 0), the UE may measure Ll-RSRP values for the 10 beams in Set B and reports them to the BS, and the ML model may use the reported measurements from Set B to predict which beam(s) in Set Jl will be optimal for the UE at N future time instances (e.g., time 1, time 2, etc.). In yet another embodiment, Set c/Z comprises narrow beams (e.g., 100 narrow DL beams for precise coverage), while Set B comprises a smaller set of wider beams (e.g., 5 wide beams from SSB signals) that provide broader, coarse-grained measurements. At the current time (time 0), the UE may measure Ll-RSRP values for the 10 wider beams in Set B and reports them to the BS, and the ML model at the BS may use the reported measurements from Set B to predict which narrow beam(s) in Set c/Z will be optimal for the UE at N future time instances. In still another embodiment, both Set c/Z and Set B comprise the same set of 100 beams, but the measurements in Set B are taken at an earlier time. For example, at time point T — 1, the UE may measure the Ll-RSRP values on all the 100 beams in Set B (which is the same as Set c/Z). Then using these past measurements from time point T — 1, the ML model can predict which of the beams in Set Jl will be optimal at N future time points: T + 1, . . . , T + N.

[0045] In both BM Case-1 and BM Case-2, 3 GPP delineated two options for the placement of the ML model. The first option involves the placement of the model on the BS (e.g. gNB) side, while the second option entails the placement of the ML model on the UE side. In some embodiments, when the ML model is placed/implemented on the BS side, the ML model can be a software program that is executed by the BS based on data stored and/or received at the BS. In some other embodiments, the ML model can be a hardware component (e.g. Field-Programmable Gate Array (FPGA) or Application-Specific Integrated Circuit (ASIC)) that is embedded in the BS hardware. In yet some other embodiments, when the ML model is placed/implemented on the UE side, the ML model can be a software program that is executed by the UE based on data stored and/or received at the UE. In still some other embodiments, the ML model can be a hardware component (e.g. FPGA or ASIC) that is embedded in the UE hardware.

[0046] FIG. 2 A illustrates a signaling diagram between a BS 202 and a UE 204 for spatial domain beam prediction with ML model placed on the BS side, in accordance with some embodiments. In some embodiments, the BS 202 may transmit one or more reference signals such as CSLRSs or Synchronization Signals (SSs) to the UE 204. The one or more reference signals may be used by the UE 204 to measure the received signal quality and strength across different beams. In some embodiments, the one or more reference signals are transmitted across multiple antenna arrays or beams, wherein each of the multiple antenna arrays or beams corresponds to a different spatial direction, allowing the UE 204 to assess the quality of each beam.

[0047] Upon receiving the one or more reference signals, the UE 204 may be configured to measure the Layer 1 Reference Signal Received Power (Ll-RSRP) for each received beam or group of beams, wherein the measured Ll-RSRP reflects the power and quality of the signals received from the BS 202 across different beams. In some embodiments, the UE 204 compiles different Reference Signal Received Power (RSRP) values into a Ll-RSRP report, which is a low-level report directly related to signal strength. Then the UE 204 may send the Ll-RSRP report back to the BS 202.

[0048] In some embodiments, upon receiving the Ll-RSRP report, the BS 202 may use an ML model 206 placed within the BS 202 to perform beam prediction from input Set B to output Set c/Z, wherein Set c/Z represents the predicted optimal beams used for future communication, and Set B may be a specific subset of narrow beams that the UE 204 has measured and reported in the Ll-RSRP report. In some embodiments, the ML model 206 may be a software program executed by a processor within the BS 202, wherein the processor is part of the BS 202’ s main computing unit. In some other embodiments, the ML model 206 may be a dedicated hardware component (e.g. FPGA or ASIC or GPU) implemented within the BS 202. In yet some other embodiments, the ML model 206 may be located outside the BS 202, wherein the ML model 206 is connected to the BS 202 via a wired or wireless connection. For instance, the ML model 206 may be hosted on a central server or cloud-based platform, communicating with the BS 202 to receive Set B and to return predictions from Set c/Z. In some embodiments, the data in Set c/Z and Set B may be stored in a memory (e.g., random-access memory) within the BS 202. The specific subset may represent the beams with the best signal strength in the Ll-RSRP report, or the beams that are expected to provide reliable communication based on historical performances. In some embodiments, Set B is a subset of Set c/Z. In this case, the prediction of Set c/Z and the measurements in Set B may be the beams in the CSLRS resources used in the communication. The ML model 206 may be configured to analyze the Ll-RSRP report from Set B (e.g. the subset of narrow beams the UE 204 evaluated). The BS 202 may then use the ML model 206 to perform model inference and predict which beams in Set c/Z (e.g. the broader set of beams) will be optimal for the upcoming data transmissions. Based on these predictions, the BS 202 can dynamically select the best beams from the full set (Set c/Z) for future transmissions. This improves efficiency by narrowing the choices down to the most relevant beams.

[0049] In some embodiments, the ML model 206 may be a separate entity that is connected to the BS 202 via wireless or wired connection. In some other embodiments, Set B comprises only Ll-RSRP measurements used as inputs of the ML model 206 for performing beam predictions, wherein the Ll-RSRP measurements may be stored in a memory of the UE and used as inputs for the ML model implemented on the UE side. In case the ML model is implemented on the BS side, the Ll-RSRP measurements may be reported from the UE to the BS, and then the Ll-RSRP measurements may serve as inputs of the ML model at the BS. In some embodiments, Set B may be obtained on the UE side, wherein the UE is configured to take Ll-RSRP measurements on a subset of (e.g. selected based on spatial beam patterns, previous history, or channel conditions) or all available beams from the BS, wherein the Ll- RSRP measurements form Set B. In yet some other embodiments, Set B comprises Ll-RSRP measurements and assistance information (such as UE location, mobility patterns, historical data, and environment-specific parameters) used as inputs of the ML model 206 to improve beam prediction accuracy. In still some other embodiments, Set B comprises Channel Impulse Response (CIR) used as inputs of the ML model 206. In still some other embodiments, the inputs of the ML model 206 comprise Ll-RSRP measurements from Set B combined with the corresponding downlink transmit (Tx) and/or receive (Rx) beam IDs.

[0050] In some embodiments, Set B is different from Set c/Z, and Set c/Z may be used for DL beam prediction. The codebook construction of Set c/Z and Set B may be clarified by users or companies. In this case, Set c/Z may comprise Narrow Beams (NBs) accompanying CSLRS resources while Set B comprises Wide Beams (WBs) on Synchronization Signal Block (SSB) resources. That is, the ML model 206 may use the Ll-RSRP report based on wide beam measurements to predict which narrow beams (Set c/Z, CSLRS beams) will provide optimal performance for the UE 204. Therefore, the wide beams may serve as a precursor for determining the more granular, directional narrow beams that the BS 202 will use for data transmission. The ML model 206 may then link the wide beam performance (from Set B) to predict the best narrow beams (Set c/Z), reducing the need for an exhaustive beam sweep of all narrow beams.

[0051] FIG. 2B illustrates another signaling diagram between a BS 212 and a UE 214 for spatial domain beam prediction with ML model placed on the UE side, in accordance with some embodiments. In some embodiments, the ML model 216 may be a software program executed by a processor within the UE 214, wherein the processor is part of the UE 214’s main computing unit. In some other embodiments, the ML model 216 may be a dedicated hardware component (e.g. FPGA or ASIC or GPU) implemented within the UE 214. In yet some other embodiments, the ML model 216 may be located outside the UE 214, wherein the ML model 216 is connected to the UE 214 via a wired or wireless connection. For instance, the ML model 216 may be hosted on a central server or cloud-based platform, communicating with the UE 214 to receive Set B and to return predictions from Set c/Z. In some embodiments, the data in Set c/Z and Set B may be stored in a memory (e.g., randomaccess memory) within the UE 214. In some embodiments, the BS 212 may transmit one or more reference signals such as CSLRSs or SSs to the UE 214. Upon receiving the one or more reference signals, the UE 214 may be configured to generate an Ll-RSRP report as discussed above with reference to FIG. 2A. In some embodiments, instead of transmitting the Ll-RSRP report back to the BS 212, the UE 214 may use the Ll-RSRP report along with assistance information to perform beam prediction based on an ML model 216 placed in the UE 214. Examples of assistance information that can be used for beam prediction with the ML model 216 include: Channel State Information (CSI), Channel Quality Indicators (CQI), Signal-to-Noise Ratio (SNR), beam indices and measurement history, UE location and mobility information, time-domain information of the network. The ML model 216 may be a component placed within the UE 214, or implemented in a different entity that is connected to the UE 214 via wireless or wired connection. In some embodiments, the ML model 216 may use the Ll-RSRP report along with the assistance information to perform model training and beam prediction, as described above with reference to FIG. 2A.

[0052] FIG. 2C illustrates yet another signaling diagram between a BS 222 and a UE 224 for temporal domain beam prediction with ML model placed on the BS side, in accordance with some embodiments. In some embodiments, the BS 222 may transmit one or more reference signals such as CSI-RSs or SSBs to the UE 224. Upon receiving the one or more reference signals, the UE 224 may be configured to generate a historical Ll-RSRP report and send the historical Ll-RSRP report back to the BS 222. In some embodiments, the historical Ll-RSRP report may include historical Ll-RSRP measurements, UE mobility patterns including UE location information and UE speed, beam performance history including beam ID and beam switching events, and historical channel condition data including signal to noise and CQI.

[0053] In some embodiments, upon receiving the historical Ll-RSRP report, the BS 222 may use an ML model 226 placed within the BS 222 to perform beam prediction from input Set B to output Set c/Z. In some embodiments, the ML model 226 is implemented in the BS 222 in a way similar to the ML model 206 described in FIG. 2A. In some embodiments, Set c/Z and Set B are different and Set B is not a subset of Set c/Z. In this case, Set c/Z may comprise a distinct set of beams used for downlink beam prediction, while Set B may comprise a separate set of beams used for measurements. The ML model 226 may then use measurements from Set B to predict the optimal beams in Set c/Z. In some other embodiments, Set B is a subset of Set c/Z. In this case, Set c/Z may comprise all the beams for downlink prediction, while Set B is a smaller subset of Set c/Z. The ML model 226 may use measurements from the narrower Set B to make predictions for beams in Set c/Z. In yet some other embodiments, Set c/Z and Set B comprise the same measurements obtained at different times. In this case, Set B may comprise historical beam measurements, and Set c/Z may comprise predicted future values for the same beam measurements from Set B.

[0054] In some embodiments, the ML model 226 may use measurement data from the latest K measurement instances (where K > 1) with the following input alternatives: Alternative 1 : The ML model 226 uses only Ll-RSRP values measured from Set B as inputs to make predictions on Set JI. Alternative 2: along with Ll-RSRP values, the ML model 226 incorporates additional assistance information (such as UE location, mobility data, or environmental conditions) to enhance prediction accuracy. Alternative 3: The Ll-RSRP measurements from Set B are combined with information on the corresponding downlink Tx and/or Rx beam IDs as model inputs to improve the prediction accuracy of the ML model 226. In some embodiments, the ML model 226 is configured to generate predictions for F future time instances (F > 1). For each future time instance, the ML model 226 may provide a prediction for the optimal beam configuration, which can be used to assist the UE 224 for anticipating changes in beam performance based on the input measurements. This approach provides dynamic prediction of beam performance over time, and the beamforming strategies can be adjusted accordingly to optimize connectivity.

[0055] FIG. 2D illustrates still another signaling diagram between a BS 232 and a UE 234 for temporal domain beam prediction with an ML model 236 placed on the UE side, in accordance with some embodiments. In some embodiments, the ML model 236 is implemented in the UE 234 in a way similar to the ML model 216 described in FIG. 2B. In some embodiments, the BS 232 may transmit one or more reference signals such as CSLRSs or SSBs to the UE 234. Upon receiving the one or more reference signals, the UE 234 may be configured to generate a historical Ll-RSRP report. In some embodiments, the historical Ll- RSRP report may be generated along with assistance information as discussed above with reference to FIG. 2C. In some embodiments, instead of transmitting the historical Ll-RSRP report back to the BS 232, the UE 234 may use the historical Ll-RSRP report along with the assistance information to perform beam prediction based on an ML model 236 placed in the UE 234. The ML model 236 may be a component placed within the UE 234, or implemented in a different entity that is connected to the UE 234 via wireless or wired connection. The ML model 236 may use the historical Ll-RSRP report along with the assistance information to perform model training and beam prediction, as described above with reference to FIG. 2C.

[0056] In some embodiments, the Ll-RSRP measurements from Set B are used as the primary inputs of the ML model used for beam prediction, wherein the Ll-RSRP measurements are discrete digital values converted from continuous RSRP values in a process named RSRP quantization. In some embodiments, 7 bits are utilized to report the absolute Ll-RSRP of each the strongest, or all, beams in Set B as measured by the UE. Alternatively, other embodiments may employ differential Ll-RSRP reporting using only 4 bits for each beam to reduce the uplink reporting overhead. This RSRP quantization strategy can be used to balance precision with the need to minimize overhead in the wireless communication system. A more precise quantization provides better information for the ML model but also increases the amount of data that needs to be processed. That is, the precision of the RSRP quantization can impact the performance of the trained ML model. Higher precision allows the ML model to make more accurate predictions but comes at the cost of increased data and computational overhead. In some embodiments, an optimal trade-off between precision and efficiency can be obtained to maximize model performance while minimizing system burden. In some embodiments, one can set a maximum acceptable computational overhead (e.g. 1 ms) for the ML model. Then the number of bits in the RSRP quantization can be incremented starting from 2 bits while the computational overhead of the ML model is measured each time the number of bits in the RSRP quantization is increased. The final number of bits in the RSRP quantization may be determined to be the maximum number of bits in this process while the computational overhead is still within the acceptable value (e.g. 1 ms). [0057] In some embodiments, the problem of determining the DL beam for a BS side model can be addressed in a Single User Multiple Input Multiple Output (SU-MIMO) scenario. That is, instead of the BS transmitting a beam Set B to the UE and the UE reporting corresponding Ll-RSRP measurements, an UL-based approach can be employed where the UE transmits Sounding Reference Signals (SRSs) using a beam set B. The BS may then collect measurements from the received signals, such as Ll-RSRP or CIR, and utilizes these measurements as inputs to an ML model for precoder calculation. A primary advantage of this approach is the elimination of the requirement for the UE to transmit quantized measurements prone to errors when received by the BS. However, this approach is limited to a specific scenario involving a single UE with a single BS (e.g. the serving BS), without considering the impact of intercell interference. The present disclosure extends the scope of the aforementioned UL-based approach by addressing scenarios where neighboring (interfering) BSs can also collect SRS measurements from the UE. Each neighboring BS may then report its measured Ll-RSRP of the transmitted SRS to the serving BS. These Ll-RSRP measurements comprise information regarding signal attenuation from each neighboring BS to the target UE, which can be used to derive a CQI for the serving BS. To model this relationship, the serving BS may compile a set of diverse Ll-RSRP values, including its own local measurement, and inputs these into a trained ML model that estimates the CQI value.

[0058] From the AI/ML mobility perspective, both UE-sided model and network (NW)- sided (e.g. BS-sided) model are considered in Rel-19 for Radio Resource Management (RRM) measurement predictions, such as signal strength predictions. However, for predicting events such as handover (HO) failure or radio link failure (RLF), only the UE-side model is used. Event prediction can also include measurement events, such as Event A3, which helps trigger UE measurement reporting when a signal quality (e.g. RSRP, Reference Signal Received Quality (RSRQ), or Signal-to-Interference-plus-Noise Ratio (SINR)) of a neighboring cell becomes offset (for example, offset can be 3dB) better than that of the Serving Primary Cell (SpCell).

[0059] In some embodiments, under 5G NR framework, a serving BS is in communication with a plurality of UEs. To schedule (e.g. to assign specific radio resources such as time and/or frequency resources) the plurality of UEs, the plurality of UEs need to report their CQI values to the serving BS for computing the downlink transmission’s Modulation and Coding Scheme (MCS). In some embodiments, a ML-based method can be employed to enable the serving BS to reconstruct the DL channel by receiving SRS transmissions from the plurality of UEs. In some embodiments, the ML-based method is tailored for the SU-MIMO case. In the context of Multi-User (MU) MIMO with intercell interference, the plurality of UEs may transmit their CQI values to the serving BS, allowing the serving BS to manage intercell interference and optimize resource allocation. The transmitted CQI values may be obtained through conventional CSI frameworks or alternative methods. However, in the aforementioned ML-based method, it is deemed undesirable for the plurality of UEs to perform additional uplink transmissions beyond the SRS transmissions.

[0060] FIG. 3 illustrates an exemplary design framework 300 for computing CQI based on RSRP reporting from neighboring cells, in accordance with some embodiments of the present disclosure. In some embodiments, a plurality of UEs may be in communication with a plurality of BSs, wherein the plurality of BSs may comprise JVBSS denoted by 302-1 to 302- N as shown in FIG. 3. Each of the plurality of BSs 302-1 to 302-N may provide coverage to a corresponding cell. The CQI values (e.g. SINR values) for the plurality of UEs may be computed using an ML model 306 instead of relying solely on UE-reported CQI values. In some embodiments, the BS 302-TVis in communication with one of the plurality of UEs (for example the Uth UE), wherein the BS 302-N may be denoted by the “serving cell”, and the BSs 302-1 to 302-7V-7 may be denoted by “neighboring cells”. In some embodiments, the serving cell (e.g. the BS 302-7V) may schedule the SRS transmission for the Uth UE and coordinate with the neighboring cells 302-1 to 302-7V-7 to perform measurements at the SRS resources allocated for the Uth UE. In some embodiments, each of the neighboring cells 302- 1 to 302-7V-7 may be configured to perform measurements on the allocated SRS resources and report the associated Ll-RSRP value back to the serving cell 302-7V, wherein the reported Ll-RSRP values along with the Ll-RSRP value of the serving cell are used as inputs to the ML model 306 to generate predicted CQI values (such as SINR values) for the plurality of UEs at the outputs of the ML model 306. The reported Ll-RSRP from each of the neighboring cells can be considered as the proxy for the intercell interference produced by the corresponding neighboring cell at the specific UE (e.g. the Uth UE). In some other embodiments, after each of the neighboring cells 302-1 to 302-7V-7 performs measurements on the allocated SRS resources, the associated Ll-RSRP value may be reported to a central processing entity, wherein the central processing entity may be the serving cell 302-N, or a different network component such as a CU, a cloud-based processing node, or another network server. In cases where the ML model 306 is implemented in the central processing entity, the serving cell 302-7V may forward its own measurements, along with reports from the neighboring cells, to the central entity for processing. This architecture supports distributed measurements while allowing centralized decision-making, which can improve scalability and resource efficiency across the network.

[0061] In some embodiments, communication between the plurality of BSs and the plurality of UEs is performed using frequency division multiplexing, such that the available spectrum is divided into smaller frequency chunks called sub-bands. In some embodiments, for the Uth UE, assume that the SINK (which can be considered as a specific type of CQI) at sub-band j is denoted by y_fcj. Then y_{fc j}- can be computed as:

[0063] where F(-) denotes the function provided by the ML model 306, RSRP_serving-ceii_ikij denotes the Ll-RSRP value of the A th UE from the serving cell on subband j, and RSRP_{i k j}- denotes the Ll-RSRP value of the Uth UE from the z-th cell on subband j and N is the set of neighboring cells.

[0064] During the training of the ML model 306, the Uth UE may report the CQI value as labeled data for the model input measurements from the neighboring and serving cells, allowing the ML model 306 to learn the relationship between RSRP values and the corresponding CQI. In some embodiments, the ML model 306 is run for each sub-band to estimate the per-sub-band CQI, wherein the estimated per-sub-band CQI is used by the serving BS to assign the MCS for DL transmission according to the specifications in 3GPP TS 38.214.

[0065] In some embodiments, instead of relying on UE-reported CQI values for evaluating channel conditions, the ML model 306 disclosed herein can provide more accurate real-time CQI predictions based on real-time RSRP measurements from both the serving cell and neighboring cells, which can improve the accuracy of DL quality estimation. The ML model 306 may also enable the BS 302-TVto anticipate variations in channel quality by predicting CQI values instead of reacting to UE-reported data. This proactive approach may allow the BS 302-N to dynamically adapt the MCS based on anticipated DL conditions.

[0066] In some other embodiments, the BS 302-N may comprise an ML model that is used to make decisions related to mobility management. That is, the ML model in the BS 302-7V can be configured to make decisions about when and how a UE should transition between cells in a wireless network. For example, the ML model can be configured to make handover (HO) decisions for a UE instead of simply generating the CQI at its outputs. The ML model may be configured to produce an output which represents a handover probability assigned to each neighboring cell. This handover probability can reflect how likely it is that a UE should be handed over to a particular neighboring cell.

[0067] FIG. 4 illustrates another exemplary design framework 400 for computing handover (HO) probabilities based on RSRP reporting from neighboring cells, in accordance with some embodiments of the present disclosure. In some embodiments, a plurality of UEs may be in communication with a plurality of BSs, wherein the plurality of BSs may comprise JVBSS denoted by 402-1 to 402-7V as shown in FIG. 4. The HO probability for the Uth UE from the plurality of UEs may be computed using an ML model 406. In some embodiments, the BS 402-7V is in communication with the Uth UE from the plurality of UEs, wherein the BS 402-7V may be referred to as the “serving cell”, and the BSs 402-1 to 402-7V-7 may be referred to as “neighboring cells”.

[0068] In some embodiments, the serving cell (e.g. the BS 402-7V) may schedule the SRS transmission for the Uth UE and coordinate with the neighboring cells 402-1 to 402-7V-7 to perform measurements at the SRS resources allocated for the Uth UE. In some embodiments, each of the neighboring cells 402-1 to 402-7V-7 may be configured to perform measurements on the allocated SRS resources and report the associated LI -RSRP value back to the serving cell 402-7V, wherein the reported LI -RSRP values along with the LI -RSRP value of the serving cell are used as inputs to the ML model 406 to generate HO probability for each of the neighboring cells as well as the serving cell. In some embodiments, the Ll-RSRP values used by the ML model 406 as inputs are taken either for a single sub-band or at the center frequency of the carrier bandwidth, and the outputs of the ML model 406 are the HO probabilities assigned to each cell. For example, let the HO probability of the Uth UE to the

/-th cell " be denoted as p_kj, then p_kj can be computed as:

[0070] where F(-) denotes the function provided by the ML model 406, RSRP_serving-ceii_iki* denotes the Ll-RSRP value of the A th UE from the serving cell with * denoting that the sub-band index is irrelevant, and RSRP_{i k} denotes the Ll-RSRP value of the &-th UE from the /-th cell with * denoting that the sub-band index is irrelevant, ^hysteresis refers to a small value used to implement hysteresis in handover decisions, and JV is the set of neighboring cells to the serving cell. In some embodiments, the ML model 406 is implemented in the serving BS 402-7V, wherein the serving BS 402-7V may determine that the Uth UE should hand over to a neighboring cell with the highest HO probability as predicted by the ML model 406.

[0071] In some embodiments, to further enhance the robustness of the handover decision process, a Time-to-Trigger (TTT) parameter may be employed in addition to the hysteresis value ^hysteresis- The TTT parameter may refer to a time window during which the signal quality of the neighboring cell must consistently exceed that of the serving cell (by at least the hysteresis margin) before the handover is triggered. In some embodiments, p_{k j}- may be also computed as:

[0073] where TTT denotes the time-to-trigger parameter used to compute the handover probability p_{k j}- for the fc-th UE to the j-th neighboring cell. By incorporating TTT, the ML model 406 may avoid initiating handovers based on brief fluctuations in signal strength. In some embodiments, the ML model 406 can dynamically adjust the TTT value based on the

UE’s context (e.g., mobility, trajectory, and historical handover data) to ensure efficient handover decisions.

[0074] In some embodiments, hysteresis may be employed to prevent rapid and frequent handovers (known as the “ping-pong” effect) due to threshold discontinuities, as illustrated by the parameter A in the equation of p_{k j}-. The “ping-pong” effect may happen when a UE repeatedly switches between two or more neighboring cells due to minor fluctuations in signal strength around a threshold. By applying hysteresis, a margin or buffer can be added to the handover decision thresholds. For example, a UE can be configured not to initiate a handover to a neighboring cell until the signal from that neighboring cell exceeds the serving cell’s signal by a certain margin (e.g. the hysteresis value denoted by A). This ensures that small, temporary changes in signal quality do not trigger unnecessary handovers, reducing signaling overhead and improving network stability.

[0075] In some embodiments, the HO probability for different cells is estimated by the ML model 406, wherein the ML model 406 is implemented on the BS side (e.g. network side). In this case, the Uth UE may provide assistance information (e.g., UE trajectory, speed, location, etc.) to the source cell (e.g. the serving cell) to assist the source cell to determine which neighbor cell should provide SRS resources to the UE, thus enhancing the accuracy of handover decisions and improving overall mobility management.

[0076] In some embodiments, when the ML model 406 uses real-time Ll-RSRP values from both the serving cell and neighboring cells as inputs, the HO probabilities can be predicted in a more accurate manner. That is, by learning patterns from historical handover data and adjusting to real-time conditions, the ML model 406 can make more accurate and context-sensitive HO predictions, reducing the chances of premature or unnecessary handovers. In such as case, the ML model 406 may be trained using historical handover data, wherein the historical handover data are used as a ground truth for training the ML model 406. Moreover, the predicted HO probabilities can be used to proactively determine which cell would offer the best connection for the UE in the near future based on current conditions and predicted trends. This predictive capability allows the network to manage the communication more effectively, particularly when the UEs are moving rapidly between coverage areas, resulting in smoother transitions and less connection disruption.

[0077] FIG. 5 illustrates a signaling diagram for SRS resource determination, in accordance with some embodiments of the present disclosure. In some embodiments, the signaling diagram illustrated in FIG. 5 shows how a serving BS (e.g. the BS 402-7V shown in FIG. 4) coordinates with neighboring cells to determine which cells should allocate SRS resources for accurate RSRP measurements, wherein the RSRP measurements are used as inputs of an ML model for predicting HO probabilities. In some embodiments, the SRS resource allocation illustrated in FIG. 5 may provide the ML model (e.g. the ML model 406 in FIG. 4) with updated real-time input data for accurately predicting the HO probabilities. In some embodiments, a UE 504 may be in communication with a BS 502-1, wherein the BS 502-1 may have a plurality of neighboring BSs 502-2 to 502-3. In some embodiments, the BS 502-1 that is in communication with the UE 504 may be referred to as “serving cell” or “source cell”, and the BSs 502-2 and 502-3 may be referred to as “neighboring cells”. In some embodiments, the BS 502-1 may transmit a signal comprising a configuration message, wherein the configuration message specifies what assistance information (e.g. UE trajectory, speed, and location data) the UE 504 should collect and report to the BS 502-1. In some embodiments, the configuration message may also specify how often and/or under what conditions the assistance information should be reported (e.g. periodically or aperiodically). In some embodiments, the BS 502-1 may use the assistance information to determine which neighboring BS (or which neighboring cell) should configure SRS resources to the UE 504.

[0078] Upon receiving the configuration message comprising the assistance information configuration, the UE 504 may transmit assistance information such as UE trajectory, speed, movement direction and location data to the BS 502-1. In some embodiments, the assistance information is transmitted periodically, wherein the UE 504 transmits the assistance information at regular intervals, allowing the network to continuously monitor changes in the UE’s status. This is useful for fast-moving UEs where conditions can change quickly. In some other embodiments, the assistance information is transmitted aperiodically, wherein the UE 504 transmits the assistance information only when certain events or thresholds are met, such as when the speed of the UE 504 exceeds a certain limit or when the UE 504 changes direction significantly. This can reduce signaling overhead by sending updates only when necessary.

[0079] In some embodiments, upon receiving the assistance information transmitted by the UE 504, the BS 502-1 may determine which neighboring cell(s) should allocate SRS resources to the UE 504. For example, in one embodiment, the BS 502-1 may determine that the BSs 502-2 and 502-3 should allocate SRS resources to the UE 504. In another embodiment, the BS 502-1 may determine that only the BS 502-2 should allocate SRS resources to the UE 504. In yet another embodiment, the BS 502-1 may determine that only the BS 502-3 should allocate SRS resources to the UE 504.

[0080] In some embodiments, when the BS 502-1 determines that both the BS 502-2 and the BS 502-3 should allocate SRS resources to the UE 504, and the BS 502-1 may transmit an SRS resource request to each of the BSs 502-2 and 502-3. Upon receiving an SRS resource request response for the UE 504 from each of the BSs 502-2 and 502-3, the BS 502-1 may configure the UE 504 with the SRS resources from 502-2 and 502-3 as well as the SRS resources from the serving cell BS 502-1, as illustrated by the signal “SRS configuration” in FIG. 5. In some embodiments, as discussed above with reference to FIG. 4, the UE 504 may be configured to transmit SRSs to the BSs 502-1, 502-2 and 502-3. Based on the SRS measurement results from the BSs 502-1 to 502-3, the serving cell BS 502-1 may determine whether a handover is needed for the UE 504 using an ML model implemented in the BS 502-1. If a handover is needed, then the serving cell BS 502-1 may determine which neighboring cell the UE 504 should hand over to using the ML model. The information on which neighboring cell the UE 504 should hand over to can be transmitted in an “HO command” signal as illustrated in FIG. 5.

[0081] In some embodiments, as discussed above with reference to FIG. 4, the UE 504 may be configured to transmit SRSs to the serving BS 502-1 and one or more neighboring BSs (e.g., 502-2 and 502-3). The decision of which neighboring BSs should receive SRS transmissions from the UE 504 may be based on coordination among network entities. In one embodiment, a centralized network entity (e.g. a CU) may determine which BSs should allocate SRS resources for the UE 504 based on assistance information, such as the UE’s trajectory, speed, and location.

[0082] In some other embodiments, the SRS measurements at the BSs (502-1, 502-2, and 502-3) may provide UL signal quality indicators, such as Ll-RSRP, which may be used by the serving BS 502-1 in conjunction with the ML model to predict HO probabilities. This approach differs from legacy HO mechanisms, where the UE’s DL RSRP measurements are used to determine the target cell for handover. In some embodiments, the ML model may consider both DL RSRP measurements at the UE 504 and UL SRS measurements at the BSs to make more robust HO decisions by accounting for potential differences in UL and DL channel conditions (e.g., due to asymmetric interference patterns). In some embodiments, the serving BS 502-1 may use the DL RSRP measurements of UE 504 for cross-verification, particularly in environments with significant differences between DL and UL channel conditions. By incorporating both SRS measurements from the BSs and DL measurements from the UE 504, the ML-based framework can make context-aware HO decisions, improving handover success rates and reducing unnecessary handovers.

[0083] In some embodiments, the UE 504 may be configured to determine which neighboring cell(s) to transmit the SRSs based on the trajectory, position, and speed of the UE 504. That is, instead of relying solely on instructions from the serving cell BS 502-1, the UE 504 may assess its trajectory, position, and speed to determine which neighboring cell(s) the UE 504 needs to transmit the SRSs. By targeting specific neighboring cell(s), the UE 504 may avoid unnecessary transmissions to all possible neighboring cells, which helps conserve power in the UE 504. In some embodiments, although the SRS resources for multiple neighboring cells may be pre-configured by a centralized network entity (e.g., a CU), the serving BS 502-1 may instruct the UE 504 to transmit SRSs selectively to only the neighboring cells most relevant for the handover decision. For example, based on assistance information (e.g., UE trajectory, speed, and location), the ML model at the serving BS 502-1 (or a centralized entity) may predict which neighboring cells are likely candidates for handover. In this case, the CU may reserve and configure SRS resources for all potential neighboring cells to ensure they are available when needed. In some embodiments, the serving BS 502-1 may only instruct the UE 504 to use the SRS resources of a subset of neighboring cells (e.g., BS 502-2 but not BS 502-3), based on real-time mobility predictions. This avoids unnecessary SRS transmissions to all neighboring cells and conserves UE power. In some other embodiments, the list of target neighboring cells can be updated dynamically as the UE 504 moves, ensuring that only relevant SRS transmissions are made while unused SRS configurations remain idle unless needed.

[0084] In some embodiments, the serving cell BS 502-1 is configured to schedule UE’s SRS transmissions, which requires coordination with neighboring cells. This coordination creates signaling overhead, as the serving cell BS 502-1 needs to communicate with neighboring cells about SRS scheduling details. When determining CQI for DL transmissions, the serving cell BS 502-1 may schedule periodic SRS transmissions for the UE 504 when the BS 502-1 is prepared to schedule the UE 504 for DL transmissions, therefore aligning SRS transmissions with upcoming DL transmissions. This helps the serving cell BS 502-1 assess channel conditions accurately in preparation for DL data scheduling.

[0085] In a handover context, the serving cell BS 502-1 may use RRM measurements as triggers for SRS transmissions. If the UE 504 reports that a neighboring cell’s signal strength surpasses a certain threshold, the serving cell BS 502-1 may begin scheduling the UE 504 to transmit SRSs periodically. The periodic SRS transmission can continue until a handover decision is made, ensuring that the serving cell BS 502-1 has up-to-date channel information for the potential handover target. To reduce the signaling overhead, the serving cell BS 502-1 may schedule multiple UEs to transmit SRSs in the same timeslot on orthogonal set of subbands. By assigning each UE an orthogonal set of sub-bands within the same timeslot, as illustrated below with reference to FIG. 6, the serving cell BS 502-1 can reduce the overall signaling overhead while still gathering the necessary channel information for all involved UEs.

[0086] In some embodiments, the serving BS 502-1 may use RRM measurement reports from the UE 504 as triggers for SRS transmissions, wherein the RRM measurement reports comprise DL RSRP, RSRQ, or SINR values measured by the UE 504 for both the serving and neighboring cells. These reports can help the network determine whether to schedule UL

SRS transmissions to obtain more precise measurements for HO decision-making.

[0087] In some other embodiments, the SRS resources for the serving cell and neighboring cells may be pre-configured in advance by a centralized entity (e.g., CU). The UE 504 may then determines to transmit SRSs to the serving and neighboring cells based on its own DL RSRP measurements without explicitly reporting those measurements back to the serving BS 502-1. The BSs (serving and neighboring) may then measure the received SRSs and report UL RSRP values to the serving BS 502-1. These values may serve as inputs to the ML model for predicting HO probabilities. This approach assumes that the SRS resources are available to the UE 504 and can be used for uplink transmission without the need for new configuration steps. In some embodiments, the ML model may combine UL RSRP measurements with the UE’s DL RSRP reports (if available) to improve cross-validation and prediction accuracy.

[0088] In yet some other embodiments, the SRS resources for the neighboring BSs may not be pre-configured initially. In this case, the serving BS 502-1 may rely on the UE’s RRM measurement reports (comprising DL RSRP/SINR values) to identify whether the signal from a neighboring cell is improving or if the signal from the serving cell is deteriorating. If the UE 504’ s measurement report indicates a significant drop in the serving cell’s RSRP or an improvement in a neighboring cell’s RSRP, the serving BS 502-1 may configure SRS resources for relevant neighboring cells and schedule the UE to transmit SRSs on those resources. The SRS transmissions can provide fine-grained UL RSRP measurements that are used as real-time inputs to the ML model to refine the HO probability prediction. This approach addresses potential limitations of using DL RSRP reports alone, ensuring that the ML model receives high-quality, real-time measurements from multiple neighboring BSs. [0089] FIG. 6 illustrates an exemplary time-frequency resource allocation diagram 600 for SRS transmission, in accordance with some embodiments of the present disclosure. The time-frequency resource allocation diagram 600 in FIG. 6 is shown along a time axis 602 and a frequency axis 604. In some embodiments, SRSs may be transmitted from each of a plurality of UEs 606-1 to 606-m to a plurality of BSs comprising a serving BS. In some embodiments, each of the plurality of UEs 606-1 to 606-m transmits its respective SRSs to the serving BS in the same time slot at different frequencies. In FIG. 6, each block with a different fill pattern corresponds to a different UE from the plurality of UEs 606-1 to 606-m, and blocks with the same fill pattern corresponds to the same UE from the plurality of UEs 606-1 to 606-m. In some embodiments, for each time duration of transmission, respective SRSs may be transmitted from each of the plurality of UEs 606-1 to 606-m to each of the plurality of BSs, wherein each of the plurality of UEs 606-1 to 606-m transmits respective SRSs with a different frequency. For example, for the time duration 612 shown in FIG. 6, each of the plurality of UE1 to UE5 transmits respective SRSs with a different frequency, wherein UE1 transmits its respective SRSs with the highest frequency as shown by block 606-1, and UE2 transmits its respective SRSs with the lowest frequency as shown by block 606-i. In some embodiments, a UE transmits SRSs at different time durations with different frequencies. In some other embodiments, a UE transmits different SRSs at different time durations with the same frequency.

[0090] In some embodiments, the time interval T between two consecutive time durations in the SRS transmission may be expressed as T > 2z_max, wherein T_max represents the maximum propagation delay between the transmitting UE and the receiving BS. An example of the time interval T is shown by the interval 608 in FIG. 6. Similarly, the frequency separation interval or frequency difference A between two consecutive frequency sub-bands in the SRS transmission may be expressed as A > 2A/_D, wherein A/_D represents the maximum Doppler shift for different moving UEs in the plurality of UEs 606-1 to 606-m. An example of the frequency separation interval A is shown by the interval 610 in FIG. 6. In some embodiments, T_mnx is determined based on the farthest distanced UE from the receiving BS among the plurality of UEs 606-1 to 606-m, and A/_D is determined based on the maximum velocity of the UE among the plurality of UEs 606-1 to 606-m. In some embodiments, the SRS transmissions for each of the plurality of UEs 606-1 to 606-m are distributed over the entire frequency band of the SRS.

[0091] In some embodiments, the time-frequency resource allocation strategy illustrated in FIG. 6 allows multiple UEs to transmit SRSs in the same time slot but on different frequency sub-bands, therefore optimizing the use of available spectrum. In addition, the frequency -based separation of SRS resources enables the network to scale up efficiently, supporting more UEs within the same time slot without needing additional time-frequency resources.

[0092] Referring back to FIG. 4, in some embodiments, the design framework 400 disclosed herein does not necessitate any modifications in the Life Cycle Management (LCM) operations as defined in TR 38.843, which provides a description of the LCM operations encompassing data acquisition, training, inference, and performance monitoring. In addition, the design framework 400 depicted in FIG. 4 can also be adaptable to any potential modifications in signaling aspects introduced by 3GPP in the future. This flexibility ensures that the design framework 400 disclosed herein can seamlessly accommodate any forthcoming changes in signaling procedures mandated by 3 GPP and allows for the continued compatibility of the design framework 400 disclosed herein in different environments. [0093] Referring back to FIG. 3, in some embodiments, the ML model 306 is configured to predict CQI values (e.g. SINR values) for a plurality of UEs at the outputs of the ML model 306, wherein the ML model 306 is trained to consider inter-cell interference as a proxy for the CQI reports. During the training stage, the serving BS 302-7V may gather the Ll-RSRP information from all the neighboring cells and measure the received signal strength from each of the plurality of UEs, while at the same time, the serving BS 302-7V may request the plurality of UEs to report their CQI values. Then the Ll-RSRP values may be provided to the ML model 306 as inputs, and the corresponding CQI information may be provided to the ML model 306 as the output label data. The ML model 306 may be then configured to generate the CQI value per sub-band to be used for the DL transmissions in the inference stage.

[0094] Referring again to FIG. 4, in some embodiments, the ML model 406 is configured to predict the HO probability for each of the plurality of BSs 402-1 to 402-7V, wherein a parameter called “LoggedMeasurementConfiguration” is used as part of the mobility mechanism to collect labels for the ML model 406. The “LoggedMeasurementConfiguration” may be referred to as a configuration parameter used in wireless communication systems such as LTE and 5G to instruct the UE to record specific measurement data (e.g. RSRP and RSRQ) for later reporting to the network. This configuration may be used to gather information about the radio environment when certain events such as a radio link failure (RLF) or a handover failure (HOF) occur. In some embodiments, the “LoggedMeasurementConfiguration” parameter enables the collection of data that can be used for training the ML model 406. For example, the logged information may be used to refine the ML model 406 that predicts handover probabilities or to optimize mobility decisions in challenging radio environments. [0095] In some embodiments, upon encountering a link failure, the Uth UE may generate a radio link failure report, which may indicate either an RLF or an HOF. The radio link failure report may include pertinent information that aids the serving BS 402-7V in computing handover probabilities for neighboring cells, wherein the pertinent information may include UE’s location, velocity, direction, and associated uncertainty as specified in TS 37.320 V18.1.0 of the Third Generation Partnership Project (3GPP). In some embodiments, the process of collecting input data for the ML model 406 (e.g. Ll-RSRP) is similar to that of the ML model 306 described above with reference to FIG. 3.

[0096] In some embodiments, after performing the data collection and training, the ML model 406 may be configured to perform inference and performance monitoring. In some embodiments, during the inference and performance monitoring, the serving BS 402-7V may collect UL data from the Uth UE and use the trained ML model 406 to predict HO probabilities for different cells. This process can be applied to either spatial beam prediction (for BM Case-1) or temporal beam prediction (for BM Case-2). Periodically, the serving BS 402-7V may monitor the performance of the trained ML model 406 to determine if the serving BS 402-7V should continue using the current trained ML model 406, switch to an alternative model, or revert to a legacy method.

[0097] In some embodiments, during the performance monitoring of the trained ML model 406, the serving BS 402-7V may first collect input data samples and labels from the Uth UE using a similar method used in the data collection and training of the ML model 406 as described above. Then, the serving BS 402-7V may be configured to apply the collected input data samples to the trained ML model 406. Next, the serving BS 402-7V may compare the collected label to the predicted label. If there is a low loss function value, the serving BS 402- N may decide to maintain the current trained ML model 406. If there are large discrepancies, the serving BS 402-7V may switch models or revert to a more traditional, non-ML method (e.g. legacy fallback).

[0098] In some embodiments, the ML models 306 and 406 use raw, unquantized UL channel measurements (such as Ll-RSRP) as inputs, which contrasts with approaches in 3GPP where quantized or pre-processed data is used. By using unquantized data, the ML models 306 and 406 may benefit from more accurate and precise channel information, which can enhance the prediction accuracy of the ML models. This approach reduces performance degradation that typically arises from quantization errors, enabling the ML model to make more reliable beam predictions. The method disclosed herein ensures that the design framework 400 continues to operate efficiently by dynamically adjusting or retraining the ML model 406 as necessary based on real-world performance data. In some other embodiments, the inference and performance monitoring can also be applied to the ML model 306 described in FIG. 3.

[0099] In some embodiments, the Ll-RSRP measurements need to be transmitted between network entities (e.g., from neighboring cells to the serving BS or from BSs to a CU). In such cases, quantization may be required to minimize signaling overhead. In one embodiment, the ML model (e.g., the ML model 406) is implemented at the serving BS (e.g., BS 402-N). In this case, Ll-RSRP measurements from neighboring cells are quantized at the neighboring cells before being transmitted to the serving BS. The serving BS may then use these quantized measurements as inputs to the ML model 406 for predicting HO probabilities or performing beam predictions. In another embodiment, the ML model (e.g., ML model 406 or 306) is implemented at a centralized network entity such as a CU. In this scenario, all Ll- RSRP measurements (including those from the serving BS and neighboring cells) are quantized before being transmitted to the CU. Quantization ensures efficient data transmission while maintaining compatibility with the centralized architecture. In yet another embodiment, the ML model is implemented locally within a BS (e.g., the serving BS 402-N) or a UE. In such a case, unquantized Ll-RSRP measurements may be used as ML model inputs. This eliminates performance degradation from quantization errors and allows the ML model to operate with higher precision. For example, the serving BS may directly utilize its own local unquantized Ll-RSRP measurements to complement quantized inputs from neighboring cells. In some embodiments, the serving BS or the CU may optimize CQI predictions based on the combination of quantized and unquantized measurements.

[00100] FIG. 7 illustrates a deep neural network (DNN) model 700 used to implement the ML model 306 or 406, in accordance with some embodiments of the present disclosure. Although a DNN example is illustrated in FIG. 7, the ML model in the present disclosure is not limited to DNN implementation, and can take any other forms of ML model, such as multilayer perceptron, feedforward neural networks, convolutional neural networks, recurrent neural networks, autoencoder, generative adversarial networks, long short-term memory and transformers. In some embodiments, the ML model 308, described in FIG.3, layer 410-k output can be arranged in the vector o(fc) = [o₁₍ ... , o_n with layer k consisting of n_k neurons and the input layer corresponding to k = 0. The relationship between the outputs of layers k — 1 and k can be represented as o(fc) = s(o(fc — l)l _k), where s(. ) is a non-linear activation function applied elementwise to the input argument and IV _k is a weight matrix that connects layers 410-(k) and 410-(k-l).

[00101] In some embodiments, the DNN model 700 is trained using a plurality of training samples, wherein each training sample comprises an input vector and a corresponding output vector. In one embodiment, to find the optimal values of the weight matrices IV _x to W_k+1 during the training of the DNN model 700, a back propagation algorithm is used by taking an error rate of a forward propagation and feeding this loss backward through the layers of the

DNN model 700 to fine-tune the weights. In another embodiment, to find the optimal values of the weight matrices IV₁ to IV_fc+1 during the training of the ANN model 700, a weight perturbation technique can be used. The weight perturbation technique may be applied in an iterative manner for a plurality of iterations, wherein in each of the plurality of iterations, a weight variation of random sign is added to each of the elements in the weight matrices I _x to IV_fc+1 and a corresponding training error is observed. If the training error is increased in a given iteration, then the elements in the weight matrices IV _x to IV_fc+1 will be changed to the opposite directions of the weight variations; if the training error is decreased in a given iteration, then the elements in the weight matrices IV _x to IV_fc+1 will be changed to the same directions of the weight variations. This iterative training can be stopped if at least one of the following conditions is met: the training error becomes smaller than a predetermined error threshold value, a maximum number of iterations is reached, and the training error does not decrease for a predetermined number of iterations. In some embodiments, a dynamic weight perturbation technique can be applied to train the DNN model 700 by decreasing the amount of weight variations in each iteration, such that the DNN model 700 is fine-tuned towards the end of the training process. In one embodiment, the weight variation in the t-th iteration v_t can be calculated as: v_t where v₀ is an initial weight variation amount, and is a user-defined parameter which controls the decrease rate of v_t.

[00102] FIG. 8 illustrates an example method 800 for CQI determination and handover decision, in accordance with some embodiments. The operations of method 800 presented below are intended to be illustrative. In some embodiments, method 800 may be accomplished with one or more additional operations not described and/or without one or more of the operations discussed. Additionally, the order in which the operations of method

800 are illustrated in FIG. 8 and described below is not intended to be limiting.

[00103] At step 802, a serving BS transmits a first signal comprising a configuration message to a UE. In some embodiments, the configuration message specifies what assistance information (e.g. UE trajectory, speed, and location data) the UE should collect and report back to the serving BS. In some embodiments, the configuration message may also specify how often and/or under what conditions the assistance information should be reported (e.g. periodically or aperiodically). In some embodiments, the serving BS may use the assistance information to determine which neighbor BSs (or which neighbor cells) should configure SRS resources to the UE.

[00104] At step 804, upon receiving the first signal, the UE may be configured to transmit a second signal comprising the assistance information such as UE trajectory, speed, movement direction and location data to the serving BS. In some embodiments, the assistance information is transmitted periodically, wherein the UE transmits the assistance information at regular intervals, allowing the network to continuously monitor changes in the UE’s status. In some other embodiments, the assistance information is transmitted aperiodically, wherein the UE transmits the assistance information only when certain events or thresholds are met, such as when the speed of the UE exceeds a certain limit or when the UE changes direction significantly.

[00105] At step 806, upon receiving the second signal, the serving BS may determine which neighbor cell(s) should allocate SRS resources to the UE based on the assistance information. For example, based on the assistance information, the serving BS may predict that the UE will soon move closer to a neighboring BS and eventually away from UE’s current serving cell's coverage provided by the serving BS. To ensure optimal connectivity and minimize potential signal degradation, the serving BS may determine that the neighboring BS should prepare to allocate SRS resources to the UE.

[00106] At step 808, the serving BS may transmit an SRS resource request to each of the neighbor cells that needs to allocate the SRS resources to the UE as determined at step 806, and receive an SRS resource request response from each of the neighbor cells.

[00107] At step 810, upon receiving the SRS resource request responses from the neighbor cells, the serving BS may configure the UE with SRS resources from the neighbor cells and the serving cell. That is, the serving BS may allocate the SRS resources from the neighbor cells and the serving cell to the UE for further transmissions.

[00108] At step 812, the UE may transmit SRSs to the serving BS as well as to the neighbor BSs (e.g. neighbor cells) from which the SRS resources are allocated to the UE.

[00109] At step 814, based on the SRS measurement results (e.g. Ll-RSRP values) from the neighboring cells and the serving cell (e.g. the serving BS), the serving BS may generate one or more output signals using an ML model implemented in the serving BS or a remote server. In alternative embodiments, the ML model may be implemented in the UE as described above with reference to Figures 2B and 2D. For example, the UE may perform SRS measurements (e.g. Ll-RSRP) from the serving cell and neighboring cells, wherein the SRS measurements are used as inputs in the locally implemented ML model within the UE. The ML model within the UE can then process the SRS measurements to predict either CQI values for DL transmissions or HO probabilities for neighboring cells. With the ML model implemented in the UE, the UE only needs to send predicted CQI or HO probabilities to the serving BS, instead of sending detailed raw measurements. This approach may reduce the amount of data transmitted back to the BS, potentially lowering signaling overhead. [00110] In some embodiments, the one or more output signals comprise predicted CQI for the DL transmissions, and the ML model used by the serving BS is the ML model 306 described in FIG. 3. In such a case, in the training stage of the ML model, the UE may report the computed CQI values for the DL, which represent the actual DL quality that the UE experiences. These reported CQI values may serve as ground truth data used to train the ML model. Additionally, the UE may continue to transmit SRSs to the neighboring cells during training of the ML model. The serving BS then may compare the predicted CQI values generated by the ML model 306 with the ground truth CQI values reported by the UE. In this way, the serving BS can continuously improve the ML model’s prediction accuracy.

[00111] In some other embodiments, the one or more output signals comprise HO probabilities for different cells predicted by the ML model implemented in the serving BS, wherein the ML model is the ML model 406 described in FIG. 4. In such a case, the UE’s RRM measurement reports may be used as the triggers for SRS transmissions. That is, the UE may perform regular RRM measurements on its serving cell and neighboring cells. These measurements may include RSRP, RSRQ, and SINR. The results of these measurements may be reported back to the serving BS, and based on these RRM measurement reports, the serving BS may decide when to schedule SRS transmissions. For instance, if the UE’s RSRP on the serving cell drops below a certain threshold or if the signal quality of some neighboring BSs improves significantly, the serving BS may schedule SRS transmissions for the neighboring BSs to gather more detailed uplink channel information. In some embodiments, based on the one or more output signals (e.g. predicted HO probabilities by the ML model), the serving BS determines if a handover is needed or not using the ML model. If a handover is needed, then the serving BS may determine which neighboring cell the UE should hand over to using the ML model. In some embodiments, multiple UEs can be scheduled for SRS transmissions in one timeslot, as described above with reference to FIG. 6. [00112] At step 816, the serving BS may transmit a third signal to the UE. In some embodiments, the ML model implemented in the serving BS is the ML model 306. In this case, the third signal comprises predicted CQI for DL transmissions. In some other embodiments, the ML model implemented in the serving BS is the ML model 406. In this case, the third signal comprises an indication to indicate which neighboring cell the UE should hand over to.

[00113] While various embodiments of the present disclosure have been described above, it should be understood that they have been presented by way of example only, and not by way of limitation. Likewise, the various diagrams may depict an example architectural or configuration, which are provided to enable persons of ordinary skill in the art to understand exemplary features and functions of the present disclosure. Such persons would understand, however, that the present disclosure is not restricted to the illustrated example architectures or configurations, but can be implemented using a variety of alternative architectures and configurations. Additionally, as would be understood by persons of ordinary skill in the art, one or more features of one embodiment can be combined with one or more features of another embodiment described herein. Thus, the breadth and scope of the present disclosure should not be limited by any of the above-described exemplary embodiments.

[00114] It is also understood that any reference to an element herein using a designation such as "first," "second," and so forth does not generally limit the quantity or order of those elements. Rather, these designations can be used herein as a convenient means of distinguishing between two or more elements or instances of an element. Thus, a reference to first and second elements does not mean that only two elements can be employed, or that the first element must precede the second element in some manner. [00115] Additionally, a person having ordinary skill in the art would understand that information and signals can be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits and symbols, for example, which may be referenced in the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[00116] A person of ordinary skill in the art would further appreciate that any of the various illustrative logical blocks, modules, processors, means, circuits, methods and functions described in connection with the aspects disclosed herein can be implemented by electronic hardware (e.g., a digital implementation, an analog implementation, or a combination of the two), firmware, various forms of program or design code incorporating instructions (which can be referred to herein, for convenience, as "software" or a "software module), or any combination of these techniques.

[00117] To clearly illustrate this interchangeability of hardware, firmware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware, firmware or software, or a combination of these techniques, depends upon the particular application and design constraints imposed on the overall system. Skilled artisans can implement the described functionality in various ways for each particular application, but such implementation decisions do not cause a departure from the scope of the present disclosure. In accordance with various embodiments, a processor, device, component, circuit, structure, machine, module, etc. can be configured to perform one or more of the functions described herein. The term “configured to” or “configured for” as used herein with respect to a specified operation or function refers to a processor, device, component, circuit, structure, machine, module, etc. that is physically constructed, programmed and/or arranged to perform the specified operation or function.

[00118] Furthermore, a person of ordinary skill in the art would understand that various illustrative logical blocks, modules, devices, components and functions described herein can be implemented within or performed by one or more circuits or circuitry. As used herein, the term “circuitry” refers to and includes any one or more of the following: discrete circuit components or devices coupled to each other to form circuit, logic circuitry, integrated circuits, application specific integrated circuits, state machines, general purpose processors, special purpose processors, digital signal processors (DSP), microprocessors, field programmable gate arrays (FPGA) or other programmable logic devices, or any combination thereof. Circuitry can further include antennas, reflectors, transmitters, receivers and/or transceivers to communicate with various components, devices or nodes within a communication network. As used herein, the term “processor” refers to a combination of structures including processing circuitry, a memory coupled to the processing circuitry, and executable code stored in the memory that when executed by the processing circuitry perform the functions or operations instructed by the executable code.

[00119] If implemented in software, the functions can be stored as one or more instructions or code on a computer-readable medium. Thus, the steps of a method or algorithm disclosed herein can be implemented as software stored on a computer-readable medium. Computer- readable media includes both computer storage media and communication media including any medium that can be enabled to transfer a computer program or code from one place to another. A storage media can be any available media that can be accessed by a computer. By way of example, and not limitation, such non-transitory computer-readable media can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store desired program code in the form of instructions or data structures and that can be accessed by a computer.

[00120] In this document, the term "module" as used herein, refers to software, firmware, hardware, and any combination of these elements for performing the associated functions described herein. Additionally, for purpose of discussion, the various modules are described as discrete modules; however, as would be apparent to one of ordinary skill in the art, two or more modules may be combined to form a single module that performs the associated functions according embodiments of the present disclosure.

[00121] Additionally, memory or other storage, as well as communication components, may be employed in embodiments of the present disclosure. It will be appreciated that, for clarity purposes, the above description has described embodiments of the present disclosure with reference to different functional units and processors. However, it will be apparent that any suitable distribution of functionality between different functional units, processing logic elements or domains may be used without detracting from the present disclosure. For example, functionality illustrated to be performed by separate processing logic elements, or controllers, may be performed by the same processing logic element, or controller. Hence, references to specific functional units are only references to a suitable means for providing the described functionality, rather than indicative of a strict logical or physical structure or organization.

[00122] Various modifications to the implementations described in this disclosure will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other implementations without departing from the scope of this disclosure. Thus, the disclosure is not intended to be limited to the implementations shown herein, but is to be accorded the widest scope consistent with the novel features and principles disclosed herein, as recited in the claims below.

Claims

CLAIMS What is claimed is:

1. A method comprising: transmitting, by a wireless communication device, a first signal to a first plurality of wireless communication nodes, wherein the first plurality of wireless communication nodes comprises a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); receiving, by the wireless communication device, a second signal from the first wireless communication node, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

2. The method of claim 1, wherein the second signal is determined from a plurality of Layer 1 Reference Signal Received Power (Ll-RSRP) values using a machine learning (ML) model, wherein each of the plurality of Ll-RSRP values is measured from the first signal at a respective one of the first plurality of wireless communication nodes.

3. The method of claim 2, wherein the ML model is trained using historical handover data, wherein the historical handover data are used as a ground truth for training the ML model.

4. The method of claim 1, wherein transmissions of the first signal are triggered by radio resource management (RRM) measurement reports transmitted from the wireless communication device to the first wireless communication node, wherein the RRM measurement reports comprise RRM measurements, wherein the RRM measurements comprise at least one of: a reference signal received power (RSRP) value; a reference signal received quality (RSRQ) value; and a signal-to-interference-plus-noise ratio (SINR) value.

5. The method of claim 1, wherein the first signal is transmitted in a time slot during which each of a plurality of wireless communication devices transmits a respective SRS to the first plurality of wireless communication nodes.

6. The method of claim 1, wherein the first signal is transmitted based on an SRS resource allocation, wherein the SRS resource allocation is determined based on assistance information of the wireless communication device, wherein the assistance information comprises at least one of: a trajectory of the wireless communication device; a speed of the wireless communication device; a movement direction of the wireless communication device; and a location of the wireless communication device.

7. The method of claim 2, wherein the indication is determined based on each of a plurality of handover probabilities associated with a respective one of the first plurality of wireless communication nodes, wherein the plurality of handover probabilities is determined by the ML model.

8. A wireless communication device comprising: a transceiver configured to: transmit a first signal to a first plurality of wireless communication nodes, wherein the first plurality of wireless communication nodes comprises a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); receive a second signal from the first wireless communication node, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

9. A non-transitory computer readable medium storing computer-executable instructions which when executed perform a method comprising: transmitting, by a wireless communication device, a first signal to a first plurality of wireless communication nodes, wherein the first plurality of wireless communication nodes comprises a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); receiving, by the wireless communication device, a second signal from the first wireless communication node, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

10. Circuitry configured to perform a method, the method comprising: transmitting, by a wireless communication device, a first signal to a first plurality of wireless communication nodes, wherein the first plurality of wireless communication nodes comprises a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); receiving, by the wireless communication device, a second signal from the first wireless communication node, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

11. A method comprising: receiving, by each of a first plurality of wireless communication nodes, a first signal from a wireless communication device, wherein the first plurality of wireless communication nodes comprises a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); transmitting, by the first wireless communication node, a second signal to the wireless communication device, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

12. The method of claim 11, wherein the second signal is determined from a plurality of Layer 1 Reference Signal Received Power (Ll-RSRP) values using a machine learning (ML) model, wherein each of the plurality of Ll-RSRP values is measured from the first signal at a respective one of the first plurality of wireless communication nodes.

13. The method of claim 12, wherein the ML model is trained using historical handover data, wherein the historical handover data are used as a ground truth for training the ML model.

14. The method of claim 11, wherein transmissions of the first signal are triggered by radio resource management (RRM) measurement reports transmitted from the wireless communication device to the first wireless communication node, wherein the RRM measurement reports comprise RRM measurements, wherein the RRM measurements comprise at least one of: a reference signal received power (RSRP) value; a reference signal received quality (RSRQ) value; and a signal-to-interference-plus-noise ratio (SINR) value.

15. The method of claim 11, wherein the first signal is received in a time slot during which the first plurality of wireless communication nodes receives a respective SRS from each of a plurality of wireless communication devices.

16. The method of claim 11, wherein the first signal is received based on an SRS resource allocation, wherein the SRS resource allocation is determined based on assistance information of the wireless communication device, wherein the assistance information comprises at least one of: a trajectory of the wireless communication device; a speed of the wireless communication device; a movement direction of the wireless communication device; and a location of the wireless communication device.

17. A first wireless communication node comprising: a transceiver configured to: receive a first signal from a wireless communication device, wherein the first signal is received at a first plurality of wireless communication nodes comprising the first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); transmit a second signal to the wireless communication device, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

18. A non-transitory computer readable medium storing computer-executable instructions which when executed perform a method comprising: receiving, by each of a first plurality of wireless communication nodes, a first signal from a wireless communication device, wherein the first plurality of wireless communication nodes comprises a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); transmitting, by the first wireless communication node, a second signal to the wireless communication device, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

19. Circuitry configured to perform a method, the method comprising: receiving, by each of a first plurality of wireless communication nodes, a first signal from a wireless communication device, wherein the first plurality of wireless communication nodes comprises a first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); transmitting, by the first wireless communication node, a second signal to the wireless communication device, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.

20. A communication system comprising a wireless communication device and a first wireless communication node, wherein: the wireless communication device comprises a first transceiver configured to transmit a first signal to a first plurality of wireless communication nodes, wherein the first plurality of wireless communication nodes comprises the first wireless communication node and a second plurality of wireless communication nodes, wherein the first signal comprises at least one sounding reference signal (SRS); the first wireless communication node comprises a second transceiver configured to transmit a second signal to the wireless communication device, wherein the second signal is determined based on the first signal, and the second signal comprises: an indication to indicate a neighboring cell to which the wireless communication device should hand over, wherein the neighboring cell is covered by one of the first plurality of wireless communication nodes.