A method for providing channel state information feedback using channel prediction, and a respective apparatus and a computer program product
By computing and reporting Channel Predictor Functions using uncompressed CSI, the method addresses the challenge of inaccurate CSI acquisition in 5G systems, enhancing precision and reducing overhead, thereby improving MIMO performance in 5G networks.
Patent Information
- Application Number
- PCT/EP2024/063469
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-17
- Filing Date
- 2024-05-16
- Publication Date
- 2024-11-21
AI Technical Summary
Current 5G wireless communication systems face challenges in achieving accurate Channel State Information (CSI) acquisition due to compression errors in CSI feedback, leading to inaccurate precoding and increased overhead, particularly in MIMO systems using mmWave frequencies.
A method where the User Equipment (UE) computes and reports Channel Predictor Functions using uncompressed CSI, reducing quantization errors and overhead by training a neural network with uncompressed data and verifying predictions through error or cosine similarity, allowing for precise CSI acquisition without the need for a neural network at the base station.
This approach enhances the precision of CSI acquisition, reduces overhead, and eliminates the need for synchronized machine learning algorithms, improving MIMO precoding performance and data throughput in 5G networks.
Smart Images

Figure EP2024063469_21112024_PF_FP_ABST
Abstract
Description
A METHOD FOR PROVIDING CHANNEL STATE INFORMATION FEEDBACK USING CHANNEL PREDICTION, AND A RESPECTIVE APPARATUS AND A COMPUTER PROGRAM PRODUCTTechnical field
[0001] The present invention relates generally to cellular radio networks and their implementation. Concerning different generations of technology, the present invention may relate to any of the following: 2G, 3G, 4G, 5G or 6G access networks.Background
[0002] Today’s and future wireless communication systems, such as Long-Term Evolution (LTE) or fifth-generation (5G), also referred to as New Radio (NR), have been envisaged to use multiple-input & multiple-output (MIMO) multi-antenna transmission techniques. Constantly increasing requirements for high throughput motivates the wireless communication systems, such as 5G, to use the mmWave (millimeter wave) frequencies due to available high bandwidth.
[0003] However, the usage of mmWave frequencies poses new challenges on MIMO performance.
[0004] Channel state information (CSI) acquisition is of predominant importance for MIMO precoding and beamforming features of NR. Currently, in 5G telecommunication networks, two different strategies have been considered for CSI report: type I CSI and type II CSI feedback; available in 3rdGeneration Partnership Project (3GPP) technical specifications. Such techniques are based on a set of codebooks known at both ends of communication (gNodeB i.e. the base station (gNB) and the user equipment (UE)), and they are differently combined to report the downlink (DL) channel geometrical features.
[0005] More precisely, the channel is estimated starting from pilots, which are transmitted by the gNB, be them periodic, semi-persistent or aperiodic. Once estimated, the channel is projected onto the codebooks through different levels of precision, e.g., type-ll CSI adopts a linear combination of the vectors in the codebook to represent the channel. Both methods (i.e. both type I and type II)include a strong compression of the CSI estimated at the UE side to reduce the over-the-air (OTA) overhead. Such compression strongly reduces the precision of the CSI, which impacts the performance of MIMO precoding. This is a drawback of the present technology.
[0006] Furthermore, CSI has been proven to possess a certain level of timeconsistency, i.e. , the CSI at time n is somehow correlated with the CSI at time n-1 , n-2, and so on. This effect is particularly true for non-mobile UEs and for UEs whose mobility pattern can be predicted (e.g., a UE on a train, on a freeway, and so on).
[0007] The problem of the prior art can be described as follows.
[0008] The compression imposed at the CSI feedback from a UE to the gNB impacts the precision of the CSI and consequently the performance of the precoding. In the literature, different strategies to predict CSI realization starting from past realization exist. Moreover, each channel estimation and report is in a sense “agnostic” and independent from previous realizations and reports.
[0009] In CSI feedback, accurate channel predictor function (CPF) is needed at both ends of communication system. However, CPF used in the literature is based on a compressed version of CSI. Thus, resultant CPF can be prone to compression errors. This will further lead to inaccurate channel prediction, causing inaccurate CSI acquisition, which is a major drawback.
[0010] In prior art, training a machine learning -based (ML-based) channel predictor at the gNB and UE is bounded to use the compressed version of the estimated CSI due to requiring identical predictions for CSI feedback. The reporting of training data from UE to gNB can cause a huge over-the-air (OTA) overhead. Further, the training data at the gNB and the UE is compressed. Hence, prediction will not be accurate. Furthermore, synchronization of two prediction algorithms for training is indispensable.
[0011] In another piece of prior art, to avoid ML at the UE, ML is used only at the gNB but such an ML algorithm has to rely on a compressed CSI for training. Though, consideration of ML at the UE has also been discussed, which is again with the assumption of using a compressed CSI.
[0012] Hence, in the prior art, there are three issues. As a first issue, one UE has to report training data, which itself requires huge OTA overhead. Secondly, training and prediction is based on a compressed CSI. Besides, prediction of compressedCSI is highly complex. Thirdly, synchronization of ML channel predictors for training a prediction model is required.
[0013] Unabated growth of cellular users and their immense data rate demands require novel advancements in the existing cellular infrastructure. To meet the requirements of cellular users, many transmission schemes such as coding, adaptive modulation, and channel-aware scheduling, etc., are bounded to have an accurate CSI at the transmitter side to achieve a significant gain, in the downlink scenario. Therefore, an accurate CSI is of pivotal importance in 5G and beyond network technologies.
[0014] Within a short span of time, ML emerged as a paradigm shift for future cellular networks. For instance, prediction of CSI has been considered by many researchers both from academia and industry. To this end, channel prediction is mainly handled by using Kalman filter (KF)-based approaches in such pieces of prior art. For example, there are publications which investigate the performance of KF in single-input single-output (SISO) and MIMO mobile radio channels, respectively. Specifically, an auto regressive (AR) model is used to model the dynamics of the channel transfer function (CTF) and the coefficients of the AR process are calculated using the temporal and spectral correlation under the assumption of Jakes’ fading model and one-sided exponential power delay profile (PDP), respectively. On the other hand, artificial intelligence ML is also considered by many authors for predicting the mobile radio channel. For instance, a piece of prior art has exploited a recurrent neural network (RNN), a class of ML algorithms, for massive MIMO channel prediction. It has shown that an RNN can predict massive MIMO channel effectively, and it comes up with a benefit of multi-step ahead prediction in comparison of KF-based approaches.
[0015] There is a couple of prior art documents which have exploited twin channel predictors and one-sided CPF computation, respectively. However, channel predictors implementation and computation of CPF is based on a compressed version of CSI, which can reduce precision of the acquired CSI. This drawback thus remains in this concept.
[0016] Hence, the problem to be solved is to have an accurate CPF to exploit for the channel prediction, consequently for the CSI acquisition.Summary
[0017] The present invention is about proposing a light-weighted solution including only one neural network (NN) that can aid CSI acquisition in terms of precision and overhead.
[0018] A possible solution for the above-mentioned problem is to exploit UE’s capabilities, i.e., computing CPF only at the UE by using an uncompressed estimated CSI. A novel way is proposed to exploit a neural network (NN) to either reduce the quantization error (hence, improving MIMO precoding performance) while keeping the OTA occupation constant, or to further reduce the overhead, without the need for implementing a NN at the gNB’s end.
[0019] According to a first aspect of the present invention, there is introduced a method for providing Channel State Information, CSI, feedback using channel prediction at a User Equipment, UE, side (102). It is characterized in that the method comprises the steps of: using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receiving the CSI-RS directly from the gNB (101 ) without request;verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and reporting an updated CPF (116) to the gNB (101 ).
[0020] According to an embodiment of the first aspect of the present invention, the method further comprises a step of: using CSI-RS sent by the gNB (101 ) for channel estimation in case the UE (102) does not have the training data available (304).
[0021] According to an embodiment of the first aspect of the present invention, the method further comprises steps of: requesting also a Precoding Matrix Indicator (PMI) of the latest estimated CSI from the gNB (101 ) in connection with requesting the CSI-RS (308), or receiving the PMI directly from the gNB (101 ) without request; and comparing (309) a PMI of the latest estimated CSI at the UE (102) with the PMI reported by the gNB (101 ).
[0022] According to an embodiment of the first aspect of the present invention, the method further comprises a step of: if the error or the CS or the PMI comparison fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ).
[0023] According to an embodiment of the first aspect of the present invention, the method further comprises steps of: if none of the error or the CS or the PMI comparison fulfils a certain criterion (310), the CPF is considered non-valid, and sending a non-acknowledgement message (NACK) (114, 313) to the gNB (101 ); andcomputing a new CPF (115) via steps (105, 106, 107; 303, 305).
[0024] According to an embodiment of the first aspect of the present invention, the method further comprises steps of: if none of the error or the CS fulfils a certain criterion (310), the CPF is considered non-valid, and sending a non-acknowledgement message (NACK) (114, 313) to the gNB (101 ); and computing a new CPF (115) via steps (105, 106, 107; 303, 305).
[0025] According to an embodiment of the first aspect of the present invention, the method further comprises a step of: predicting a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information (110).
[0026] According to an embodiment of the first aspect of the present invention, the method further comprises a step of: applying a Recurrent Neural Network, RNN, based Channel Predictor Function at the UE (102).
[0027] According to an embodiment of the first aspect of the present invention, the method further comprises a step of: instead of training a Neural Network, selecting an available function from a list using an autoregressive, AR, model.
[0028] According to an embodiment of the first aspect of the present invention, the method further comprises a step of: using an identical CPF in both the gNB (101 ) and the UE (102).
[0029] According to an embodiment of the first aspect of the present invention, the method further comprises steps of:at the start of the process, requesting the gNB (101 ) to send CSI-RS (304, 103) if the estimated CSI is not available at the UE (102); and receiving an ACK and the requested CSI-RS (104) from the gNB (101 ).
[0030] According to an embodiment of the first aspect of the present invention, the method further comprises a step of: in connection with step (116), reporting also the latest compressed CSI by the UE (102) to the gNB (101 ).
[0031] According to a second aspect of the present invention, there is introduced a method for providing Channel State Information, CSI, feedback using channel prediction at a base station, gNB, side (101 ). It is characterized in that the method comprises the steps of: receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); sending (104) an acknowledgement message (ACK) to the UE (102); receiving (108) a Channel Predictor Function, CPF, from the UE (102); receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receiving (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); and receiving (116, 311) an updated CPF from the UE (102).
[0032] According to an embodiment of the second aspect of the present invention, the method further comprises steps of:receiving (103) a request of Channel State Information Reference Signals, CSI-RS, from the UE (102) in case that information is not available at the UE (102); and sending (104) the CSI-RS to the UE (102).
[0033] According to an embodiment of the second aspect of the present invention, the method further comprises steps of: receiving (112, 308) also a request of a Precoding Matrix Indicator (PMI) of the latest estimated CSI from the UE (102) in connection with the request of the CSI- RS (308); and sending (308) the PMI from the gNB (101 ) to the UE (102) after the request; or sending (308) the PMI directly from the gNB (101 ) to the UE (102) without any received request.
[0034] According to an embodiment of the second aspect of the present invention, the method further comprises a step of: after a received NACK in step (313), waiting for the UE (102) to compute a new CPF (115) via steps (105, 106, 107; 303, 305).
[0035] According to a third aspect of the present invention, there is introduced a User Equipment, UE, (102) for providing Channel State Information, CSI, feedback using channel prediction. It is characterized in that the UE (102) comprises: means for using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); means for computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); means for reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 );means for using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); means for requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or means for receiving the CSI-RS directly from the gNB (101 ) without request; means for verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, the UE (102) further comprises means for sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and means for reporting an updated CPF (116) to the gNB (101 ).
[0036] According to a fourth aspect of the present invention, there is introduced a base station, gNB, (101 ) for providing Channel State Information, CSI, feedback using channel prediction. It is characterized in that the gNB (101 ) comprises: means for receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); means for sending (104) an acknowledgement message (ACK) to the UE (102); means for receiving (108) a Channel Predictor Function, CPF, from the UE (102); means for receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); means for predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; means for sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102);means for receiving (114) an acknowledgement message (ACK) (311 ) or a non-acknowledgement message (NACK) (313) from the UE (102); and means for receiving (116, 311 ) an updated CPF from the UE (102).
[0037] According to a fifth aspect of the present invention, there is introduced a User Equipment, UE, (102) for providing Channel State Information, CSI, feedback using channel prediction. It is characterized in that the UE (102) comprises: at least one processor; and at least one memory, storing instructions that, when executed by the at least one processor, cause the UE (102) at least to: use uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); compute a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); report the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); use the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); request Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receive the CSI-RS directly from the gNB (101 ) without request; verify the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, send an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and report an updated CPF (116) to the gNB (101 ).
[0038] According to a sixth aspect of the present invention, there is introduced a base station, gNB, (101) for providing Channel State Information, CSI, feedback using channel prediction. It is characterized in that the gNB (101 ) comprises: at least one processor; and at least one memory, storing instructions that, when executed by the at least one processor, cause the gNB (101 ) at least to: receive (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); send (104) an acknowledgement message (ACK) to the UE (102); receive (108) a Channel Predictor Function, CPF, from the UE (102); receive (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predict (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; send (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receive (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); and receive (116, 311) an updated CPF from the UE (102).
[0039] According to a seventh aspect of the present invention, there is introduced a computer program comprising instructions which, when executed by a User Equipment, UE, (102), cause the UE to perform at least the following: using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305);reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receiving the CSI-RS directly from the gNB (101 ) without request; verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and reporting an updated CPF (116) to the gNB (101 ).
[0040] According to an eighth aspect of the present invention, there is introduced a computer program comprising instructions which, when executed by a base station, gNB, (101 ), cause the gNB to perform at least the following: receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); sending (104) an acknowledgement message (ACK) to the UE (102); receiving (108) a Channel Predictor Function, CPF, from the UE (102); receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receiving (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); andreceiving (116, 311 ) an updated CPF from the UE (102).
[0041] According to a ninth aspect of the present invention, there is introduced a non-transitory computer readable medium comprising program instructions that, when executed by a User Equipment, UE (102), cause the UE to perform at least the following: using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receiving the CSI-RS directly from the gNB (101 ) without request; verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and reporting an updated CPF (116) to the gNB (101 ).
[0042] According to a tenth aspect of the present invention, there is introduced a non-transitory computer readable medium comprising program instructions that, when executed by a base station, gNB (101 ), cause the gNB to perform at least the following:receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); sending (104) an acknowledgement message (ACK) to the UE (102); receiving (108) a Channel Predictor Function, CPF, from the UE (102); receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receiving (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); and receiving (116, 311 ) an updated CPF from the UE (102).Brief description of the drawings
[0043] FIG. 1 illustrates an example of a process chart involving a signaling diagram, according to an embodiment of the present invention,
[0044] FIG. 2 illustrates an example of an implementation of a Channel Predictor Function (CPF) at the UE, according to an embodiment of the present invention,
[0045] FIG. 3 illustrates a flow chart of the implementation at the UE and at the gNB, according to an embodiment of the present invention,
[0046] FIG. 4 illustrates an exemplary radio network scenario in accordance with at least some embodiments of the present invention, and
[0047] FIG. 5 illustrates an example apparatus capable of supporting at least some embodiments of the present invention.Detailed description
[0048] As specified in the previous sections (in the summary and the problem statements in the background), the present invention concerns the utilization of aCPF, which is computed only at the UE, to help the CSI reporting function in 5G networks. Such a CPF will be more precise for CSI acquisition as it is computed based on uncompressed CSI. Next, it is discussed, how the CPF will be more accurate in the end. UE is the only source that can have uncompressed version of the estimated CSI; therefore, a more accurate CPF may be calculated by exploiting an uncompressed CSI and with the help of a NN at UE. Later, the CPF along with the latest compressed CSI or uncompressed CSI (to assist channel prediction at the gNB) may be reported to the gNB.
[0049] The present invention focuses on using ML at the UE only, in its various embodiments. Specifically, the UE will train the neural network (NN) using an uncompressed estimated CSI. Thus, inputs and outputs of the NN are different from prior art.
[0050] Once the NN at UE is trained, it computes a CPF and reports the CPF along with the latest compressed / uncompressed CSI to the gNB to start the process of CSI acquisition. Both entities can exploit the CPF to make a prediction on compressed / uncompressed data.
[0051] After a certain amount of time, the UE can request the gNB for Channel State Information Reference Signals (CSI-RS) and Precoding Matrix Indicator (PMI; optional) to estimate the CSI and verify the CPF, respectively, through, e.g., Mean- Squared-Error (MSE), cosine similarity (CS), or PMI’s comparisons. In consequence, the updated CPF is reported if needed, and otherwise, an acknowledgement (ACK) is sent. In addition to ACK, the UE can send the compressed update function, i.e. , difference between prediction and estimation. The evaluation of the update function can be dependent on, e.g., a threshold value. For example, if the error in the update function is less than a threshold value, then the UE reports ACK and the update function. Reporting the update function can help the gNB to generate a better prediction for future values. Otherwise, the UE reports NACK and updated CPF. The threshold value can be the error in the conventional approach, i.e., difference between current estimated CSI and its compressed version. If the error produced by the conventional approach goes below the present invention, then the UE feedbacks using the conventional mechanism until a new CPF is computed, and otherwise it follows the present invention.
[0052] The present invention brings the advantage of eliminating overhead cost, required for reporting of training data to the gNB. Besides, computation and reporting of the CPF based on an uncompressed CSI can help a NN to train itself easily and enhancing the precision of the CSI. These form a further advantage of the present invention.
[0053] The basic idea of the present invention is to have an accurate and same CPF at both ends of communications (gNB and UE). Such CPF is used at a time (t- 1 ) to predict the channel realization at time t (Ht). The gNB is then transmitting pilots to allow the UE to estimate the CSI, Ht. When the UE estimates the CSI, instead of feeding back the value, it will feedback the output of a function g Ht,The gNB will use this value g(... ) and the predicted channel Ht-rto compute Ht. And for the future prediction, the gNB and the UE can use Htto predict the channel. In case of not receiving g(... ), the gNB and the UE use predicted value for future predictions. Though the UE has Ht, it uses predicted CSI for next prediction so that predictions can be identical on both sides, i.e., gNB and UE.
[0054] For simplicity, consider a gNB-UE link, in which the gNB transmits pilot symbols to the UE to estimate the CSI. The UE estimates the CSI using the classic approach for a given number of time steps, and uses such highly refined information to compute a CPF. This can be done either 1 ) training a NN, 2) selecting an available function from a list, or 3) using an auto-regressive model (ARM).
[0055] The UE will hence feedback the model and the latest com pressed / uncom pressed CSI to the gNB, where it will be used to predict the channel. As a result, the gNB and the UE will use identical CPF, and however, it will be more accurate as compared to the prior art, for CSI acquisition. As a bit theoretical example, if the prediction at both ends is perfect, then the UE does not need to feedback anything, and hence, overhead cost can be eliminated.
[0056] As mentioned above, implementation of a CPF is done at the UE with the help of uncompressed CSI. By exploiting CSI-RS, the UE can estimate the CSI and use it (without any need to compress) for the input and corresponding output of a NN, where the job of a NN is either to predict the CSI or to compute a CPF. In both scenarios, a CPF may be computed by using the predicted channel and past CSI realizations. Finally, the computed CPF and the latest com pressed / uncom pressedCSI at the UE can be reported to the gNB for CSI acquisition. How this is achieved is explained below in more detail.
[0057] A process chart involving a signaling diagram is now explained according to an embodiment of the present invention.
[0058] The respective signaling diagram 100 is also shown in FIG. 1, where the new signals required for the present invention are marked with a small circle. Base station gNB 101 and User Equipment UE 102 are shown in the left-hand side and the right-hand side of the chart, respectively. The requests, messages, acknowledgments, and in other words, the communication, is implemented between these two network entities.
[0059] At a first step of the presented signaling method in an embodiment of the invention, the UE informs 103 the use of computing / implementing CPF at its end. Therefore, the UE requests the gNB to send CSI-RS if training data (estimated CSI) is not available at the UE.
[0060] At a second step, the gNB sends 104 an acknowledgment message (ACK) for the request given in the first step. Also, the gNB sends CSI-RS if requested by the UE. In prior art, in case of receiving CSI-RS, the UE can generate some CSI samples by using a channel estimation algorithm. In this way, the prior art makes channel estimation at the UE 105.
[0061] At a third step, and back to the described embodiment of the present invention, once the CSI is estimated at the UE, the UE divides training and testing samples, and inputs uncompressed CSI training samples and corresponding uncompressed CSI labels to the neural network (NN) for training. In other words, the neural network is trained 106 for channel prediction using uncompressed CSI. Thus, the UE computes 107 a CPF by exploiting the predicted CSI and available uncompressed CSI. This may be embodied by applying the neural network or by using statistical methods, such as e.g., Kalman Filter, in various embodiments of the present invention.
[0062] At a fourth step, the UE reports 108 the CPF to the gNB. In different embodiments, reporting of the CPF can be done in different ways, where the objective can be to report smallest model (requiring minimum overhead, for example). For instance, if underlying model is evolving through an auto-regressive process of order one, AR (1 ), then the CPF can be reported as:
[0064] where CpEis the CPF.
[0065] At a fifth step, both the gNB and the UE agree to use 109 the CPF for CSI feedback. The UE reports the latest uncompressed or compressed instance of the CSI for twin channel prediction, in an embodiment. Then the CSI acquisition process begins. In prior art, both the gNB 110 and the UE 111 have predicted the CSI by exploiting the CPF and the past compressed CSI; however noting that the CPF of the present invention is different than what is used in the prior art. In an embodiment of the present invention, both the gNB and the UE predict the CSI by exploiting the CPF, and the past compressed CSI or the past uncompressed CSI.
[0066] At a sixth step, at a certain time instant ‘t’, either the UE can request the gNB to send 112 CSI-RS and optionally also Precoding Matrix Indicator (PMI) of the last predicted CSI, or the gNB sends the same (CSI-RS and PMI) by itself (i.e., without any request from the UE. CSI-RS and PMI can be sent at periodic intervals). The purpose is to check the validity of the CPF. The PMI is optional because the UE can also do verification without it. As a result, the UE is able to verify 113 the validity of the CPF locally, e.g., by computing difference between prediction and estimation, or by comparing the PMI of the gNB with the locally calculated one at the UE by exploiting CSI-RS.
[0067] Then, at a seventh step, the UE reports 114 ACK and the update function if needed as described before (see par. 0028), if the CPF is correct, and the same is thus indicated. On the other hand, the UE reports 114 NACK, if the CPF has expired, and the same is thus indicated. These reports are sent to the gNB in this step.
[0068] At an eighth step, if the response message was NACK, the UE will compute 1 15 a new CPF applying the steps 2a, 3 and 3a as numbered in FIG. 1 . These are steps 105-107 in FIG. 1 .
[0069] Then in the end, at a ninth step, the UE reports 116 an updated CPF and the latest com pressed / uncom pressed CSI to the gNB.
[0070] Next, some implementation details are discussed, in relation to respective embodiments of the present invention.
[0071] At first, implementation of the procedure in the UE is discussed, in an embodiment of the respective method.
[0072] There are three major implementations required in this embodiment, out of which one (channel estimation) is prior. Briefly, first implementation is channel estimation, second is channel prediction and CPF computation, and third is validation of CPF.
[0073] Briefly, in the first implementation, the UE exploits CSI-RS and estimates the channel using a dedicated channel estimation algorithm, where the respective channel estimation algorithm may be taken from methods already applied in the prior art, in the described embodiment of the invention.
[0074] Second implementation, i.e., channel prediction and CPF computation, exploits the estimated CSI (uncompressed). This is substantially different, than what is done in the respective prior art. The estimated CSI is divided into training, validation, and testing data. Correspondingly, labels are created for training the neural network (NN). At the input of a NN, uncompressed CSI is injected and at the output side corresponding uncompressed labels are provided for the purpose of channel prediction. Once the channel is predicted, CPF is computed by using predicted CSI and past estimated (uncompressed) CSI realizations. Another possibility of implementation is to input uncompressed CSI and output a well refined CPF.
[0075] For illustration of this embodiment, see FIG. 2, which shows implementation of CPF at the UE as a simplified logic chart 200. The uncompressed Channel State Information (CSI) 201 is input into the computation logic. The steps described in the following are shown as steps 202, 203 and 204. As a result of the computation logic, an output is the predicted CSI or the CPF 205.
[0076] An example of an implemented CPF may be performed with the following steps:
[0077] 1 ) Collect a set of predicted channels, and stack them into a matrix:P = [H^E, ... , H^E] (2)
[0078] 2) Collect the corresponding set of estimated channels:P = [H^i . HE(3)
[0079] Since p = F1P, where Fjis a CPF obtained using the present invention, it results that:Fi = P pinv(P) (4)
[0080] where pinv(-) represents the right pseudo inverse. The matrix F±is fed back to the gNB along with the latest com pressed / uncom pressed CSI, and will be used henceforth to predict the channel at both the gNB and the UE, using the following equation (H
[0081] The third implementation at the UE is the validation of the CPF. At a time instant “t”, the UE gets CSI-RS and PMI (optional) of the predicted CSI from the gNB and correspondingly validates the CPF. For example, the former (CSI-RS) is used to compute the estimated CSI using the first implementation. Once the estimation is done, the UE may perform one of the following actions. a. The UE computes an update function, e.g., a difference between the prediction and the estimation of the channel realization. If such a difference is smaller than a threshold (e.g., the quantization error of the conventional approach), then the UE feedbacks the update function:Ut= H9E- Ht'JE, (6) where to reduce OTA, update is compressed, however, overhead will be small compared to reporting only the estimated CSI due to reporting difference. This is due to the fact that the amplitude of the entries in update function can be much smaller in comparison to the estimated channel, thanks to prediction. b. The UE computes cosine similarity of the predicted and the estimated CSI, and if similarity is high, the UE feedbacks ACK (= acknowledgment message) and update function if needed. c. The UE computes PMI by mapping the estimated CSI to the codebook and compares the resultant PMI with the one sent by the gNB. If both are the same, ACK is then sent and update function if needed. d. If none of the above is satisfied, then the UE reports NACK (= nonacknowledgment message) and repeats steps 2a to 3a (= 105-107) ofFIG. 1 and reports the updated CPF and the latest com pressed / uncom pressed CSI to the gNB.
[0082] Next, an implementation possibility at the gNB (= base station) is discussed in an embodiment of the present invention.
[0083] Thanks to the accurate CPF reported by the UE, the gNB has to do only one implementation, i.e., computing the predicted channel with the help of the reported CPF and the compressed / uncompressed CSI. This can be formulated mathematically as:
[0084] In FIG. 3, a flow chart 300 of the implementation at the gNB and the UE is provided.
[0085] At step 301 , the process starts concerning the generation of CPF at the UE side 301.
[0086] At step 302, the process checks whether there is training data available at the UE or not.
[0087] If not, the process goes to step 304, and the channel estimation is performed using CSI-RS sent by the gNB.
[0088] After this step, in step 303, the process uses estimated CSI for channel prediction. This same step is performed also in the case, if there indeed is training data available at the UE in step 302. Now we have a predicted CSI.
[0089] Next the process progresses into step 305, where the CPF is computed using the predicted CSI and past uncompressed CSI realizations. This may be embodied by applying the neural network or by using statistical methods, such as e.g., Kalman Filter, in various embodiments of the present invention (as mentioned earlier in connection with step 107 of Fig. 1 ).
[0090] After this, in step 306, the process reports the CPF and the latest compressed CSI to the gNB.
[0091] Thereafter, in step 307, the gNB and the UE use the CPF and the latest compressed CSI to predict the CSI.
[0092] Then, in step 308, after a certain amount of time, the gNB sends CSI-RS and the PMI. Alternatively, this sending step is performed when the UE so requests.
[0093] Thereafter, in step 309, the UE verifies the CPF. According to a first optional solution, this verification is performed through error (predicted vs. estimated) or through Cosine Similarity (CS). According to a second optional solution, this verification is performed by comparing the PMI of the latest estimated CSI at the UE and the PMI reported by the gNB.
[0094] Then, in step 310, some checks (= comparisons) are performed. If the error is q or 1NB== lEE, the process sends ACK (acknowledgment) to the gNB and feedback update function 311 , where the update function is the compressed version of the difference between the predicted CSI and the estimated CSI at the UE. After ACK, the process ends 312. If none of the conditions of step 310 are fulfilled, the process sends NACK (non-acknowledgment) to the gNB in step 313. After this step, the process goes back to step 302, and goes on from there.
[0095] As a clarification to the terms in the previous paragraph, A and q are predefined thresholds for the error and CS. INBand IEEare PMI of the gNB and the UE for the predicted and estimated channel at the gNB and the UE, respectively.
[0096] Next, a channel predictor example is presented. It forms an embodiment within the inventive concept.
[0097] An example of a channel predictor is a machine learning -based (ML) channel predictor. For instance, within the domain of ML, a Recurrent Neural Network (RNN) is one of the powerful methods of making time-series prediction. Therefore, it may be used as a channel predictor. RNN is different from a traditional neural network (TNN) in the aspect that its output is dependent on the current as well as the previous state, hence the name recurrent. On the other hand, TNN’s inputs are independent from each other, which means no relationship between them.
[0098] Therefore, an RNN-based channel predictor function is assumed at the UE. With the help of predicted channel using the RNN, the CPF is reported to the gNB. Mathematically, acquired CSI at the gNB with the help of the reported CPF and the compressed update is given as:
[0099] Proceeding now to the obtained advantages of the present invention and its various embodiments, the following is concluded.
[0100] Firstly, it eliminates the need of reporting training data to the gNB. Hence, training-related overhead can be eliminated.
[0101] Secondly, by exploiting uncompressed CSI, a more accurate version of the CPF can be reported to the gNB.
[0102] Thirdly, the verification of the CPF can be done locally at the UE, and correspondingly update can be sent if needed.
[0103] Fourthly, there is no need to synchronize the artificial intelligence I machine learning (Al I ML) algorithms for training, as the UE will do everything.
[0104] In general, CSI feedback overhead can be eliminated if prediction is perfect at the UE and alternatively the precoding performance can be enhanced.
[0105] In summary, the present invention can obtain an accurate CPF by exploiting UE’s capability, i.e., using uncompressed CSI. In addition, there is no need to report training data to the gNB; hence, overhead of training-related data is eliminated. Verification of the CPF can be done locally. Furthermore, there is no need to synchronize the channel prediction algorithms for training. With the help of an accurate CPF, CSI feedback performance can be enhanced.
[0106] By summarizing in yet another words, the present invention brings benefits of eliminating training overhead and it helps to train AI / ML algorithms using uncompressed data.
[0107] Thus, there are various tangible advantages, which distinguish the present invention from the respective methods in the prior art.
[0108] In an embodiment, the present invention may be applied in 5G gNodeB:s, i.e. the respective base stations in 5G systems. In an embodiment, the present invention may be applied in 5G UEs, i.e. the respective mobile terminals. In other embodiments, it is possible to apply the described procedure (i.e. method) and the described device (i.e. apparatus) with some other generation of the telecommunication technologies, such as with 2G, 3G, 4G or6G telecommunication systems, for instance.
[0109] The present invention may be implemented in various network elements and in various network configurations. For instance, the presented procedure may be applied in a base station of the applied network architecture. In an embodiment,the base station may be a BTS (a base station in 2G), a NodeB (a base station in 3G), an eNB (evolved NodeB; a base station in 4G), a gNB (Next Generation NodeB; a base station of the 5G system), a CU (a Centralized Unit part of the 5G gNB) or a DU (a Distributed Unit part of the 5G gNB), for instance. In an embodiment, the base station may be a 6G base station. In an embodiment, the presented procedure may be applied in a UE of the applied network architecture. In the following, there is discussion on a certain network scenario, and applied device elements within that network. It is meant to represent a certain embodiment on an environment where the method for providing Channel State Information, CSI, feedback using channel prediction at a User Equipment, UE, side 102 according to certain embodiments of the present invention may be implemented and realized.
[0110] Figure 4 illustrates an exemplary radio network scenario in accordance with at least some embodiments of the present invention. According to the example scenario of Figure 4, there may be a wireless communication network, which comprises User Equipment, UE 410, an access node, such as a Base Station, BS, 420, and core network element 430.
[0111] UE 410 may comprise, for example, a smartphone, a cellular phone, a Machine-to-Machine (M2M) node, Machine-Type Communications (MTC) node, an Internet of Things (loT) node, a car, a car telemetry unit, a laptop computer, a tablet computer or another kind of suitable UE or mobile station. Generally, UE refers to any end device that may be capable of wireless communication. It can be either a mobile device or a stationary device. By way of example rather than limitation, a UE may also be referred to as a communication device, a terminal device, a Subscriber Station (SS), a Portable Subscriber Station, a Mobile Station (MS), or an Access Terminal (AT). In the example system of Figure 4, UE 410 may communicate wirelessly with BS 420, or with a cell of BS 420, via air interface 415. In some example embodiments, BS 420 may be considered as a serving BS, for UE 410. UE 410 may also communicate simultaneously with more than one BS 420 and / or more than one cell of BS 420.
[0112] BS 420 may be connected, directly or via at least one intermediate node, with core network 430 via interface 425. Core network 430 may be, in turn, coupled via interface 435 with another network (not shown in Figure 4), via whichever connectivity to further networks may be obtained, for example via a worldwideinterconnection network. BS 420 may be connected with one or multiple other BS as well via an inter-base station interface (not shown in Figure 4).
[0113] UE 410 may be connected to BS 420 via air interface 415. Air interface 415 between UE 410 and BS 420 may be configured in accordance with a Radio Access Technology, RAT, which UE 410 and BS 420 are configured to support. Examples of cellular RATs include Long Term Evolution (LTE), New Radio (NR), which may also be known as fifth generation (5G) radio access technology. For example, in the context of LTE, BS 420 may be referred to as an eNB while in the context of NR, BS 420 may be referred to as a gNB. In any case, example embodiments are not restricted to any particular radio technology. Instead, example embodiments may be exploited in any wireless communication network (which may be a cellular or a non-cellular technology) operating in accordance to 3GPP standard, IEEE standard (such as for example IEEE 802.11 based local area networks), or it can be some other radio technology, wherein it is desirable to achieve an improved transmission and / or reception performance between an access node such as a BS, and a UE, and in particular, an improved MIMO performance.
[0114] Ever increasing data throughput requirements in wireless communication networks requires usage of wide frequency spectrum. Thus, for example, frequency bands for 5G (also referred to as NR) are currently separated into different frequency ranges. Frequency Range 1 (FR1 ) includes sub-6 GHz frequency bands, some of which are frequency bands traditionally used by previous standards, but the range has been extended to cover potential new spectrum offerings up to 7125 MHz. Another range is Frequency Range 2 (FR2) that at the moment includes frequency bands from 24.25 GHz to 52.6 GHz. Frequencies of this range and above are referred to as mmWave frequencies. MmWave frequency ranges are attractive because of higher available bandwidth than frequency bands in the FR1 , which helps to offer data rates that satisfy 5G demands.
[0115] The described embodiments can be particularly beneficial for mmWave frequencies, including FR2 ranges, but can be applied equally to FR1 or any other frequencies. As said earlier, although applicable to any wireless networks, the 5G is mainly focused in the discussed examples, for the sake of simplicity. 5G has been envisaged to use more base stations or nodes than the current networkdeployments of LTE (a so-called small cell concept), including macro sites operating in co-operation with smaller local area access nodes and perhaps also employing a variety of radio technologies for better coverage and enhanced data rates. 5G will likely be comprised of more than one radio access technology I radio access network (RAT / RAN), each optimized for certain use cases and / or spectrum. 5G mobile communications may have a wider range of use cases and related applications including video streaming, augmented reality, different ways of data sharing and various forms of machine type applications, including vehicular safety, different sensors and real-time control.
[0116] MIMO is one of the key enabling techniques for 5G wireless technology. The underlying principle of MIMO is to use multiple transmit and receive antennas to increase throughput and / or reliability of data transmission. Increased throughput can be achieved by transmitting / receiving different data streams over multiple antennas, while increased reliability can be achieved by using multiple antennas for transmitting / receiving multiple versions of the same data.
[0117] Beamforming antenna arrays play an important role in 5G implementations. Although offering high bandwidth, mmWave frequencies have higher propagation losses that greatly vary depending on the environment. Smaller wavelength at higher carrier frequencies allows smaller antenna element sizes which gives an opportunity to place one or more (for example, two, three or more) relatively large antenna arrays at a UE. This in turn leads to various challenges to maintain the expected performance.
[0118] Considering as a non-limiting example 2x2 MIMO, downlink (DL) MIMO performance (for example, in mmWave frequency ranges, such as FR2) may be achieved by using polarization split (co-polar and cross-polar) of a dual feed antenna array at a base station (referred to also as a gNB) and / or at a UE, where each polarization corresponds to one MIMO branch. The reasoning behind this approach is to achieve high and similar antenna gain performance in both MIMO channels, while maintaining a compact spatial antenna design.
[0119] When designing dual-polarized antenna arrays, it is important to achieve high Cross Polarization Discrimination (XPD). XPD may be defined as a ratio of the co-polar component of the specified polarization and the orthogonal cross-polar component over a sector or beamwidth angle.
[0120] De-correlation at an antenna array may be obtained by ensuring that each antenna feed corresponds to a single polarization and that the resulting dual feed polarizations are designed to be orthogonal. This way, an antenna array with high XPD at the feed points may be designed. This approach will ensure full utilization of two MIMO channels for Line of Sight (LoS) and / or Non-Line of Sight (NLoS) operation provided that the maximum gain direction and the orientation of the orthogonal polarizations are aligned between antenna arrays at the UE and at the gNB.
[0121] In addition, high antenna gain requirements for mmWave (e.g. FR2) frequencies will reduce its radiation beam width, whereby beam steering at an antenna array (or arrays) is required to cover the needed angular space. The beam steering capabilities may be implemented using tunable phase shifters at each element in the antenna array, whereby the direction of the beam can be controlled electrically (phased array) instead of mechanical control.
[0122] The XPD of any antenna (or antenna array) depends on its radiation pattern and may change dynamically as a function of the Angle of Departure (AoD) and / or Angle of Arrival (AoA). This dependency increases as the variations in the radiation pattern change and as the radiation patterns change electrically. Higher antenna gain pattern leads to larger XPD variations over the angular space. Phase controlled arrays also have increased XPD variations over the angular space.
[0123] As such, the physical orientation of antennas at mmWave frequencies will affect MIMO throughput much more than what is seen at Sub-6 GHz frequencies, where the decorrelation at the UE is achieved by physical separation between two receiving antennas (each with random and different radiation pattern). Instead, the mmWave architecture may utilize dual orthogonally polarized antennas (or antenna arrays), designed for equal high gain radiation patterns.
[0124] Figure 5 illustrates an example apparatus capable of supporting at least some embodiments of the present invention. There is illustrated an apparatus 500, which may be, or comprised in, for example, a UE 410. Comprised in apparatus 500 is processor 510, which may comprise, for example, a single- or multi-core processor wherein a single-core processor comprises one processing core and a multi-core processor comprises more than one processing core. Processor 510 may comprise, in general, a control device. Processor 510 may comprise more than oneprocessor. Processor 510 may be a control device. A processing core may comprise, for example, a Cortex-A8 processing core manufactured by ARM Holdings or a Steamroller processing core produced by Advanced Micro Devices Corporation. Processor 510 may comprise at least one Qualcomm Snapdragon and / or Intel Atom processor. Processor 510 may comprise at least one Application- Specific Integrated Circuit, ASIC. Processor 510 may comprise at least one Field- Programmable Gate Array, FPGA. Processor 510 may be means for performing method steps in device 500. Processor 510 may be configured, at least in part by computer instructions, to perform actions.
[0125] A processor may comprise circuitry, or be constituted as circuitry or circuitries, the circuitry or circuitries being configured to perform phases of methods in accordance with embodiments described herein. As used in this application, the term “circuitry” may refer to one or more or all of the following: (a) hardware-only circuit implementations, such as implementations in only analog and / or digital circuitry, and (b) combinations of hardware circuits and software, such as, as applicable: (i) a combination of analog and / or digital hardware circuit(s) with software / firmware and (ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as UE 410 or BS 420, to perform various functions) and (c) hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation.
[0126] This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and / or firmware. The term circuitry also covers, for example and if applicable to the particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in server, a cellular network device, or other computing or network device.
[0127] Device 500 may comprise memory 520. Memory 520 may comprise random-access memory and / or permanent memory. Memory 520 may comprise at least one RAM (random access memory) chip, and / or at least one ROM (read-onlymemory). Memory 520 may comprise solid-state, magnetic, optical and / or holographic memory, for example. Memory 520 may be at least in part accessible to processor 510. Memory 520 may be at least in part comprised in processor 510. Memory 520 may be means for storing information. Memory 520 may comprise computer instructions that processor 510 is configured to execute. When computer instructions configured to cause processor 510 to perform certain actions are stored in memory 520, and device 500 overall is configured to run under the direction of processor 510 using computer instructions from memory 520, processor 510 and / or its at least one processing core may be considered to be configured to perform said certain actions. Memory 520 may be at least in part comprised in processor 510. Memory 520 may be at least in part external to device 500 but accessible to device 500.
[0128] Device 500 may comprise a transmitter 530. Device 500 may comprise a receiver 540. Transmitter 530 and receiver 540 may be configured to transmit and receive, respectively, information in accordance with at least one cellular or non- cellular standard. Transmitter 530 may comprise more than one transmitter. Receiver 540 may comprise more than one receiver. Transmitter 530 and / or receiver 540 may be configured to operate in accordance with Global System for Mobile Communication, GSM, Wideband Code Division Multiple Access, WCDMA, 5G / NR, Long Term Evolution, LTE, IS-95, Wireless Local Area Network, WLAN, Ethernet and / or Worldwide Interoperability for Microwave Access, WiMAX, standards, for example.
[0129] Device 500 may comprise a Near-Field Communication, NFC, transceiver 550. NFC transceiver 550 may support at least one NFC technology, such as NFC, Bluetooth, Wibree or similar technologies.
[0130] Device 500 may comprise User Interface, Ul, 560. Ul 560 may comprise at least one of a display, a keyboard, a touchscreen, a vibrator arranged to signal to a user by causing device 500 to vibrate, a speaker and a microphone. A user may be able to operate device 500 via Ul 560, for example to accept incoming telephone calls, to originate telephone calls or video calls, to browse the Internet, to manage digital files stored in memory 520 or on a cloud accessible via transmitter 530 and receiver 540, or via NFC transceiver 550, and / or to play games.
[0131] Device 500 may comprise or be arranged to accept a user identity module 570. User identity module 570 may comprise, for example, a Subscriber Identity Module, SIM, card installable in device 500. A user identity module 570 may comprise information identifying a subscription of a user of device 500. A user identity module 570 may comprise cryptographic information usable to verify the identity of a user of device 500 and / or to facilitate encryption of communicated information and billing of the user of device 500 for communication effected via device 500.
[0132] Processor 510 may be furnished with a transmitter arranged to output information from processor 510, via electrical leads internal to device 500, to other devices comprised in device 500. Such a transmitter may comprise a serial bus transmitter arranged to, for example, output information via at least one electrical lead to memory 520 for storage therein. Alternatively to a serial bus, the transmitter may comprise a parallel bus transmitter. Likewise processor 510 may comprise a receiver arranged to receive information in processor 510, via electrical leads internal to device 500, from other devices comprised in device 500. Such a receiver may comprise a serial bus receiver arranged to, for example, receive information via at least one electrical lead from receiver 540 for processing in processor 510. Alternatively to a serial bus, the receiver may comprise a parallel bus receiver.
[0133] Device 500 may comprise further devices not illustrated in Figure 5. For example, where device 500 comprises a smartphone, it may comprise at least one digital camera. Some devices 500 may comprise a back-facing camera and a frontfacing camera, wherein the back-facing camera may be intended for digital photography and the front-facing camera for video telephony. Device 500 may comprise a fingerprint sensor arranged to authenticate, at least in part, a user of device 500. In some embodiments, device 500 lacks at least one device described above. For example, some devices 500 may lack an NFC transceiver 550 and / or user identity module 570.
[0134] Processor 510, memory 520, transmitter 530, receiver 540, NFC transceiver 550, III 560 and / or user identity module 570 may be interconnected by electrical leads internal to device 500 in a multitude of different ways. For example, each of the aforementioned devices may be separately connected to a master bus internal to device 500, to allow for the devices to exchange information. However,as the skilled person will appreciate, this is only one example and depending on the embodiment various ways of interconnecting at least two of the aforementioned devices may be selected without departing from the scope of the present invention.
[0135] According to an embodiment of the invention, UE 410 may monitor and / or evaluate its current MIMO quality. For example, UE 410 may monitor a certain parameter or a set of parameters related to MIMO quality (in a non-limited embodiment this can be MIMO rank, a parameter related to MIMO rank or used to obtain MIMO rank value which UE then reports to gNB, or any other suitable quality metric or metrics). For example, UE 410 may apply various algorithms to estimate the current MIMO rank, based on the measurements on reference signals (such as DMRS). In this context, invertibility of the MIMO channel matrix may be used, and different metrics may be considered, such as singular values, condition number, as well as any other metric that may qualify the invertibility of the MIMO channel matrix. Averages of various metrics, or any filtered values, could also be used, such as sliding window, finite impulse response filter, infinite impulse response filter or any order (e.g. first, second, ... , n-th order) should also be considered. The Frequency average of channel estimates of the MIMO channel may be also taken into consideration, such as narrow band on single subcarrier DMRS or full-blown bandwidth allocated to UE, with different averaging techniques.
[0136] According to an embodiment of the invention, UE 410 may correlate MIMO transmission and / or reception quality with a Power Delay Profile (PDP) of an incoming signal. The PDP characterizes an intensity of a signal received through a multipath channel as a function of time delay, as described earlier. PDP thus may indicate multipath component(s) based on power peaks above threshold level over a delay interval.
[0137] In an embodiment, the UE 410 may determine if a possible better antenna configuration at the UE 410 could improve transmission and / or reception quality (such as MIMO rank or any other suitable quality metric as described above) and / or throughput, based on a parameter associated with the current MIMO quality and a PDP.
[0138] In an embodiment, the UE may use reference signals to scan for a new component carrier and / or Angular Power Group (i.e. APG). This could be channel state information reference signal (i.e. CSI-RS) (for example, with repetition set to“On”), synchronization signal block (SSB), position reference signal or some other existing or future suitable reference signal or signals. Also dedicated reference signals may be defined in the standard for these purposes. In an embodiment, the UE may then configure a single or multiple (two or more) antenna arrays for MIMO reception of multiple component carriers of different Angle of Arrival (AoA), where the chosen polarization configuration is optimized for optimal MIMO rank and / or combined power.
[0139] Thus, according to at least some embodiments, UE may determine if it is receiving a different signal component (e.g. a second component) by evaluating radio channel estimated PDP of the received signal. The PDP estimation can be performed for each polarization, wherein each polarization may correspond to a particular MIMO branch. The PDP estimation can be done with wide beams. The UE can thereby avoid blindly doing a time-consuming full beam scan on the whole antenna array if it is not needed.
[0140] According to an embodiment, the findings of the present invention are utilized in order to configure the antennas on the UE for better MIMO transmission and / or reception, for example in the case where the envisioned direct dual polarized MIMO configuration fails to achieve a sufficiently high MIMO rank in either LoS or NLoS environments.
[0141] In at least some embodiments, a UE 410 may receive one or more reference signal from a base station (BS), such as demodulation reference signal (DMRS), CSI-RS, or some other reference signal or signals. The UE 410 may further use one or more received reference signal for evaluating MIMO quality (in a nonlimiting embodiment, this can be MIMO rank, a parameter related to or indicative of MIMO rank, or any other suitable quality metric or metrics). Alternatively, or in addition to using reference signals, the UE 410 may use some other method or combination of methods for evaluating MIMO quality.
[0142] In an embodiment, UE 410 may calculate its current MIMO quality (such as rank, a parameter indicating the rank, or any other suitable quality metric as described above) with current antenna configuration. Next, the UE 410 may determine whether the current MIMO quality is sufficient. For example, in an embodiment, UE 410 may compare its current MIMO quality with a pre-determined threshold. In an embodiment, the threshold can be for example UE-driven. In someother embodiments, the threshold can be controlled by the network. Embodiments of the present invention may be applied to various MIMO configurations, such as 2x2 MIMO, 4x4 MIMO, or any other MIMO configuration.
[0143] If MIMO quality is above the threshold, in an embodiment this may indicate that current MIMO quality (e.g. rank) is sufficient. The UE 410 may in this case keep the current antenna configuration unchanged. The UE 410 may further wait for the next scheduled reference signal (for example, DMRS, CSI-RS or some other reference signal). Comparing to a threshold can also be implemented in a form of evaluating TRUE / FALSE condition.
[0144] In an exemplary embodiment, an apparatus, such as, for example, UE 410 or BS 420, may comprise means for carrying out the embodiments described above and any combination thereof.
[0145] In an exemplary embodiment, a computer program may be configured to cause a method in accordance with the embodiments described above and any combination thereof. In an exemplary embodiment, a computer program product, embodied on a non-transitory computer readable medium, may be configured to control a processor to perform a process (i.e. a method) comprising the embodiments described above and any combination thereof.
[0146] In an exemplary embodiment, an apparatus, such as, for example, UE 410 or BS 420, may comprise at least one processor, and at least one memory including computer program code, wherein the at least one memory and the computer program code are configured to, with the at least one processor, cause the apparatus at least to perform the embodiments described above and any combination thereof.
[0147] In the following, different exemplifying embodiments will be described using, as an example of an access architecture to which the embodiments may be applied, a radio access architecture based on Long Term Evolution Advanced (LTE Advanced, LTE-A) or new radio (NR, 5G), without restricting the embodiments to such an architecture, however. A person skilled in the art appreciates that the embodiments may also be applied to other kinds of communications networks having suitable means by adjusting parameters and procedures appropriately. Some examples of other options for suitable systems are the universal mobile telecommunications system (UMTS) radio access network (UTRAN or E-UTRAN),long term evolution (LTE, the same as E-LITRA), wireless local area network (WLAN or WiFi), worldwide interoperability for microwave access (WiMAX), Bluetooth®, personal communications services (PCS), ZigBee®, wideband code division multiple access (WCDMA), systems using ultra-wideband (UWB) technology, sensor networks, mobile ad-hoc networks (MANETs) and Internet protocol multimedia subsystems (IMS) or any combination thereof.
[0148] According to a further aspect of the present invention, there is provided an apparatus comprising at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform the method of the above-described aspect of the present invention, or an embodiment thereof.
[0149] According to some further aspects, there is provided a computer program, a computer program product, a computer readable medium, or a non-transitory computer readable medium comprising program instructions for causing an apparatus to perform the method according to any one of the above-mentioned aspects or according to an embodiment thereof.
[0150] In general, the various embodiments of the present invention may be implemented in hardware or special purpose circuits or any combination thereof. While various aspects of the invention may be illustrated and described as block diagrams or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.
[0151] The term “non-transitory,” as used herein, is a limitation of the medium itself (i.e., tangible, not a signal) as opposed to a limitation on data storage persistency (e.g., RAM vs. ROM).
[0152] The foregoing description has provided by way of exemplary and nonlimiting examples a full and informative description of the exemplary embodiments of the present invention. However, various modifications and adaptations may become apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the appended examples. However, all such and similar modifications of the teachingsof the present invention will still fall within the scope of the present invention, represented by the appended claims.
[0153] In other words: While the foregoing examples are illustrative of the principles of the embodiments in one or more particular applications, it will be apparent to those of ordinary skill in the art that numerous modifications in form, usage and details of implementation can be made without the exercise of inventive faculty, and without departing from the principles and concepts of the present invention. Accordingly, it is not intended that the present invention be limited, except as by the claims set forth below.
[0154] Yet in other words, the present invention is not restricted merely to the embodiments presented above, but the present invention may vary within the scope of the claims.
Claims
CLAIMS1. A method for providing Channel State Information, CSI, feedback using channel prediction at a User Equipment, UE, side (102), characterized in that the method comprises the steps of: using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receiving the CSI-RS directly from the gNB (101 ) without request; verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and reporting an updated CPF (116) to the gNB (101 ).
2. The method according to claim 1 , characterized in that the method further comprises a step of: using CSI-RS sent by the gNB (101 ) for channel estimation in case the UE (102) does not have the training data available (304).
3. The method according to claim 1 , characterized in that the method further comprises steps of: requesting also a Precoding Matrix Indicator (PMI) of the latest estimated CSI from the gNB (101 ) in connection with requesting the CSI-RS (308), or receiving the PMI directly from the gNB (101 ) without request; and comparing (309) a PMI of the latest estimated CSI at the UE (102) with the PMI reported by the gNB (101 ).
4. The method according to claim 3, characterized in that the method further comprises a step of: if the error or the CS or the PMI comparison fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ).
5. The method according to claim 4, characterized in that the method further comprises steps of: if none of the error or the CS or the PMI comparison fulfils a certain criterion (310), the CPF is considered non-valid, and sending a non-acknowledgement message (NACK) (114, 313) to the gNB (101 ); and computing a new CPF (115) via steps (105, 106, 107; 303, 305).
6. The method according to claim 1 , characterized in that the method further comprises steps of: if none of the error or the CS fulfils a certain criterion (310), the CPF is considered non-valid, and sending a non-acknowledgement message (NACK) (114, 313) to the gNB (101 ); and computing a new CPF (115) via steps (105, 106, 107; 303, 305).
7. The method according to claim 1 , characterized in that the method further comprises a step of:predicting a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information (110).
8. The method according to claim 1 , characterized in that the method further comprises a step of: applying a Recurrent Neural Network, RNN, based Channel Predictor Function at the UE (102).
9. The method according to claim 1 , characterized in that the method further comprises a step of: instead of training a Neural Network, selecting an available function from a list using an autoregressive, AR, model.
10. The method according to claim 1 , characterized in that the method further comprises a step of: using an identical CPF in both the gNB (101 ) and the UE (102).11 . The method according to claim 1 , characterized in that the method further comprises steps of: at the start of the process, requesting the gNB (101 ) to send CSI-RS (304, 103) if the estimated CSI is not available at the UE (102); and receiving an ACK and the requested CSI-RS (104) from the gNB (101 ).
12. The method according to claim 1 , characterized in that the method further comprises a step of: in connection with step (116), reporting also the latest compressed CSI by the UE (102) to the gNB (101 ).
13. A method for providing Channel State Information, CSI, feedback using channel prediction at a base station, gNB, side (101 ), characterized in that the method comprises the steps of:receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); sending (104) an acknowledgement message (ACK) to the UE (102); receiving (108) a Channel Predictor Function, CPF, from the UE (102); receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receiving (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); and receiving (116, 311 ) an updated CPF from the UE (102).
14. The method according to claim 13, characterized in that the method further comprises steps of: receiving (103) a request of Channel State Information Reference Signals, CSI-RS, from the UE (102) in case that information is not available at the UE (102); and sending (104) the CSI-RS to the UE (102).
15. The method according to claim 13, characterized in that the method further comprises steps of: receiving (112, 308) also a request of a Precoding Matrix Indicator (PMI) of the latest estimated CSI from the UE (102) in connection with the request of the CSI- RS (308); and sending (308) the PMI from the gNB (101 ) to the UE (102) after the request; or sending (308) the PMI directly from the gNB (101 ) to the UE (102) without any received request.
16. The method according to claim 13, characterized in that the method further comprises a step of:after a received NACK in step (313), waiting for the UE (102) to compute a new CPF (115) via steps (105, 106, 107; 303, 305).
17. A User Equipment, UE, (102) for providing Channel State Information, CSI, feedback using channel prediction, characterized in that the UE (102) comprises: means for using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); means for computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); means for reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); means for using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); means for requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or means for receiving the CSI-RS directly from the gNB (101 ) without request; means for verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, the UE (102) further comprises means for sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and means for reporting an updated CPF (116) to the gNB (101 ).
18. A base station, gNB, (101 ) for providing Channel State Information, CSI, feedback using channel prediction, characterized in that the gNB (101 ) comprises:means for receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); means for sending (104) an acknowledgement message (ACK) to the UE (102); means for receiving (108) a Channel Predictor Function, CPF, from the UE (102); means for receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); means for predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; means for sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); means for receiving (114) an acknowledgement message (ACK) (311 ) or a non-acknowledgement message (NACK) (313) from the UE (102); and means for receiving (116, 311 ) an updated CPF from the UE (102).
19. A User Equipment, UE, (102) for providing Channel State Information, CSI, feedback using channel prediction, characterized in that the UE (102) comprises: at least one processor; and at least one memory, storing instructions that, when executed by the at least one processor, cause the UE (102) at least to: use uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); compute a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305);report the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); use the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); request Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receive the CSI-RS directly from the gNB (101 ) without request; verify the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, send an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and report an updated CPF (116) to the gNB (101 ).
20. A base station, gNB, (101 ) for providing Channel State Information, CSI, feedback using channel prediction, characterized in that the gNB (101 ) comprises: at least one processor; and at least one memory, storing instructions that, when executed by the at least one processor, cause the gNB (101 ) at least to: receive (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); send (104) an acknowledgement message (ACK) to the UE (102); receive (108) a Channel Predictor Function, CPF, from the UE (102); receive (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predict (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information;send (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receive (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); and receive (116, 311 ) an updated CPF from the UE (102).
21. A computer program comprising instructions which, when executed by a User Equipment, UE, (102), cause the UE to perform at least the following: using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102); requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receiving the CSI-RS directly from the gNB (101 ) without request; verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and reporting an updated CPF (116) to the gNB (101 ).
22. A computer program comprising instructions which, when executed by a base station, gNB, (101 ), cause the gNB to perform at least the following: receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); sending (104) an acknowledgement message (ACK) to the UE (102); receiving (108) a Channel Predictor Function, CPF, from the UE (102); receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receiving (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); and receiving (116, 311) an updated CPF from the UE (102).
23. A non-transitory computer readable medium comprising program instructions that, when executed by a User Equipment, UE (102), cause the UE to perform at least the following: using uncompressed estimated Channel State Information in the UE (102) for training a Neural Network (106, 303); computing a Channel Predictor Function in the UE (102) using predicted Channel State Information and past uncompressed Channel State Information realizations (107, 305); reporting the Channel Predictor Function, CPF, (108) and the latest compressed or uncompressed Channel State Information (109, 306) to a base station, gNB, (101 ) used for predicting Channel State Information in the gNB (101 ); using the CPF and the latest compressed or uncompressed Channel State Information to predict the Channel State Information (111 , 307) in the UE (102);requesting Channel State Information Reference Signals, CSI-RS, (112, 308) from the gNB (101 ), or receiving the CSI-RS directly from the gNB (101 ) without request; verifying the Channel Predictor Function (113, 309) in the UE (102) either through an error between predicted and estimated ones, or through Cosine Similarity, CS; if the error or the CS fulfils a certain criterion (310) to indicate a verified CPF, sending an acknowledgement message (ACK) (114, 311 ) to the gNB (101 ); and reporting an updated CPF (116) to the gNB (101 ).
24. A non-transitory computer readable medium comprising program instructions that, when executed by a base station, gNB (101 ), cause the gNB to perform at least the following: receiving (103), from a User Equipment, UE, (102), information of an implementation of a Channel Predictor Function at the UE (102); sending (104) an acknowledgement message (ACK) to the UE (102); receiving (108) a Channel Predictor Function, CPF, from the UE (102); receiving (109) a latest compressed or uncompressed Channel State Information, CSI, (109, 306) from the UE (102); predicting (110) a Channel State Information in the gNB (101 ) by exploiting the Channel Predictor Function, and the latest compressed or uncompressed Channel State Information; sending (112, 308) Channel State Information Reference Signals, CSI-RS, to the UE (102) either directly or after a specific request from the UE (102); receiving (114) an acknowledgement message (ACK) (311 ) or a nonacknowledgement message (NACK) (313) from the UE (102); and receiving (116, 311 ) an updated CPF from the UE (102).
Citation Information
Patent Citations
Channel state information reporting
US20220123807A1
Cited By
Channel state information refinement
WO2026182861A1
