Overhead allocation for machine learning based CSI feedback

Abstract

Methods and apparatus are provided for machine learning (ML)-based channel state information (CSI) feedback from a user equipment (UE) to a wireless network. For a neural network (NN) model type and a method for deriving CSI for multi-rank CSI feedback, the UE reports one or more feedback sizes to the wireless network in a UE capability message. For the NN model type and the method, the UE processes a configuration of a plurality of NN models associated with the one or more feedback sizes from the wireless network. The UE processes an overhead allocation for one or more feedback sizes, and generates the multi-rank CSI feedback using a different NN model of the plurality of NN models for each spatial layer or group of spatial layers or rank based on the overhead allocation.

Description

Technical Field

[0001] The present application generally relates to wireless communication systems, including channel state information (CSI) feedback. Background Art

[0002] Wireless mobile communication technology uses various standards and protocols to send data between base stations and wireless communication devices. Wireless communication system standards and protocols may include, for example, the 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE) (e.g., 4G), 3GPP New Radio (NR) (e.g., 5G), and the IEEE 802.11 standard for wireless local area networks (WLANs) (commonly referred to within industry organizations as WLANs). ).

[0003] As envisioned by 3GPP, different wireless communication system standards and protocols may use various radio access networks (RANs) to communicate between base stations of the RAN (which may also sometimes be referred to as RAN nodes, network nodes, or simply nodes) and wireless communication devices referred to as user equipment (UE). 3GPP RANs may include, for example, Global System for Mobile Communications (GSM), Enhanced Data Rates for GSM Evolution (EDGE) RAN (GERAN), Universal Terrestrial Radio Access Network (UTRAN), Evolved Universal Terrestrial Radio Access Network (E-UTRAN), and / or Next Generation Radio Access Network (NG-RAN).

[0004] Each RAN may use one or more radio access technologies (RATs) to perform communications between base stations and UEs. For example, GERAN implements GSM and / or EDGE RAT, UTRAN implements Universal Mobile Telecommunications System (UMTS) RAT or other 3GPP RAT, E-UTRAN implements LTE RAT (sometimes referred to as LTE), and NG-RAN implements NR RAT (sometimes referred to herein as 5G RAT, 5G NR RAT, or NR). In some deployments, E-UTRAN may also implement NR RAT. In some deployments, NG-RAN may also implement LTE RAT.

[0005] The base stations used by the RAN may correspond to the RAN. An example of an E-UTRAN base station is an Evolved Universal Terrestrial Radio Access Network (E-UTRAN) Node B (also commonly denoted as an evolved Node B, enhanced Node B, eNode B, or eNB). An example of an NG-RAN base station is a Next Generation Node B (sometimes also referred to as a gNode B or gNB).

[0006] The RAN provides communication services together with external entities through its connection with the Core Network (CN). For example, E-UTRAN may utilize the Evolved Packet Core (EPC) and NG-RAN may utilize the 5G Core Network (5GC). BRIEF DESCRIPTION OF THE DRAWINGS

[0007] To easily identify the discussion of any particular element or action, the most significant digit(s) in a reference number refers to the drawing number that first introduces the element.

[0008] Figure 1 An encoder and a decoder in CSI feedback operation according to certain embodiments are illustrated.

[0009] Figure 2 A table of average cosine similarities between different spatial layers for different codebook configurations is illustrated.

[0010] Figure 3 is a table describing the four neural network model categories.

[0011] Figure 4A Three methods are illustrated that may be used with the embodiments disclosed herein to derive CSI for multi-rank CSI feedback.

[0012] Figure 4B Another method including a spatial layer group common neural network model that can be used with the embodiments disclosed herein is illustrated.

[0013] Figure 5 is a flow chart of a method for overhead allocation for ML-based CSI feedback according to one embodiment.

[0014] Figure 6 is a flow chart of a method for overhead allocation for ML-based CSI feedback according to another embodiment.

[0015] Figure 7 is a flow chart of a method for a UE to provide ML-based CSI feedback to a wireless network according to one embodiment.

[0016] Figure 8 is a flow chart of a method for a base station to configure ML-based CSI feedback in a wireless network according to one embodiment.

[0017] Fig. 9 An example architecture of a wireless communication system according to embodiments disclosed herein is illustrated.

[0018] Fig.10 A system for performing signaling between a wireless device and a network device according to embodiments disclosed herein is illustrated. DETAILED DESCRIPTION

[0019] Various embodiments are described with respect to UE. However, reference to UE is provided for illustrative purposes only. The example embodiments may be used with any electronic component that can establish a connection with a network and is configured with hardware, software, and / or firmware for exchanging information and data with the network. Therefore, UE as described herein is used to represent any suitable electronic component.

[0020] Downlink CSI (e.g., for frequency division duplex (FDD) operation) can be sent from the UE to the base station via a feedback channel. For example, the base station can use CSI feedback to reduce interference and increase throughput for large-scale multiple-input multiple-output (MIMO) communications. However, such feedback uses too much overhead. Vector quantization or codebook-based feedback can be used to reduce feedback overhead. However, the amount of feedback generated by these methods scales linearly with the number of transmit antennas, which can be difficult when using hundreds or thousands of centralized or distributed transmit antennas.

[0021] Artificial intelligence (AI) and / or machine learning (ML) can be used for CSI feedback enhancement to reduce overhead, improve accuracy and / or generate predictions. AI and / or ML can also be used, for example, for beam management (e.g., beam prediction in the time / spatial domain for overhead and latency reduction and beam selection accuracy improvement) and / or positioning accuracy enhancement.

[0022] CSI feedback using AI and / or ML can be formulated as a joint optimization of the encoder and decoder. For example, see Chao-Kai Wen, Wan-Ting Shih, and Shi Jin, "Deep Learning for Massive MIMO CSI Feedback," IEEE Wireless Communications Letters, Vol. 7, No. 5, October 2018. Since this early paper by Chao-Kai Wen et al., autoencoders and many variants have been considered. Image processing / video processing techniques have been used for CSI compression, which can be a natural choice considering the latest wave of applications of ML in image processing / video processing. In addition, when formulated in the right domain, CSI feedback has similarities with image / video streams.

[0023] At a high level, Figure 1The encoder 102 of the UE and the decoder 104 of the base station (e.g., gNB) in the AI-based CSI feedback operation according to certain embodiments are illustrated. As shown, the encoder 102 receives a downlink (DL) channel H or a DL precoder and outputs an AI-based CSI feedback. The encoder 102 learns the transformation from the original transformation matrix to the compressed representation (codeword) through training data. The decoder 104 learns the inverse transformation from the codeword to the original channel. Therefore, the decoder 104 may receive AI-based CSI feedback (codeword) from the encoder 102 and output a reconstructed channel H or a DL precoder (which may be represented by H). End-to-end learning (e.g., using an unsupervised learning algorithm) may be used to train the encoder 102 and the decoder 104. Typically, normalized mean square error (NMSE) or cosine similarity is an optimization metric. In some designs, the DL channel H may be replaced with a DL precoder. Thus, the encoder 102 takes the DL precoder as input and generates AI-based CSI feedback, and the decoder 104 takes the AI-based CSI feedback and reconstructs the DL precoder.

[0024] Various types of neural network (NN) encoders / decoders can be trained for different purposes, with different tradeoffs in complexity, overhead, and performance. Convolutional neural networks (CNNs) can be used, for example, for CSI feedback for frequency domain and spatial domain CSI reference signal (CSI-RS) compression. Other examples include the use of transformers or generative adversarial networks (GANs). Depending on the number and rank of receive antennas, channel feedback or precoding matrix feedback can be used. For example, in the case of four receive antennas, rank 1 and rank 2 feedback can potentially use an AI NN trained using feature vectors as input, while rank 3 and rank 4 can potentially use a channel matrix as input to the trained AI NN. Data preprocessing can be used for input to the AI ​​model. Preprocessing from the frequency domain to the time domain can be used, and some small paths can be removed before input to the AI ​​NN. The maximum rank indicates the maximum number of layers per UE, which corresponds to the lack of correlation or interference between the antennas of the UE. For example, rank 1 corresponds to a maximum of one spatial layer of the UE, rank 2 corresponds to a maximum of two spatial layers of the UE, rank 3 corresponds to a maximum of three layers of the UE, and rank 4 corresponds to a maximum of four layers of the UE.

[0025] CNN+RNN (recurrent NN) based NN can be used for CSI-RS compression in time, frequency and space domains. The input can be a time series with a set of CSI-RS configurations. Preprocessed time series such as frequency domain preprocessing (to time domain and remove small channel taps) and Doppler domain preprocessing can also be applied as AI input. Angular domain preprocessing is also possible, however, angular domain preprocessing may not be efficient in some specific implementations.

[0026] Channel Description

[0027] For a wireless channel, between a base station (e.g., gNB) and a UE, the highest few singular values ​​of the channel matrix or wideband covariance matrix are often much larger than the remaining singular values. For example, the precoder can be obtained using a singular value decomposition of the channel matrix H=USV, where H is a 4×32 matrix, V is a 32×32 matrix, U is a 4×4 matrix, and S is a 4×32 matrix for 32 transmit antennas at the base station and 4 receive antennas at the UE. The channel may not support full rank transmission. Therefore, the singular values ​​along the diagonal of the S matrix can be arranged from largest to smallest. As shown in the following example S matrix, the largest singular value can be much larger than the remaining singular values ​​along the diagonal:

[0028]

[0029] In addition, from the perspective of the strongest spatial layers, a compact or parsimonious representation may usually be sufficient because they can be well represented by a few clusters with significant influence. The V matrix can be represented as 32 column vectors (V = [v1, .., v32]. For example, the vector v1 corresponding to spatial layer 1 can be easily represented because it corresponds to the main deviation angle. Therefore, the vector v1 has a clear physical meaning and can be described in a compact way. However, for other spatial layers (e.g., layers 3 and 4), this may not be the case because the singular value decomposition may include leakage from the strongest deviation angle and weaker signals from other deviation angles. That is, there may be contributions from weaker clusters.

[0030] In the precoder calculation, it can also be observed that higher rank precoders are susceptible to subspace rotation. Therefore, the physical excitation representation of higher rank precoders (e.g., rank 3 and rank 4) may have more working problems than lower rank precoders (rank 1 and rank 2).

[0031] As an example, Figure 2A table of average cosine similarities between different spatial layers (spatial layer 1, spatial layer 2, spatial layer 3, and spatial layer 4) for different NR Release 16 (Rel. 16) Type II codebook configurations (confi-1, confi-2, confi-3, confi-4, confi-5, confi-6, confi-7, confi-8) is illustrated. The feedback overhead increases with the configuration value. Therefore, Type II codebook configuration 1 (config-1) uses the lowest feedback overhead, and Type II codebook configuration 8 (config-8) uses the highest feedback overhead. An average cosine similarity close to or above 0.9 can be considered to have a low quantization error, while a lower average cosine similarity value corresponds to a high quantization error. Thus, for example, a type II codebook configuration 1 (config-1) with low feedback overhead has a good average cosine similarity for spatial layer 1 (0.9274), but a poor average cosine similarity for spatial layer 3 (0.4969) and a poor average cosine similarity for spatial layer 4 (0.3696). As shown, for each of the type II codebook configurations, spatial layer 1 and spatial layer 2 have relatively good average cosine similarity values. However, spatial layer 3 and spatial layer 4 benefit from the increased feedback overhead at higher type II codebook configurations (e.g., the average cosine similarity increases to 0.7928 for spatial layer 3 and 0.7162 for spatial layer 4). Thus, the amount of benefit of increasing the feedback overhead depends on the spatial layer.

[0032] NN model types and methods

[0033] In general, there may be multiple NN model types and various methods to derive CSI for multi-rank CSI feedback. For example, Figure 3 A table describing a general classification of NN models that may be used with certain embodiments is illustrated. In particular, Figure 3 is a table describing four NN model categories which may be referred to herein as Type 1, Type 2, Type 3, and Type 4.

[0034] NN Model Type 1 includes a single network-side (e.g., gNB) trained ML model adopted by the UE. The network (NW) can select optimized loss functions for multi-user MIMO (MU-MIMO), coherent joint transmission (C-JT), etc. However, NN Model Type 1 may have high requirements for UE specific implementation. For example, the UE hardware may not support the NW-designed model and / or optimize for the NW-designed model. In addition, the model representation format (MRF) may not be compatible at the UE.

[0035] NN model type 2 includes a trained ML model on the UE side, which may be simpler for the UE and may allow for optimized hardware and model design for the UE. A single UE model may work with any base station (gNB). However, the loss function may not match the NW specific implementation. In a time slot, multiple ML models may need to be executed on the gNB side to receive from multiple UEs. MRF at the gNB may be an issue.

[0036] NN model type 3 is optimized separately for UE and base station (e.g., gNB) hardware, and the NW is allowed to choose the optimized loss function. The ML model can be kept proprietary for UE / NW separately. However, NN model type 3 may use a large amount of storage for UE and / or NW. Scalability may also be an issue for multiple vendors. In addition, it may not be possible to fine-tune the model. NN model type 3 can be trained at a facility where the NW and UE exchange gradient information for forward and backward propagation.

[0037] NN model type 4 is optimized separately for UE and base station (e.g., gNB) hardware. ML models can be kept proprietary separately for UE / NW. However, NN model type 4 may use a large training overhead caused by the sharing of intermediate training labels. In addition, there is a potential performance loss caused by mismatched models. For training, one side (e.g., gNB or UE) generates "labels" with CSI for its own trained AI model (Model A), and the other side provides "labels" and CSI for its own AI model (Model B). Note that Model A and Model B can be different (e.g., one model can be a transformer and the other model can be a CNN).

[0038] Figure 4A Three methods that can be used with the embodiments disclosed herein to derive CSI for multi-rank CSI feedback are illustrated. The first method (method 1) includes using a spatial layer common NN model. As shown in the figure, in method 1, the same single-rank NN model is used for each spatial layer (spatial layer 1, spatial layer 2, spatial layer 3, and spatial layer 4) to derive the precoder for each spatial layer.

[0039] The second method (method 2) involves using a rank-specific NN model. Figure 4A As shown, for example, if the maximum rank is limited to four, there are four NN models (NN-1, NN-2, NN-3, and NN-4). The first NN model (NN-1) is used for rank 1 corresponding to spatial layer 1. The second NN model (NN-2) is used for rank 2 corresponding to spatial layer 1 and spatial layer 2. The third NN model (NN-3) is used for rank 3 corresponding to spatial layer 1, spatial layer 2, and spatial layer 3. The fourth NN model (NN-4) is used for rank 4 corresponding to spatial layer 1, spatial layer 2, spatial layer 3, and spatial layer 4.

[0040] The third approach (Method 3) involves using a spatial layer specific NN model. Figure 4A As shown, for example, if the maximum rank is limited to four, there are four NN models (NN-1, NN-2, NN-3, and NN-4). The first NN model (NN-1) is used for spatial layer 1 for rank 1, rank 2, rank 3, and rank 4. The second NN model (NN-2) is used for spatial layer 2 for rank 2, rank 3, and rank 4. The third NN model (NN-3) is used for spatial layer 3 for rank 3 and rank 4. The fourth NN model (NN-4) is used for spatial layer 4 for rank 4.

[0041] In certain embodiments, other methods may also be used to derive CSI for multi-rank CSI feedback. For example, spatial layer group and / or rank group specific methods may be used. For example, Figure 4B Another method (method 1A) including a common NN model for a spatial layer group according to one embodiment is illustrated. In the illustrated example, four spatial layers are divided into two spatial layer groups. A first NN model (NN-1) is used for a first spatial layer group including spatial layer 1 and spatial layer 2. A second NN model (NN-2) is used for a second spatial layer group including spatial layer 3 and spatial layer 4.

[0042] For NN model type 1 and NN model type 2, one or more NN models may be deployed or configured at the UE and NW. The encoder of the NN model resides at the UE, and the decoder of the NN model resides at the NW. For method 1, a single NN model is deployed or configured for the UE and NW. For methods 1A, 2, and 3, there may be multiple NN models deployed or configured for the UE and NW.

[0043] For NN model type 3 and NN model type 4, because the UE-side NN model is trained or optimized separately from the NW-side NN model at the UE, one or more NN model pairs may be deployed at the UE and the NW. The encoder of the UE-side NN model resides at the UE. The decoder of the NW-side NN model, which is the counterpart of the UE-side NN model, resides at the NW. For method 1, a single NN model pair is deployed or configured for the UE and the NW. For methods 1A, 2, and 3, there may be multiple NN model pairs deployed or configured for the UE and the NW.

[0044] In certain embodiments, the NW may configure one or more NN models (or model pairs) corresponding to a plurality of different feedback overhead sizes for the UE. Alternatively, from another perspective, the number of NN models is not increased. Instead, the feedback overhead size is considered to be a property of the NN model. For example, the NN model in method 1 may be associated with 30 bits or 60 bits. As a further example, in method 2, NN-3 may be associated with 80 bits or 100 bits, and / or NN-4 may be associated with 120 bits or 240 bits.

[0045] Selective feedback overhead per rank, spatial layer or spatial layer group

[0046] In certain embodiments, the feedback overhead for different ranks, spatial layers, or spatial layer groups may be different for AI-stimulated CSI feedback. In one such embodiment, the overhead allocation may be configured by a base station (e.g., a gNB). The base station may explicitly indicate the overhead allocation for different spatial layers to the UE. For example, the base station may indicate 40 bits allocated for each of spatial layers 1 and 2, and 60 bits allocated for each of spatial layers 3 and 4. The UE then inputs the spatial layer 1 information into a first model associated with 40 bits of feedback overhead, the spatial layer 2 information into a first model associated with 40 bits of feedback overhead, the spatial layer 3 information into a second model associated with 60 bits of feedback overhead, and the spatial layer 4 information into a second model associated with 60 bits of feedback overhead. Therefore, the total feedback overhead in this example is 40 bits + 40 bits + 60 bits + 60 bits = 200 bits. In another embodiment, the base station indicates to the UE the ratio between the feedback overhead for each spatial layer and the total number of feedback bits. In yet another embodiment, the base station indicates to the UE the ratio between spatial layer groups.

[0047] In other embodiments, the overhead allocation is selected by the UE and reported to the base station. For example, the UE may allocate 40 bits for each of spatial layers 1 and 2, and the UE may allocate 60 bits for each of spatial layers 3 and 4. The UE reports the amount of overhead allocation for each spatial layer in the CSI report, or the UE reports the feedback overhead for at least two groups of spatial layers in the CSI feedback. The UE may report the ratio of the feedback overhead between at least two groups of spatial ranks. For example, the UE may report that the feedback overhead for {rank 1, rank 2} is 3 / 4 of the feedback overhead for {rank 3, rank 4}.

[0048] Figure 5Flowchart of a method 500 for overhead allocation for ML-based CSI feedback according to one embodiment. At box 502, for an NN model type (e.g., type 1, 2, 3, or 4), and for a method, the UE reports a preferred or capable feedback size in UE capability signaling. At box 504, for an NN model type (e.g., type 1, 2, 3, or 4), and for a method, the NW configures the UE with NN models for multiple feedback sizes (or NN model pairs for type 3 or type 4) based on the reported feedback size from the UE. In box 506, box 508, or box 510, depending on the method, the NW indicates the overhead allocation to the UE. In box 512, the UE indicates to use a different NN model for each spatial layer or spatial layer group or rank based on the NW indication of the overhead allocation. Figure 5 Example NW indications are shown for Method 1A (block 506), Method 2 (block 508), and Method 3 (block 510). However, the present disclosure is not limited thereto, and the NW indication may also be applied to other methods of deriving CSI for multi-rank CSI feedback.

[0049] As shown in block 506, when the method includes method 1A discussed above, NW indicates a plurality of feedback sizes for a spatial layer group. Figure 4B , NW may indicate a first feedback size for a first spatial layer group including spatial layer 1 and spatial layer 2, and NW may indicate a second feedback size for a second spatial layer group including spatial layer 3 and spatial layer 4. Alternatively, NW may indicate a total feedback size for each spatial layer group and a ratio of the allocated feedback size to the total feedback size. At block 512 of method 500, the UE uses a first NN model (NN-1) associated with a first feedback size indicated by the NW for spatial layer 1 (for rank 1, rank 2, rank 3, and rank 4) and spatial layer 2 (for rank 2, rank 3, and rank 4). The UE also uses a second NN model (NN-2) associated with a second feedback size indicated by the NW for spatial layer 3 (for rank 3 and rank 4) and spatial layer 4 (for rank 4).

[0050] As shown in block 508, when the method includes method 2 discussed above, NW indicates that bits are divided between different spatial layers in the rank (where the division may be unequal). Figure 4AIn the method 2 shown, for example, NW may indicate that a first number of feedback bits (e.g., 30 bits) is used for spatial layer 1 in rank 1. For rank 2, NW may indicate that a second number of feedback bits (e.g., 60 bits) is used, which is divided between spatial layer 1 and spatial layer 2. For rank 3, NW may indicate that a third number of feedback bits (e.g., 120 bits) is used, which is divided between spatial layer 1, spatial layer 2, and spatial layer 3 in rank 3. As an example, for rank 3, NW may indicate that 30 bits are used for each of spatial layers 1 and spatial layer 2 and 60 bits are used for spatial layer 3. For rank 4, NW may indicate that a fourth number of feedback bits (e.g., 240 bits) is used, which is divided between spatial layer 1, spatial layer 2, spatial layer 3, and spatial layer 4. As an example, for rank 4, NW may indicate that 30 bits are used for each of spatial layers 1 and spatial layer 2, 60 bits are used for spatial layer 3, and 120 bits are used for spatial layer 4. However, a person skilled in the art will recognize from the disclosure herein that any partitioning may be used for rank 2, rank 3, or rank 4, including equal partitioning between spatial layers. Alternatively, the NW may indicate a total number of feedback bits for each spatial layer and a ratio of the allocated bits to the total number of feedback bits. At box 512 of method 500, the UE uses a first NN model (NN-1) associated with a first number of feedback bits indicated by the NW for rank 1 (spatial layer 1), a second NN model (NN-2) associated with a second number of feedback bits indicated by the NW for rank 2 (spatial layer 1 and spatial layer 2), a third NN model (NN-3) associated with a third number of feedback bits indicated by the NW for rank 3 (spatial layer 1, spatial layer 2, and spatial layer 3), and a fourth NN model (NN-4) associated with a fourth number of feedback bits indicated by the NW for rank 4 (spatial layer 1, spatial layer 2, spatial layer 3, and spatial layer 4).

[0051] As shown in block 510, when the method includes method 3 discussed above, NW indicates feedback bits for different spatial layers (where the number of feedback bits for each spatial layer may not be equal). Figure 4AIn the method 3 shown, for example, the NW may indicate that a first number of feedback bits (e.g., 30 bits) is used for spatial layer 1 in rank 1, rank 2, rank 3, and rank 4. The NW may indicate that a second number of feedback bits (e.g., 30 bits) is used for spatial layer 2 in rank 2, rank 3, and rank 4. The NW may indicate that a third number of feedback bits (e.g., 60 bits) is used for spatial layer 3 in rank 3 and rank 4. The NW may indicate that a fourth number of feedback bits (e.g., 90 bits) is used for spatial layer 4 in rank 4. Alternatively, the NW may indicate the total number of feedback bits for each spatial layer and the ratio of the allocated bits to the total number of feedback bits. At block 512 of method 500, the UE uses a first NN model (NN-1) associated with a first number of feedback bits indicated by the NW for spatial layer 1 (for rank 1, rank 2, rank 3, and rank 4). The UE uses a second NN model (NN-2) associated with a second number of feedback bits indicated by the NW for spatial layer 2 (for rank 2, rank 3, and rank 4). The UE uses a third NN model (NN-3) associated with a third number of feedback bits indicated by the NW for spatial layer 3 (for rank 3 and rank 4). The UE uses a fourth NN model (NN-4) associated with a fourth number of feedback bits indicated by the NW for spatial layer 4 (for rank 4).

[0052] Figure 6 is a flow chart of a method 600 for overhead allocation for ML-based CSI feedback according to another implementation. Figure 6 The method 600 shown is similar to Figure 5 Method 500 is shown, except that in method 600 the UE determines and indicates the overhead allocation to the NW.

[0053] At box 602, for a NN model type (e.g., type 1, 2, 3, or 4), and for a method, the UE reports a preferred or capable feedback size in UE capability signaling. At box 604, for a NN model type (e.g., type 1, 2, 3, or 4), and for the method, the NW configures the UE with NN models for multiple feedback sizes (or NN model pairs for type 3 or type 4) based on the reported feedback size from the UE. In box 606, box 608, or box 610, depending on the method, the UE determines and indicates an overhead allocation to the base station. In box 612, the UE uses a different NN model for each spatial layer or spatial layer group or rank based on the UE indication of the overhead allocation provided to the base station. Figure 6 Example UE indications are shown for Method 1A (block 606), Method 2 (block 608), and Method 3 (block 610). However, the present disclosure is not limited thereto, and UE indications may also be applied to other methods of deriving CSI for multi-rank CSI feedback.

[0054] As shown in block 606, when the method includes method 1A discussed above, the UE indicates multiple feedback sizes for the spatial layer groups. Figure 4B , the UE may indicate a first feedback size for a first spatial layer group including spatial layer 1 and spatial layer 2, and the UE may indicate a second feedback size for a second spatial layer group including spatial layer 3 and spatial layer 4. Alternatively, the UE may indicate a total feedback size for each spatial layer group and a ratio of the total. At block 612 of method 600, the UE uses a first NN model (NN-1) associated with a first feedback size indicated to the NW for spatial layer 1 (for rank 1, rank 2, rank 3, and rank 4) and spatial layer 2 (for rank 2, rank 3, and rank 4). The UE uses a second NN model (NN-2) associated with a second feedback size indicated to the NW for spatial layer 3 (for rank 3 and rank 4) and spatial layer 4 (for rank 4).

[0055] As shown in block 608, when the method includes method 2 discussed above, the UE indicates that bits are divided between different spatial layers in the rank (where the division may be unequal). Figure 4AIn the method 2 shown, for example, the UE may indicate the use of a first number of feedback bits (e.g., 30 bits) for spatial layer 1 in rank 1. For rank 2, the UE may indicate the use of a second number of feedback bits (e.g., 60 bits) divided between spatial layer 1 and spatial layer 2. For rank 3, the UE may indicate the use of a third number of feedback bits (e.g., 120 bits) divided between spatial layer 1, spatial layer 2, and spatial layer 3 in rank 3. As an example, for rank 3, the UE may indicate the use of 30 bits for each of spatial layers 1 and 2, and 60 bits for spatial layer 3. For rank 4, the UE may indicate the use of a fourth number of feedback bits (e.g., 240 bits) divided between spatial layer 1, spatial layer 2, spatial layer 3, and spatial layer 4. As an example, for rank 4, the UE may indicate the use of 30 bits for each of spatial layers 1 and 2, 60 bits for spatial layer 3, and 120 bits for spatial layer 4. However, a person skilled in the art will recognize from the disclosure herein that any partitioning may be used for rank 2, rank 3, or rank 4, including equal partitioning between spatial layers. Alternatively, the UE may indicate the total number of feedback bits for each spatial layer and the ratio of the total number. At box 612 of method 600, the UE uses a first NN model (NN-1) associated with a first number of feedback bits indicated to the NW for rank 1 (spatial layer 1), a second NN model (NN-2) associated with a second number of feedback bits indicated to the NW for rank 2 (spatial layer 1 and spatial layer 2), a third NN model (NN-3) associated with a third number of feedback bits indicated to the NW for rank 3 (spatial layer 1, spatial layer 2, and spatial layer 3), and a fourth NN model (NN-4) associated with a fourth number of feedback bits indicated to the NW for rank 4 (spatial layer 1, spatial layer 2, spatial layer 3, and spatial layer 4).

[0056] As shown in block 610, when the method includes method 3 discussed above, the UE indicates feedback bits for different spatial layers (where the number of feedback bits for each spatial layer may not be equal). Figure 4AIn the method 3 shown, for example, the UE may indicate the use of a first number of feedback bits (e.g., 30 bits) for spatial layer 1 in rank 1, rank 2, rank 3, and rank 4. The UE may indicate the use of a second number of feedback bits (e.g., 30 bits) for spatial layer 2 in rank 2, rank 3, and rank 4. The UE may indicate the use of a third number of feedback bits (e.g., 60 bits) for spatial layer 3 in rank 3 and rank 4. The UE may indicate the use of a fourth number of feedback bits (e.g., 90 bits) for spatial layer 4 in rank 4. Alternatively, the UE may indicate the total number of feedback bits for each spatial layer and the ratio of the total number. At block 612 of method 600, the UE uses a first NN model (NN-1) associated with the first number of feedback bits indicated to the NW for spatial layer 1 (for rank 1, rank 2, rank 3, and rank 4). The UE uses a second NN model (NN-2) associated with the second number of feedback bits indicated to the NW for spatial layer 2 (for rank 2, rank 3, and rank 4). The UE uses a third NN model (NN-3) associated with a third number of feedback bits indicated to the NW for spatial layer 3 (for rank 3 and rank 4). The UE uses a fourth NN model (NN-4) associated with a fourth number of feedback bits indicated to the NW for spatial layer 4 (for rank 4).

[0057] Implementing differential feedback overhead

[0058] In some embodiments, the use of different feedback overheads (e.g., per rank, per spatial layer, or per spatial layer group) is triggered in response to a predetermined condition, threshold, or formula. For example, it may be triggered when the total feedback overhead is low. Figure 2 , it can be seen that when the total feedback overhead is large (e.g., when type II codebook configuration 8 (config-8) is used), the need to support different feedback overheads for different spatial layers may be reduced due to the relatively high average cosine similarity of each spatial layer in the spatial layer. However, when the total feedback overhead is low (e.g., when type II codebook configuration 1 (config-1) is used), there is a greater need to support different feedback overheads for different spatial layers (e.g., more feedback bits may be used for spatial layer 3 and spatial layer 4 to increase the average cosine similarity of these layers). Therefore, when, for example, the type II codebook configuration is associated with a relatively low total feedback overhead, the NW or UE may trigger the use of different numbers of feedback bits. Additionally, or in other embodiments, the trigger may be responsive to determining low channel quality or low average cosine similarity of one or more spatial layers compared to a predetermined threshold.

[0059] Figure 7700 is a flowchart of a method 700 for a UE to provide ML-based CSI feedback to a wireless network according to one embodiment. In box 702, for NN model types and methods for deriving CSI for multi-rank CSI feedback, method 700 includes reporting one or more feedback sizes to the wireless network in a UE capability message. In certain embodiments, the NN model type and / or method is predetermined (e.g., defined in a specification or standard). In other embodiments, the UE and / or network may selectively be selected from a plurality of different NN model types and / or methods. In box 704, for NN model types and methods, method 700 includes processing the configuration of a plurality of NN models associated with one or more feedback sizes from the wireless network. In box 706, method 700 includes processing overhead allocation for one or more feedback sizes. In box 708, method 700 includes generating multi-rank CSI feedback using different NN models from a plurality of NN models for each spatial layer or spatial layer group or rank based on the overhead allocation.

[0060] In certain embodiments of method 700, processing the overhead allocation includes receiving an indication of the overhead allocation from the wireless network.

[0061] In certain embodiments of method 700, processing the overhead allocation includes: determining the overhead allocation at the UE based on a plurality of NN models configured by the wireless network for one or more feedback sizes; and transmitting an indication of the overhead allocation from the UE to the wireless network.

[0062] In certain embodiments of method 700, one or more feedback sizes are selected from the group consisting of at least one of: a total feedback size, a per spatial layer feedback size, a per spatial layer group feedback size, a per rank feedback size, and a per NN model feedback size.

[0063] In certain embodiments of method 700, the method includes using a spatial layer group common NN model, and the overhead allocation includes an indication of: different feedback sizes for different spatial layer groups; or a total feedback size for each spatial layer group in the different spatial layer groups and a ratio of the allocated feedback size to the total feedback size. Generating multi-rank CSI feedback using different NN models may include using at least: a first NN model associated with a first feedback size for a first spatial layer group; and a second NN model associated with a second feedback size for a second spatial layer group.

[0064] In certain embodiments of method 700, the method includes using a rank-specific NN model, and wherein the overhead allocation includes an indication of: a division of feedback bits between different spatial layers in an indicated rank; or a total number of feedback bits for each spatial layer in the different spatial layers in the indicated rank and a ratio of the allocated bits to the total number of feedback bits. In one such embodiment, the division includes an unequal number of feedback bits between different spatial layers in the indicated rank. Additionally, or in other embodiments, using different NN models to generate multi-rank CSI feedback includes using two or more of the following: a first NN model associated with a first number of feedback bits for a first rank; a second NN model associated with a second number of feedback bits divided between different spatial layers for a second rank; a third NN model associated with a third number of feedback bits divided between different spatial layers for a third rank; and a fourth NN model associated with a fourth number of feedback bits divided between different spatial layers for a fourth rank.

[0065] In certain embodiments of method 700, the method includes using spatial layer-specific NN models, and wherein the overhead allocation includes an indication of: a number of feedback bits for different spatial layers; a total number of feedback bits for each spatial layer in the different spatial layers and a ratio of the allocated bits to the total number of feedback bits. In one such embodiment, the indication indicates that the number of feedback bits is an unequal number between different spatial layers. Additionally, or in other embodiments, using different NN models to generate multi-rank CSI feedback includes using two or more of: a first NN model associated with a first number of feedback bits for a first spatial layer; a second NN model associated with a second number of feedback bits for a second spatial layer; a third NN model associated with a third number of feedback bits for a third spatial layer; and a fourth NN model associated with a fourth number of feedback bits for a fourth spatial layer.

[0066] In certain embodiments of method 700, NN model types are optimized separately for UE hardware and base station hardware, and multiple NN models each include an NN model pair corresponding to an encoder at the UE and a decoder at the base station.

[0067] In certain embodiments, method 700 further includes: determining a total feedback overhead for codebook configuration; comparing the total feedback overhead to a threshold to determine a trigger event; and in response to the trigger event, processing overhead allocation and generating multi-rank CSI feedback using different NNs based on the overhead allocation.

[0068] Embodiments contemplated herein include an apparatus comprising means for performing one or more elements of the method 700. The apparatus may be, for example, an apparatus that is a UE (such as the wireless device 1002 as a UE, as described herein).

[0069] Embodiments contemplated herein include one or more non-transitory computer-readable media including instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform one or more elements of method 700. The non-transitory computer-readable medium may be, for example, a memory of a UE (such as memory 1006 of wireless device 1002 as a UE, as described herein).

[0070] Embodiments contemplated herein include an apparatus comprising logical components, modules, or circuits for performing one or more elements of the method 700. The apparatus may be, for example, an apparatus of a UE (such as the wireless device 1002 as a UE, as described herein).

[0071] Embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of the method 700. The apparatus may be, for example, an apparatus of a UE (such as the wireless device 1002 as a UE, as described herein).

[0072] Embodiments contemplated herein include a signal as described in or in connection with one or more elements of method 700 .

[0073] Embodiments contemplated herein include a computer program or computer program product including instructions, wherein execution of the program by a processor causes the processor to perform one or more elements of the method 700. The processor may be a processor of a UE (such as the processor 1004 of the wireless device 1002 as a UE, as described herein). These instructions may be located, for example, in a processor of a UE and / or on a memory (such as the memory 1006 of the wireless device 1002 as a UE, as described herein).

[0074] Figure 8800 is a flow chart of a method 800 for a base station to configure ML-based CSI feedback in a wireless network according to one embodiment. In box 802, for NN model types and methods for deriving CSI for multi-rank CSI feedback, method 800 includes receiving one or more feedback sizes from a user equipment (UE) in a UE capability message. In certain embodiments, the NN model type and / or method is predetermined (e.g., defined in a specification or standard). In other embodiments, the UE and / or network may selectively select from a plurality of different NN model types and / or methods. In box 804, for NN model types and methods, method 800 includes configuring a plurality of NN models associated with one or more feedback sizes for the UE. In box 806, method 800 includes processing overhead allocation for one or more feedback sizes.

[0075] In one embodiment of the method 800, processing the overhead allocation includes receiving, at a base station, an indication of the overhead allocation from a UE.

[0076] In one embodiment of the method 800, processing the overhead allocation includes: determining the overhead allocation at the base station based on a plurality of NN models configured by the wireless network for one or more feedback sizes; and transmitting an indication of the overhead allocation from the base station to the UE.

[0077] In one embodiment of method 800, the one or more feedback sizes are selected from the group consisting of at least one of: a total feedback size, a per spatial layer feedback size, a per spatial layer group feedback size, a per rank feedback size, and a per NN model feedback size.

[0078] In one embodiment of method 800, the method includes using a spatial layer group common NN model, and the overhead allocation includes an indication of: different feedback sizes for different spatial layer groups; or a total feedback size for each spatial layer group in different spatial layer groups and a ratio of the allocated feedback size to the total feedback size.

[0079] In one embodiment of method 800, the method includes using a rank-specific NN model, and wherein the overhead allocation includes an indication of: a division of feedback bits between different spatial layers in the indicated rank; or a total number of feedback bits for each of the different spatial layers in the indicated rank and a ratio of the allocated bits to the total number of feedback bits. In one such embodiment, the division includes an unequal number of feedback bits between different spatial layers in the indicated rank.

[0080] In one embodiment of method 800, the method includes using a spatial layer specific NN model, and wherein the overhead allocation includes an indication of: a number of feedback bits for different spatial layers; a total number of feedback bits for each of the different spatial layers and a ratio of the allocated bits to the total number of feedback bits. In one such embodiment, the indication indicates that the number of feedback bits is an unequal number between different spatial layers.

[0081] In one embodiment of method 800, NN model types are optimized separately for UE hardware and base station hardware, and multiple NN models each include an NN model pair corresponding to an encoder at the UE and a decoder at the base station.

[0082] In one embodiment, the method 800 further includes: determining a total feedback overhead for codebook configuration; comparing the total feedback overhead to a threshold to determine a triggering event; and processing the overhead allocation in response to the triggering event.

[0083] Embodiments contemplated herein include an apparatus comprising means for performing one or more elements of method 800. For example, the apparatus may be an apparatus of a base station (such as network device 1018 as a base station, as described herein).

[0084] Embodiments contemplated herein include one or more non-transitory computer-readable media including instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform one or more elements of method 800. For example, the non-transitory computer-readable medium may be a memory of a base station (such as memory 1022 of network device 1018 acting as a base station, as described herein).

[0085] Embodiments contemplated herein include an apparatus comprising logic, modules, or circuits operable to perform one or more elements of method 800. For example, the apparatus may be an apparatus of a base station (such as network device 1018 as a base station, as described herein).

[0086] Embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of method 800. For example, the apparatus may be an apparatus of a base station (such as network device 1018 as a base station, as described herein).

[0087] Embodiments contemplated herein include a signal as described in or in connection with one or more elements of method 800 .

[0088] Embodiments contemplated herein include a computer program or computer program product including instructions, wherein execution of the program by a processing element causes the processing element to perform one or more elements of method 800. The processor may be a processor of a base station (such as processor 1020 of network device 1018 as a base station, as described herein). For example, these instructions may be located in a processor and / or on a memory of a base station (such as memory 1022 of network device 1018 as a base station, as described herein).

[0089] Fig. 9 An example architecture of a wireless communication system 900 according to an embodiment disclosed herein is illustrated. The description provided below is for an example wireless communication system 900 operating in conjunction with the LTE system standard and / or the 5G or NR system standard provided in the 3GPP technical specifications.

[0090] like Fig. 9 As shown, wireless communication system 900 includes UE 902 and UE 904 (although any number of UEs may be used). In this example, UE 902 and UE 904 are illustrated as smartphones (e.g., handheld touch screen mobile computing devices capable of connecting to one or more cellular networks), but may also include any mobile or non-mobile computing device configured for wireless communication.

[0091] UE 902 and UE 904 may be configured to be communicatively coupled to RAN 906. In an embodiment, RAN 906 may be NG-RAN, E-UTRAN, etc. UE 902 and UE 904 utilize connections (or channels) (shown as connection 908 and connection 910, respectively) with RAN 906, where each connection (or channel) includes a physical communication interface. RAN 906 may include one or more base stations, such as base station 912 and base station 914, to implement connection 908 and connection 910.

[0092] In this example, connection 908 and connection 910 are air interfaces that enable such communicative coupling and may conform to the RAT used by the RAN 906 , such as, for example, LTE and / or NR.

[0093] In some embodiments, UE 902 and UE 904 may also communicate data directly via side link interface 916. UE 904 is shown as being configured to access an access point (shown as AP 918) via connection 920. As an example, connection 920 may include a local wireless connection, such as any connection that complies with the IEEE 802.11 protocol, wherein AP 918 may include In this example, AP 918 may be connected to another network (eg, the Internet) without going through CN 924.

[0094] In an embodiment, UE 902 and UE 904 may be configured to communicate with each other or with base station 912 and / or base station 914 over a multi-carrier communication channel using orthogonal frequency division multiplexing (OFDM) communication signals according to various communication techniques, such as, but not limited to, orthogonal frequency division multiple access (OFDMA) communication techniques (e.g., for downlink communication) or single carrier frequency division multiple access (SC-FDMA) communication techniques (e.g., for uplink and ProSe or sidelink communication), although the scope of the embodiments is not limited in this respect. The OFDM signal may include multiple orthogonal subcarriers.

[0095] In some embodiments, all or part of the base station 912 or the base station 914 may be implemented as one or more software entities running on a server computer as part of a virtual network. In addition, or in other embodiments, the base station 912 or the base station 914 may be configured to communicate with each other via the interface 922. In an embodiment where the wireless communication system 900 is an LTE system (e.g., when the CN 924 is an EPC), the interface 922 may be an X2 interface. The X2 interface may be defined between two or more base stations (e.g., two or more eNBs, etc.) connected to the EPC and / or between two eNBs connected to the EPC. In an embodiment where the wireless communication system 900 is an NR system (e.g., when the CN 924 is a 5GC), the interface 922 may be an Xn interface. The Xn interface is defined between two or more base stations (e.g., two or more gNBs, etc.) connected to the 5GC, between the base station 912 (e.g., gNB) and the eNB connected to the 5GC, and / or between two eNBs connected to the 5GC (e.g., CN 924).

[0096] The RAN 906 is shown communicatively coupled to the CN 924. The CN 924 may include one or more network elements 926 configured to provide various data and telecommunication services to customers / subscribers (e.g., UE 902 and users of UE 904) connected to the CN 924 via the RAN 906. The components of the CN 924 may be implemented in one physical device or separate physical devices including components for reading and executing instructions from a machine-readable or computer-readable medium (e.g., a non-transitory machine-readable storage medium).

[0097] In an embodiment, CN 924 may be an EPC, and RAN 906 may be connected to CN 924 via an S1 interface 928. In an embodiment, S1 interface 928 may be divided into two parts: an S1 user plane (S1-U) interface that carries service data between base station 912 or base station 914 and a serving gateway (S-GW); and an S1-MME interface that is a signaling interface between base station 912 or base station 914 and a mobility management entity (MME).

[0098] In an embodiment, CN 924 may be a 5GC, and RAN 906 may be connected to CN 924 via an NG interface 928. In an embodiment, NG interface 928 may be divided into two parts: an NG user plane (NG-U) interface, which carries service data between base station 912 or base station 914 and a user plane function (UPF); and an S1 control plane (NG-C) interface, which is a signaling interface between base station 912 or base station 914 and an access and mobility management function (AMF).

[0099] Generally, the application server 930 may be an element that provides applications that use Internet Protocol (IP) bearer resources with the CN 924 (e.g., packet-switched data services). The application server 930 may also be configured to support one or more communication services (e.g., VoIP sessions, group communication sessions, etc.) for the UE 902 and UE 904 via the CN 924. The application server 930 may communicate with the CN 924 via an IP communication interface 932.

[0100] Fig.10 A system 1000 for performing signaling 1034 between a wireless device 1002 and a network device 1018 according to an embodiment disclosed herein is illustrated. The system 1000 may be part of a wireless communication system as described herein. The wireless device 1002 may be, for example, a UE of a wireless communication system. The network device 1018 may be, for example, a base station (e.g., an eNB or a gNB) of a wireless communication system.

[0101] The wireless device 1002 may include one or more processors 1004. The processor 1004 may execute instructions to perform various operations of the wireless device 1002, as described herein. The processor 1004 may include one or more baseband processors implemented using, for example, a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a controller, a field programmable gate array (FPGA) device, another hardware device, a firmware device, or any combination thereof configured to perform the operations described herein.

[0102] The wireless device 1002 may include a memory 1006. The memory 1006 may be a non-transitory computer-readable storage medium that stores instructions 1008 (which may include, for example, instructions executed by the processor 1004). The instructions 1008 may also be referred to as program code or a computer program. The memory 1006 may also store data used by the processor 1004 and results computed by the processor.

[0103] The wireless device 1002 may include one or more transceivers 1010, which may include radio frequency (RF) transmitter and / or receiver circuitry that uses an antenna 1012 of the wireless device 1002 to facilitate signaling (e.g., signaling 1034) sent or received by the wireless device 1002 with other devices (e.g., network device 1018) according to a corresponding RAT.

[0104] The wireless device 1002 may include one or more antennas 1012 (e.g., one, two, four, or more). For implementations with multiple antennas 1012, the wireless device 1002 may take advantage of the spatial diversity of these multiple antennas 1012 to transmit and / or receive multiple different data streams on the same time-frequency resources. This behavior may be referred to as, for example, MIMO behavior (referring to the multiple antennas used at each of the transmitting device and the receiving device to implement this aspect). MIMO transmission by the wireless device 1002 may be implemented based on precoding (or digital beamforming) applied at the wireless device 1002, which multiplexes the data streams between the antennas 1012 based on known or assumed channel characteristics so that each data stream is received at an appropriate signal strength relative to the other streams and at a desired location in the spatial domain (e.g., the location of the receiver associated with the data stream). Certain implementations may use a single-user MIMO (SU-MIMO) approach (where the data streams are all directed to a single receiver) and / or a multi-user MIMO (MU-MIMO) approach (where individual data streams may be directed to individual (different) receivers in different locations in the spatial domain).

[0105] In certain embodiments with multiple antennas, the wireless device 1002 may implement analog beamforming techniques whereby the phases of signals transmitted by antennas 1012 are relatively adjusted such that the (joint) transmissions of antennas 1012 are directional (this is sometimes referred to as beam steering).

[0106] The wireless device 1002 may include one or more interfaces 1014. The interfaces 1014 may be used to provide input to or output from the wireless device 1002. For example, a wireless device 1002 that is a UE may include an interface 1014, such as a microphone, a speaker, a touch screen, buttons, etc., to allow a user of the UE to provide input and / or output to the UE. Other interfaces of such a UE may be composed of transmitters, receivers, and other circuit systems that allow the UE to communicate with other devices (e.g., in addition to the transceiver 1010 / antenna 1012 described above), and may communicate according to known protocols (e.g., etc.) to perform the operation.

[0107] The wireless device 1002 may include a CSI feedback module 1016. The CSI feedback module 1016 may be implemented via hardware, software, or a combination thereof. For example, the CSI feedback module 1016 may be implemented as a processor, circuit, and / or instructions 1008 stored in the memory 1006 and executed by the processor 1004. In some examples, the CSI feedback module 1016 may be integrated within the processor 1004 and / or the transceiver 1010. For example, the CSI feedback module 1016 may be implemented by a combination of software components (e.g., executed by a DSP or a general purpose processor) and hardware components (e.g., logic gates and circuits) within the processor 1004 or the transceiver 1010.

[0108] The CSI feedback module 1016 may be used in various aspects of the present disclosure, such as Figure 5 , Figure 6 and Figure 7 In addition, the CSI feedback module 1016 may include an encoder, such as Figure 1 Encoder 102 is shown.

[0109] The network device 1018 may include one or more processors 1020. The processor 1020 may execute instructions to perform various operations of the network device 1018 as described herein. The processor 1020 may include one or more baseband processors implemented using, for example, a CPU, a DSP, an ASIC, a controller, an FPGA device, another hardware device, a firmware device, or any combination thereof configured to perform the operations described herein.

[0110] The network device 1018 may include a memory 1022. The memory 1022 may be a non-transitory computer-readable storage medium that stores instructions 1024 (which may include, for example, instructions executed by the processor 1020). The instructions 1024 may also be referred to as program code or a computer program. The memory 1022 may also store data used by the processor 1020 and results calculated by the processor.

[0111] The network device 1018 may include one or more transceivers 1026, which may include RF transmitter and / or receiver circuits that use an antenna 1028 of the network device 1018 to facilitate signaling (e.g., signaling 1034) sent or received by the network device 1018 with other devices (e.g., wireless device 1002) according to the corresponding RAT.

[0112] The network device 1018 may include one or more antennas 1028 (e.g., one, two, four, or more). In embodiments with multiple antennas 1028, the network device 1018 may perform MIMO, digital beamforming, analog beamforming, beam steering, etc. as described above.

[0113] The network device 1018 may include one or more interfaces 1030. The interface 1030 may be used to provide input to or output from the network device 1018. For example, the network device 1018 as a base station may include an interface 1030 composed of a transmitter, a receiver and other circuits (e.g., in addition to the transceiver 1026 / antenna 1028 already described), which enables the base station to communicate with other equipment in the core network and / or enables the base station to communicate with external networks, computers, databases, etc., for the purpose of operating, managing and maintaining the base station or other equipment operably connected to the base station.

[0114] The network device 1018 may include a CSI feedback module 1032. The CSI feedback module 1032 may be implemented via hardware, software, or a combination thereof. For example, the CSI feedback module 1032 may be implemented as a processor, circuit, and / or instructions 1024 stored in the memory 1022 and executed by the processor 1020. In some examples, the CSI feedback module 1032 may be integrated within the processor 1020 and / or the transceiver 1026. For example, the CSI feedback module 1032 may be implemented by a combination of software components (e.g., executed by a DSP or general purpose processor) and hardware components (e.g., logic gates and circuits) within the processor 1020 or the transceiver 1026.

[0115] The CSI feedback module 1032 may be used in various aspects of the present disclosure, such as Figure 5 , Figure 6 and Figure 8 In addition, the CSI feedback module 1032 may include a decoder, such as Figure 1 Decoder 104 is shown.

[0116] For one or more embodiments, at least one of the components set forth in one or more of the foregoing figures may be configured to perform one or more operations, techniques, processes, and / or methods as described herein. For example, a baseband processor as described herein in conjunction with one or more of the foregoing figures may be configured to operate according to one or more of the examples set forth herein. For another example, a circuit system associated with a UE, a base station, a network element, etc. as described above in conjunction with one or more of the foregoing figures may be configured to operate according to one or more of the examples set forth herein.

[0117] Unless otherwise expressly stated, any of the above embodiments may be combined with any other embodiment (or combination of embodiments). The foregoing description of one or more specific implementations provides illustration and description, but is not intended to be exhaustive or to limit the scope of the embodiments to the precise form disclosed. In view of the above teachings, modifications and variations are possible or can be obtained from the practice of various embodiments.

[0118] Embodiments and implementations of the systems and methods described herein may include various operations that may be embodied in machine executable instructions to be executed by a computer system. A computer system may include one or more general or special purpose computers (or other electronic devices). A computer system may include hardware components that include specific logic components for performing operations; or may include a combination of hardware, software, and / or firmware.

[0119] It should be appreciated that the systems described herein include descriptions of specific embodiments. These embodiments may be combined into a single system, partially combined into other systems, separated into multiple systems, or otherwise divided or combined. In addition, it is contemplated that parameters, attributes, aspects, etc. of another embodiment may be used in one embodiment. For clarity, these parameters, attributes, aspects, etc. are described only in one or more embodiments, and it should be appreciated that these parameters, attributes, aspects, etc. may be combined with or substituted for parameters, attributes, aspects, etc. of another embodiment unless expressly stated herein.

[0120] It is understood that the use of personally identifiable information should be subject to privacy policies and practices that are generally recognized to meet or exceed industry or government requirements for maintaining user privacy. Specifically, personally identifiable information data should be managed and processed to minimize the risk of unintentional or unauthorized access or use, and the nature of the authorized use should be clearly stated to users.

[0121] Although the foregoing has been described in considerable detail for the sake of clarity, it will be apparent that certain changes and modifications may be made without departing from the principles of the invention. It should be noted that there are many alternative ways to implement both the processes and the apparatus described herein. Therefore, the embodiments of the present invention should be regarded as illustrative rather than restrictive, and the specification is not limited to the details given herein, but may be modified within the scope and equivalents of the appended claims.

Claims

1. A method for a user equipment (UE) to provide machine learning (ML) based channel state information (CSI) feedback to a wireless network, the method comprising: for a neural network (NN) model type and a method for deriving CSI for multi-rank CSI feedback, reporting one or more feedback sizes to the wireless network in a UE capabilities message; For the NN model type and the method, processing a configuration of a plurality of NN models associated with the one or more feedback sizes from the wireless network; processing an overhead allocation for the one or more feedback sizes; as well as The multi-rank CSI feedback is generated using a different NN model among the multiple NN models for each spatial layer or spatial layer group or rank based on the overhead allocation.

2. The method of claim 1, wherein processing the overhead allocation comprises receiving an indication of the overhead allocation from the wireless network.

3. The method of claim 1 , wherein processing the overhead allocation comprises: determining, at the UE, the overhead allocation based on the plurality of NN models configured by the wireless network for the one or more feedback sizes; as well as An indication of the overhead allocation is transmitted from the UE to the wireless network.

4. The method of claim 1, wherein the one or more feedback sizes are selected from the group consisting of at least one of: a total feedback size, a per-spatial layer feedback size, a per-spatial layer group feedback size, a per-rank feedback size, and a per-NN model feedback size.

5. The method of claim 1, wherein the method comprises using a spatial layer group common NN model, and wherein the overhead allocation comprises an indication of: Different feedback sizes for different spatial layer groups; or A total feedback size for each of the different spatial layer groups and a ratio of the allocated feedback size to the total feedback size.

6. The method of claim 5, wherein using the different NN models to generate the multi-rank CSI feedback comprises using at least: a first NN model associated with a first feedback size for a first spatial layer group; and A second NN model is associated with a second feedback size for a second spatial layer group.

7. The method of claim 1, wherein the method comprises using a rank-specific NN model, and wherein the overhead allocation comprises an indication of: dividing the feedback bits among different spatial layers in the indicated rank; or A total number of feedback bits for each of the different spatial layers in the indicated rank and a ratio of the allocated bits to the total number of feedback bits.

8. The method of claim 7, wherein the partitioning comprises unequal numbers of the feedback bits between the different spatial layers in the indicated rank.

9. The method of claim 7, wherein using the different NN models to generate the multi-rank CSI feedback comprises using two or more of: a first NN model associated with a first number of feedback bits for a first rank; a second NN model associated with a second number of feedback bits divided between the different spatial layers for a second rank; a third NN model associated with a third number of feedback bits divided between the different spatial layers for a third rank; and A fourth NN model is associated with a fourth number of feedback bits divided among the different spatial layers for a fourth rank.

10. The method of claim 1, wherein the method comprises using a spatial layer specific NN model, and wherein the overhead allocation comprises an indication of: The number of feedback bits used for different spatial layers; or a total number of feedback bits for each of the different spatial layers and a ratio of the allocated bits to the total number of feedback bits.

11. The method of claim 10, wherein the indication indicates that the number of feedback bits between the different spatial layers is an unequal number.

12. The method of claim 10, wherein using the different NN models to generate the multi-rank CSI feedback comprises using two or more of: a first NN model associated with a first number of feedback bits for a first spatial layer; a second NN model associated with a second number of feedback bits for a second spatial layer; a third NN model associated with a third number of feedback bits for a third spatial layer; and A fourth NN model is associated with a fourth number of feedback bits for a fourth spatial layer.

13. The method of claim 1, wherein the NN model types are optimized separately for UE hardware and base station hardware, and wherein the multiple NN models each include an NN model pair corresponding to an encoder at the UE and a decoder at the base station.

14. The method according to claim 1, further comprising: determining a total feedback overhead for codebook configuration; comparing the total feedback overhead to a threshold to determine a triggering event; as well as In response to the triggering event, the overhead allocation is processed and the multi-rank CSI feedback is generated using the different NNs based on the overhead allocation.

15. A method for a base station to configure machine learning (ML) based channel state information (CSI) feedback in a wireless network, the method comprising: For a neural network (NN) model type and a method for deriving CSI for multi-rank CSI feedback, one or more feedback sizes are received from a user equipment (UE) in a UE capabilities message; For the NN model type and the method, configuring a plurality of NN models associated with the one or more feedback sizes for the UE; as well as Overhead allocation for the one or more feedback sizes is processed.

16. The method of claim 15, wherein processing the overhead allocation comprises receiving, at the base station, an indication of the overhead allocation from the UE.

17. The method of claim 15, wherein processing the overhead allocation comprises: determining, at the base station, the overhead allocation based on the plurality of NN models configured by the wireless network for the one or more feedback sizes; as well as An indication of the overhead allocation is transmitted from the base station to the UE.

18. The method of claim 15, wherein the one or more feedback sizes are selected from the group consisting of at least one of: a total feedback size, a per-spatial layer feedback size, a per-spatial layer group feedback size, a per-rank feedback size, and a per-NN model feedback size.

19. The method of claim 15, wherein the method comprises using a spatial layer group common NN model, and wherein the overhead allocation comprises an indication of: Different feedback sizes for different spatial layer groups; or A total feedback size for each of the different spatial layer groups and a ratio of the allocated feedback size to the total feedback size.

20. The method of claim 15, wherein the method comprises using a rank-specific NN model, and wherein the overhead allocation comprises an indication of: dividing the feedback bits among different spatial layers in the indicated rank; or A total number of feedback bits for each of the different spatial layers in the indicated rank and a ratio of the allocated bits to the total number of feedback bits.

21. The method of claim 20, wherein the partitioning comprises unequal numbers of the feedback bits between the different spatial layers in the indicated rank.

22. The method of claim 15, wherein the method comprises using a spatial layer specific NN model, and wherein the overhead allocation comprises an indication of: The number of feedback bits used for different spatial layers; or a total number of feedback bits for each of the different spatial layers and a ratio of the allocated bits to the total number of feedback bits.

23. The method of claim 22, wherein the indication indicates that the number of feedback bits between the different spatial layers is an unequal number.

24. The method of claim 15, wherein the NN model types are optimized separately for UE hardware and base station hardware, and wherein the multiple NN models each comprise an NN model pair corresponding to an encoder at the UE and a decoder at the base station.

25. The method according to claim 15, further comprising: determining a total feedback overhead for codebook configuration; comparing the total feedback overhead to a threshold to determine a triggering event; as well as In response to the triggering event, the overhead allocation is processed.

26. An apparatus comprising means for performing the method according to any one of claims 1 to 25.

27. A computer-readable medium comprising instructions, which, when executed by one or more processors of an electronic device, cause the electronic device to perform the method according to any one of claims 1 to 25.

28. An apparatus comprising logic components, modules or circuits for performing the method according to any one of claims 1 to 25.

Images