WO2024179655A1

WO2024179655A1 - Enhanced channel state information reporting

Info

Publication number: WO2024179655A1
Application number: PCT/EP2023/054792
Authority: WO
Inventors: Mehmet ILTER; Salah Eddine HAJRI; Thorsten Schier
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2023-02-27
Filing date: 2023-02-27
Publication date: 2024-09-06
Anticipated expiration: 2025-08-27

Abstract

Embodiments of the invention relate to enhanced channel state information (CSI) reporting for a client device (100) in a communication system. The client device (100) is enabled to determine the predictability of coefficients of a first CSI quantity for the client device (100) and report the coefficients of the first CSI quantity to a network access node (300) based on the determined predictability. The predictability indicates how useful, e.g., unique, the information provided by the coefficients of the first CSI quantity is and allows the client device (100) to prioritize which coefficients of the first CSI quantity to report the network access node (300). In this way, the client device (100) can e.g., only report coefficients of the first CSI quantity providing useful information and omit coefficients of the first CSI quantity which do not provide any new information. Furthermore, embodiments of the invention also relate to corresponding methods and a computer program.

Description

ENHANCED CHANNEL STATE INFORMATION REPORTING

Technical Field

Embodiments of the invention relate to enhanced channel state information (CSI) reporting for a client device in a communication system. Furthermore, embodiments of the invention also relate to corresponding methods and a computer program.

Background

Several of 5G new radio (NR) main features make extensive use of multi-antenna techniques to achieve its key performance indicators in terms of reliability, coverage, latency and throughput, among others. By leveraging multiple antennas at the network node and the user equipment (UE), beamforming can be used to improve system performance.

To select the proper beamforming weights and link adaptation decisions CSI is needed. From the network perspective, CSI can be obtained following UE feedback and/or uplink reference signal measurements. CSI can be computed with different assumptions on the transmission/interference hypotheses.

To adapt to different conditions and use cases, 5G NR specifies a quite versatile CSI reporting framework, including e.g., different reporting behavior and reporting resources, multiple CSI reporting quantities and multiple options for reference signal mapping and transmission behavior in the time domain.

The restrictions applicable to different reporting behaviors vary and have an impact on the CSI timeline, i.e., the time from the CSI reporting was triggered to the CSI report is decoded at the network node. Aperiodic and semi-persistent CSI reporting provide a high degree of flexibility in scheduling CSI reports as they rely on triggering CSI reporting using dynamic signaling. Of course, this means that correct decoding of the trigger message is mandatory for the CSI report to be prepared within the expected timeline.

Summary

An objective of embodiments of the invention is to provide a solution which mitigates or solves the drawbacks and problems of conventional solutions. Another objective of embodiments of the invention is to provide a solution for enhanced reporting of CSI quantities.

The above and further objectives are solved by the subject matter of the independent claims. Further embodiments of the invention can be found in the dependent claims.

According to a first aspect of the invention, the above mentioned and other objectives are achieved with a client device for a communication system, the client device being configured to: determine a predictability metric for a set of coefficients of a first channel state information, CSI, quantity for the client device based on the first CSI quantity and a second CSI quantity; and transmit a CSI report to a network access node, the CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

An advantage of the client device according to the first aspect is that the coefficients in a CSI report can be grouped according to their predictability, e.g., correlation/di stance with respect to previous CSI reports or other coefficients within the CSI report. When CSI omission is needed, e.g., when multiplexing CSI report with other uplink payloads or for prioritization among CSI reports, CSI coefficients can be dropped from the CSI report based on their predictability group. Consequently, minimal information loss occurs which ultimately benefits CSI data collection for machine learning (ML) and non-ML based CSI procedures.

In an implementation form of a client device according to the first aspect, the predictability metric is based on a distance between the set of coefficients of the first CSI quantity and a set of coefficients of the second CSI quantity.

An advantage with this implementation form is that predictability can be quantified by the distance of subsets of coefficients with respect to previous measurements or statistics of previous measurements. The higher the distance, the more informative the coefficients are e.g., as training data to develop capable ML models. Consequently, coefficients associated with higher distances can be reported with priority. In an implementation form of a client device according to the first aspect, the predictability metric is based on a correlation between the set of coefficients of the first CSI quantity and a set of coefficients of the second CSI quantity.

An advantage with this implementation form is that predictability can be quantified by the correlation of subsets of coefficients with respect to previous measurements or statistics of previous measurements. The higher the correlation, the more predictable the coefficients are. Predictable coefficients are easily obtainable with high accuracy, using adequate interpolation methods. Consequently, less correlated coefficients can be reported with priority in order to properly design CSI prediction algorithms at the network side.

In an implementation form of a client device according to the first aspect, the predictability metric is based on a training performance and/or an inference performance for a machine learning model associated with the first CSI quantity and/or the second CSI quantity.

An advantage with this implementation form is that the predictability metric can take into consideration the monitoring of ML models during training and/or inference. If subsets of coefficient were found to cause errors in prediction, there reporting can be prioritized so that the ML model could be adapted accordingly.

In an implementation form of a client device according to the first aspect, the predictability metric is based on a divergence between an estimated distribution based on the first CSI quantity and a reference distribution based on the second CSI quantity.

An advantage with this implementation form is that statistical measurements, determined based on prior CSI reporting or other CSI quantities, can be used as a reference to discriminate coefficients based on their predictability. For example, the farther the coefficients are from the estimated distribution the higher their reporting priority.

In an implementation form of a client device according to the first aspect, the first CSI quantity and the second CSI quantity are based on measurements of reference signals transmitted at a first time instance; or the first CSI quantity is based on measurements of reference signals transmitted at the first time instance and the second CSI quantity is based on measurements of reference signals transmitted at a second time instance; or the first CSI quantity is based on measurements of reference signals transmitted at the first time instance and the second CSI quantity is based on reference measurements.

An advantage with this implementation form is that different predictability metrics can be supported, enabling to address the specific needs of different ML procedures. For example, when time domain prediction for beam or precoding matrix indicators (PMIs) are used, the predictability metric can reflect a distance or a correlation over time. In the case of beam prediction in the spatial domain, predictability metric can reflect correlation in the space domain between beams or PMI supports. Additionally, reference CSI quantities can be derived from statistics of previous CSI reports and used to quantify predictability.

In an implementation form of a client device according to the first aspect, the CSI report indicates the subset of coefficients of the first CSI quantity when the predictability metric fulfills a criterion.

An advantage with this implementation form is that the reporting of subset of coefficients of the first CSI quantity can be performed as a function of their predictability. Coefficients which are most prone to be predictable, with high probability, are removed and only the most informative coefficients are transmitted. This behavior can be used as a CSI omission rule, when uplink channel multiplexing is problematic due to limited uplink resources. Additionally, it can be used to reduce the overhead when collecting training data for ML models at the network side.

In an implementation form of a client device according to the first aspect, the client device is configured to: transmit the CSI report to the network access node based on one or more of an uplink resource for the CSI report, a bit width of the CSI report and a bit width of the subset of coefficients of the first CSI quantity.

An advantage with this implementation form is that selecting the subsets of coefficients to report, based on predictability, can be used as a CSI omission rule, when uplink channel multiplexing is problematic due to limited uplink resources. Additionally, it can be used to reduce the overhead when collecting training data for ML models at the network side.

In an implementation form of a client device according to the first aspect, the client device is configured to: transmit a first control signal to the network access node, the first control signal indicating that the CSI report is based on the predictability metric and/or a reporting format for the CSI report.

An advantage with this implementation form is that the first control signal can be used by the network access node to distinguish the cases where dropping of CSI coefficients happened due to high predictability. Which enables the network access node to perform the correct processing of the received subset of coefficients of the first CSI quantity.

In an implementation form of a client device according to the first aspect, the client device is configured to: receive a second control signal from the network access node, the second control signal indicating a measurement configuration for the predictability metric and/or a format of the predictability metric; and determine the predictability metric based on the second control signal.

An advantage with this implementation form is that the network access node can indicate, in the second control signal, a configuration which indicates the predictability metric that is suitable, given the implemented prediction algorithms at the network side, without disclosing the implemented prediction algorithms.

According to a second aspect of the invention, the above mentioned and other objectives are achieved with a network access node for a communication system, the network access node being configured to: receive a CSI report from a client device, the CSI report indicating a subset of coefficients of a first CSI quantity for the client device, the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity. An advantage of the network access node according to the second aspect is that the network access node can perform ML-based predictions based on the received subset of coefficients, without considerable performance loss, even if coefficients have been dropped by the client device. As the subset of coefficients is determined based on their predictability, the most predictable coefficients can be dropped, in the event of a CSI omission.

In an implementation form of a network access node according to the second aspect, the network access node is configured to: receive a first control signal from the client device, the first control signal indicating that the CSI report is based on a predictability metric for the set of coefficients of the first CSI quantity and/or a reporting format for the CSI report.

An advantage with this implementation form is that the first control signal can be used by the network access node to distinguish the cases where dropping of CSI coefficients happened due to high predictability. Which enables the network access node to perform the correct processing of the received subset of coefficients.

In an implementation form of a network access node according to the second aspect, the network access node is configured to: transmit a second control signal to the client device, the second control signal indicating a measurement configuration for the predictability metric and/or a format of the predictability metric.

An advantage with this implementation form is that the network access node indicates, in the second control signal, a configuration which indicates the predictability metric that is suitable, given the implemented prediction algorithms at the network side, without disclosing the implemented prediction algorithms.

According to a third aspect of the invention, the above mentioned and other objectives are achieved with a method for a client device, the method comprises: determining a predictability metric for a set of coefficients of a first channel state information, CSI, quantity for the client device based on the first CSI quantity and a second CSI quantity; and transmitting a CSI report to a network access node, the CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

The method according to the third aspect can be extended into implementation forms corresponding to the implementation forms of the client device according to the first aspect. Hence, an implementation form of the method comprises the feature(s) of the corresponding implementation form of the client device.

The advantages of the methods according to the third aspect are the same as those for the corresponding implementation forms of the client device according to the first aspect.

According to a fourth aspect of the invention, the above mentioned and other objectives are achieved with a method for a network access node, the method comprises: receiving a CSI report from a client device, the CSI report indicating a subset of coefficients of a first CSI quantity for the client device, the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity.

The method according to the fourth aspect can be extended into implementation forms corresponding to the implementation forms of the network access node according to the second aspect. Hence, an implementation form of the method comprises the feature(s) of the corresponding implementation form of the network access node.

The advantages of the methods according to the fourth aspect are the same as those for the corresponding implementation forms of the network access node according to the second aspect.

Embodiments of the invention also relate to a computer program, characterized in program code, which when run by at least one processor causes the at least one processor to execute any method according to embodiments of the invention. Further, embodiments of the invention also relate to a computer program product comprising a computer readable medium and the mentioned computer program, wherein the computer program is included in the computer readable medium, and may comprises one or more from the group of read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), flash memory, electrically erasable PROM (EEPROM), hard disk drive, etc.

Further applications and advantages of embodiments of the invention will be apparent from the following detailed description.

Brief Description of the Drawings

The appended drawings are intended to clarify and explain different embodiments of the invention, in which:

- Fig. 1 shows a client device according to an embodiment of the invention;

- Fig. 2 shows a flow chart of a method for a client device according to an embodiment of the invention;

- Fig. 3 shows a network access node according to an embodiment of the invention;

- Fig. 4 shows a flow chart of a method for a network access node according to an embodiment of the invention;

- Fig. 5 shows a communication system according to an embodiment of the invention;

- Fig. 6 shows signaling for CSI reporting based on a predictability metric according to an embodiment of the invention;

- Fig. 7 shows correlation with previous CSI reports according to an embodiment of the invention; and

- Fig. 8 shows physical resource usage for CSI reporting according to an embodiment of the invention.

Detailed Description

The current CSI framework suffers from several impairments related to the content of the CSI data, such as e.g., outdated channel matrices due to computation and transmission delays, limited number of measurements, limited PMI accuracy, limited data volumes and timeliness of data. These aspects may hinder the achievable performance with the current CSI framework, particularly performance related to artificial intelligence (AI)/machine learning (ML) algorithms. In 3GPP, these aspects of the CSI framework were discussed and solutions to minimize the usage of physical resource and improve the granularity of the information were proposed. The payload of type II PMI codebooks, up to 3GPP release 17, is arranged in two parts. Part 1 is characterized by a fixed size, regardless of the number of layers, and contains information that enables the next generation NodeB (gNB) to derive the size of part 2. The main payload of the reported PMI is contained in part 2 which has a varying size, depending on e.g., the number of layers, configured codebook parameters, etc. In one slot, a UE can be configured to report multiple CSI reports, each comprising CSI quantities having different formats.

The configuration of CSI reporting is semi-static. Consequently, the CSI payload can change more rapidly than the available uplink resources for CSI reporting. Ultimately, this may result in having scheduled uplink resources that fall short from fitting the entire payload of all the scheduled CSI reports. When the payload of a CSI report, exceeds the allocated uplink resources for reporting, CSI omission rules are applicable. CSI omission rules define a mechanism that enables the UE to drop sections of CSI reports when the CSI payload exceeds the available allocated uplink resources.

The conventional CSI omission rules are based on a subdivision of each CSI report in a number of groups, where wideband information is given a higher priority than frequency selective information. When a UE is scheduled to transmit a transport block on physical uplink shared channel (PUSCH) multiplexed with CSI reports, payload of the CSI reports can be omitted when the PUSCH resource is not sufficient to contain all the scheduled CSI reports. The payload of the CSI reports can be omitted level by level, beginning with the lowest priority level until reaching the lowest priority level which causes the uplink payload to fit the scheduled PUSCH resource.

The standardized priority rules do not consider informational criteria, beyond the current reporting instance. Indeed, one of the major design principles for CSI reporting in long term evolution (LTE) and NR, was that the CSI reports should be self-contained. This means that information redundancy, which may exist between consecutive CSI reports, is not considered. Thus, timeliness of channel measurements, correlation with previous and future CSI reports and frequency and time granularity of channel measurements are not properly leveraged, especially during CSI omission. In the case where the gNB has predictive capabilities and/or when collecting data for ML model training, it would be beneficial if CSI omission rules would consider data properties relevant to the proper functioning of the ML models, such as bias and uniqueness, over multiple reporting instances. ML models work best when the training data is representative of all possible channel conditions that may be encountered during prediction.

According to embodiments of the invention a solution to determine a predictability metric for coefficients of a first CSI quantity is therefore provided. The predictability metric indicates a variation between the coefficients of the first CSI quantity and a second CSI quantity and can hence indicate if the information provided by the coefficients of the first CSI quantity is useful to a network access node, e.g., for ML model training. Using the determined predictability metric, the client device can prioritize which coefficients of the first CSI quantity to report to the network access node when limited CSI reporting resources are available.

Fig. 1 shows a client device 100 according to an embodiment of the invention. In the embodiment shown in Fig. 1, the client device 100 comprises a processor 102, a transceiver 104 and a memory 106. The processor 102 is coupled to the transceiver 104 and the memory 106 by communication means 108 known in the art. The client device 100 further comprises an antenna or antenna array 110 coupled to the transceiver 104, which means that the client device 100 is configured for wireless communications in a communication system.

The processor 102 may be referred to as one or more general-purpose central processing units (CPUs), one or more digital signal processors (DSPs), one or more application-specific integrated circuits (ASICs), one or more field programmable gate arrays (FPGAs), one or more programmable logic devices, one or more discrete gates, one or more transistor logic devices, one or more discrete hardware components, or one or more chipsets. The memory 106 may be a read-only memory, a random access memory (RAM), or a non-volatile RAM (NVRAM). The transceiver 104 may be a transceiver circuit, a power controller, or an interface providing capability to communicate with other communication modules or communication devices. The transceiver 104, memory 106 and/or processor 102 may be implemented in separate chipsets or may be implemented in a common chipset. That the client device 100 is configured to perform certain actions can in this disclosure be understood to mean that the client device 100 comprises suitable means, such as e.g., the processor 102 and the transceiver 104, configured to perform the actions.

According to embodiments of the invention the client device 100 is configured to determine a predictability metric for a set of coefficients of a first CSI quantity for the client device 100 based on the first CSI quantity and a second CSI quantity. The client device 100 is further configured to transmit a CSI report to a network access node 300. The CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

Furthermore, in an embodiment of the invention, the client device 100 for a communication system 500 comprises: a processor configured to determine a predictability metric for a set of coefficients of a first CSI quantity for the client device 100 based on the first CSI quantity and a second CSI quantity; and a transceiver/processor configured to transmit a CSI report to a network access node 300, the CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

Moreover, in yet another embodiment of the invention, the client device 100 for a communication system 500 comprises a processor and a memory having computer readable instructions stored thereon which, when executed by the processor, cause the processor to: determine a predictability metric for a set of coefficients of a first CSI quantity for the client device 100 based on the first CSI quantity and a second CSI quantity; and transmit a CSI report to a network access node 300, the CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

Fig. 2 shows a flow chart of a corresponding method 200 which may be executed in a client device 100, such as the one shown in Fig. 1. The method 200 comprises determining 202 a predictability metric for a set of coefficients of a first CSI quantity for the client device 100 based on the first CSI quantity and a second CSI quantity. The method 200 further comprises transmitting 204 a CSI report to a network access node 300, the CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

Fig. 3 shows a network access node 300 according to an embodiment of the invention. In the embodiment shown in Fig. 3, the network access node 300 comprises a processor 302, a transceiver 304 and a memory 306. The processor 302 is coupled to the transceiver 304 and the memory 306 by communication means 308 known in the art. The network access node 300 may be configured for wireless and/or wired communications in a communication system. The wireless communication capability may be provided with an antenna or antenna array 310 coupled to the transceiver 304, while the wired communication capability may be provided with a wired communication interface 312 e.g., coupled to the transceiver 304.

The processor 302 may be referred to as one or more general-purpose CPU, one or more DSPs, one or more ASICs, one or more FPGAs, one or more programmable logic devices, one or more discrete gates, one or more transistor logic devices, one or more discrete hardware components, one or more chipsets. The memory 306 may be a read-only memory, a RAM, or a NVRAM. The transceiver 304 may be a transceiver circuit, a power controller, or an interface providing capability to communicate with other communication modules or communication devices, such as network nodes and network servers. The transceiver 304, the memory 306 and/or the processor 302 may be implemented in separate chipsets or may be implemented in a common chipset.

That the network access node 300 is configured to perform certain actions can in this disclosure be understood to mean that the network access node 300 comprises suitable means, such as e.g., the processor 302 and the transceiver 304, configured to perform the actions.

According to embodiments of the invention the network access node 300 is configured to receive a CSI report from a client device 100, the CSI report indicating a subset of coefficients of a first CSI quantity for the client device 100, the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity.

Furthermore, in an embodiment of the invention, the network access node 300 for a communication system 500 comprises: a transceiver configured to receive a CSI report from a client device 100, the CSI report indicating a subset of coefficients of a first CSI quantity for the client device 100, the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity.

Moreover, in yet another embodiment of the invention, the network access node 300 for a communication system 500 comprises a processor and a memory having computer readable instructions stored thereon which, when executed by the processor, cause the processor to: receive a CSI report from a client device 100, the CSI report indicating a subset of coefficients of a first CSI quantity for the client device 100, the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity.

Fig. 4 shows a flow chart of a corresponding method 400 which may be executed in a network access node 300, such as the one shown in Fig. 3. The method 400 comprises receiving 402 a CSI report from a client device 100, the CSI report indicating a subset of coefficients of a first CSI quantity for the client device 100, the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity.

Fig. 5 shows a communication system 500 according to an embodiment of the invention. The communication system 500 in the disclosed embodiment comprises a client device 100 and a network access node 300 configured to communicate and operate in the communication system 500. The network access node 300 may be connected to a network NW such as e.g., a core network over a communication interface. The communication system 500 may be a communication system according to the 3GPP standard such as e.g., a 5G system in which case the client device 100 may be a UE and the network access node 300 may be a gNB but the invention is not limited thereto.

In the shown embodiment, it is assumed that the client device 100 has been configured by the network access node 300 to perform CSI measurements and reporting in a conventional way. As previously described, the payload of the CSI reports can change more rapidly than the available uplink resources for CSI reporting. This may result in that the entire payload of the scheduled CSI reports cannot fit into the scheduled uplink resources. The client device 100 then determines CSI reports or parts of CSI reports to omit or skip to adapt the payload of the CSI reports to the available allocated uplink resources.

According to embodiments of the invention the client device 100 is enabled to omit or skip CSI reports or parts of CSI reports based on a predictability metric for a set of coefficients of a first CSI quantity for the client device 100. The client device 100 determines the predictability metric for the set of coefficients of the first CSI quantity and transmit a CSI report to the network access node 300 based on the predictability metric, as shown in Fig. 5. The predictability metric indicates how the set of coefficients of the first CSI quantity vary from a second CSI quantity, e.g., a set of previously computed coefficients and/or reference coefficients, and hence whether the set of coefficients of the first CI quantity provides information which is useful to the network access node 300 or not. Based on the predictability metric, the client device 100 can determine a subset of coefficients of the first CSI quantity from the set of coefficients of the first CSI quantity to include in the CSI report. The subset of coefficients of the first CSI quantity may comprise the coefficients providing the most useful information, e.g., the coefficients with the lowest predictability. In this way, the client device 100 can prioritize CSI information to report to the network access node 300, when there are limited reporting resources.

Fig. 6 shows signaling for CSI reporting based on the predictability metric according to an embodiment of the invention. In the shown embodiment, the client device 100 determines the predictability metric at least partly based on information received from the network access node 300, i.e., the network access node 300 configures the client device 100 to determine the predictability metric. However, the client device 100 may in embodiments instead be preconfigured to determine the predictability metric. The client device 100 may further perform or trigger the determination and use of the predictability metric either autonomously, e.g., based on one or more conditions, or based on a control signal from the network access node 300 or another network node. In other words, step I in Fig. 6 is optional as indicated with a dashed arrow.

In optional step I in Fig. 6, the network access node 300 transmits a second control signal 520 to client device 100. The second control signal 520 indicates a measurement configuration for the predictability metric and/or a format of the predictability metric. The measurement configuration for the predictability metric may indicate a number of and/or interval of measurements for determining a first CSI quantity and/or a second CSI quantity for the client device 100. The format of the predictability metric may indicate the type of predictability metric to use, i.e., whether the predictability metric is to be determined based on a distance or correlation between the first CSI quantity and the second CSI quantity, a performance of a ML model, or a divergence between the first CSI quantity and the second CSI quantity, as further described with reference to step II in Fig. 6. The second control signal 520 may e.g., be a radio resource control (RRC) message provided to the client device 100 during RRC configuration and/or reconfiguration.

The client device 100 receives the second control signal 520 from the network access node 300 and hence obtains the information indicated in the second control signal 520, i.e., the measurement configuration for the predictability metric and/or the format of the predictability metric. Based on the received second control signal 520, the client device 100 determines the predictability metric in step II in Fig. 6. The client device 100 may e.g., determine the first CSI quantity and/or the second CSI quantity based on the measurement configuration for the predictability metric and/or determine the predictability metric based on the format of the predictability metric indicated in the second control signal 520. However, the client device 100 may in embodiments instead perform step II in Fig. 6 without first receiving the second control signal 520 from the network access node 300.

In embodiments, CSI reporting based on the predictability metric may be semi-persistent and e.g., activated and/or de-activated by the network access node 300. In this case, the client device 100 may start to determine the predictability metric and/or transmit the CSI report based on the predictability metric upon receiving an activation command from the network access node 300 (not shown in Figs.). The client device 100 may then cease to use the predictability metric based on a de-activation command. The activation and/or de-activation of CSI reporting based on the predictability metric may be conveyed from the network access node 300 to the client device 100 as part of a RRC message or dynamic layer one LI and/or layer two L2 signaling, e.g., in a downlink control information (DCI) or a medium access control - control element (MAC- CE).

In step II in Fig. 6, the client device 100 determines a predictability metric for a set of coefficients of a first CSI quantity for the client device 100 based on the first CSI quantity and a second CSI quantity. The first CSI quantity and/or the second CSI quantity may be channel estimations for CSI computation based on downlink reference signals measurements. The first CSI quantity and/or the second CSI quantity may e.g., be a channel matrix, a PMI or a channel covariance matrix, either wideband or computed per subband, or a channel support in the spatial and/or delay domains, but is not limited thereto. In embodiments, the first CSI quantity and the second CSI quantity are the same CSI quantity, e.g., both the first CSI quantity and the second CSI quantity are PMIs. The first CSI quantity and/or the second CSI quantity may comprise a number of coefficients, each coefficient representing a computed value for the CSI quantity for a specific layer and/or subband, e.g., a PMI for a subband, amplitude and cophasing coefficients per subband or discrete Fourier transform (DFT) component. The set of coefficients of the first CSI quantity may comprise one or more of the coefficients of the first CSI quantity, i.e., one or more of the coefficients of the first CSI quantity that the client device 100 has been configured to report.

The first CSI quantity and the second CSI quantity may be channel estimations from the same or different time instance. Thus, the first CSI quantity and the second CSI quantity may be based on measurements of reference signals transmitted at a first time instance; or the first CSI quantity may be based on measurements of reference signals transmitted at the first time instance and the second CSI quantity may be based on measurements of reference signals transmitted at a second time instance. For example, the first CSI quantity and the second CSI quantity may e.g., be different channel estimates for the same CSI report at the same time instance or be different channel estimates for different CSI reports at different time instances.

Furthermore, the first CSI quantity may be compared against reference values. The first CSI quantity may in embodiments hence be based on measurements of reference signals transmitted at the first time instance and the second CSI quantity may be based on reference measurements. The reference measurements may e.g., be based on statistical values or configured values, either computed by the client device 100 or obtained in another way, e.g., from the network access node 300.

The client device 100 may determine the predictability metric in a number of different ways. In embodiments, the predictability metric is based on a distance between the set of coefficients of the first CSI quantity and a set of coefficients of the second CSI quantity. Thus, the predictability of the set of coefficients of the first CSI quantity can be quantified by the distance, where larger distance may indicate more informative coefficients. The predictability metric may e.g., be a distance metric based on a Log Euclidean distance or a Jensen-Bregman LogDet Divergence considering one or multiple reference signal measurement occasions. When the second CSI quantity is based on reference measurements, the predictability metric may indicate a distance from the set of coefficients of the first CSI quantity to reference coefficients such as e.g., a wideband covariance matrix, or a distance from the set of coefficients of the first CSI quantity to a statistical median and/or mean of CSI coefficients. In this way, outliers, i.e., coefficients with large distance from a given reference coefficient, can be identified. For example, coefficients with larger or lower amplitude than the wideband amplitude. Outliers can be very informative and may hence be prioritized to be reported to the network access node 300. When ML is used at the network access node 300, e.g., for PMI or beam prediction, training is performed based on CSI data. In this case, the collected CSI data needs to reflect to the best extend possible, the measurements that may be encountered during inference operation. Outliers are very informative and their availability in the training data enable to develop capable ML models. For example, take into consideration a realistic distribution over the output space of ML models.

The predictability metric may further be based on a correlation between the set of coefficients of the first CSI quantity and the set of coefficients of the second CSI quantity. The predictability metric may e.g., be a correlation metric based on a Pearson correlation coefficient or 2-D crosscorrelation coefficients. The correlation may be an autocorrelation overtime or correlation with previous CSI reports over a given look-back window. The correlation with respect to previously collected CSI reports may indicate the uniqueness of the set of coefficients of the first CSI quantity, i.e., whether the set of coefficients of the first CSI quantity are rarely seen compared to the previous CSI coefficients or not. The client device 100 may e.g., omit coefficients of the first CSI quantity that retain the highest correlation with one or more previous CSI reports, over a look-back window.

Fig. 7 shows correlation with previous CSI reports over a given look-back window according to an embodiment of the invention. In Fig. 7, the client device 100 determines the correlation between a first set of coefficients Set 1 for a first CSI report 1 and a second CSI report 2 at different time instances T1-T5 over a look-back window Wl, W2, W3. Each look-back window Wl, W2, W3 comprises three different time instances. In the shown embodiment, it is assumed that in the first lookback window Wl, the first set of coefficients Set 1 for the first CSI report 1 at the third time instance T3 has the highest correlation with the previous CSI reports. The first set of coefficients Set 1 for the first CSI report 1 at the third time instance T3 are hence omitted, i.e., not reported, as indicated with the black cross in Fig. 7. In the third lookback window W3, the first set of coefficients Set 1 for the second CSI report 2 at the fifth time instance T5 has the highest correlation with the previous CSI reports. The first set of coefficients Set 1 for the second CSI report 2 at the fifth time instance T5 are hence omitted, i.e., not reported.

The predictability metric may further be based on a training performance and/or an inference performance for a ML model associated with the first CSI quantity and/or the second CSI quantity. The training performance and/or an inference performance may indicate the importance of the first CSI quantity for training of the ML model and may e.g., be determined based on a training loss, a validation loss or an accuracy for the ML model. When the training performance and/or an inference performance indicates high confidence on the set of coefficients of the first CSI quantity, the set of coefficients of the first CSI quantity is less important for training the ML model and may hence be skipped without loss of accuracy for the ML model. In embodiments, the predictability metric can take into consideration the monitoring of ML models during training and/or inference. If subsets of coefficient were found to cause errors in prediction, there reporting can be prioritized so that the ML model could be adapted accordingly.

In embodiments, the predictability metric may be based on a divergence between an estimated distribution based on the first CSI quantity and a reference distribution based on the second CSI quantity. For example, Kullback-Leibler divergence or relative entropy can be calculated between two probability distributions, one obtained from previous results and one obtained for the set of coefficients of the first CSI quantity. As these measurements measures the separation in a statistical manifold between probability distributions, they can be used to measure how predictable a set of coefficients are, relative to previous CSI reports.

In step II in Fig. 6, the client device 100 determines and transmits a CSI report to the network access node 300 based on the predictability metric. The CSI report indicates a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric. In other words, the client device 100 uses the predictability metric for the set of coefficients of the first CSI quantity to determine which coefficient of the first CSI quantity to include in the CSI report to the network access node 300. The subset of coefficients of the first CSI quantity may comprise one or more of the coefficients in the set of coefficients of the first CSI quantity. The CSI report may further indicate other subsets of coefficients of the first CSI quantity, i.e., multiple subsets of coefficients of the first CSI quantity. The client device 100 may determine to include all the coefficients of the first CSI quantity, e.g., as configured to be reported, or only some the coefficients of the first CSI quantity in the CSI report to the network access node 300.

The predictability metric may indicate the importance of the set of coefficients of first CSI quantity e.g., for training of a ML model. The predictability metric may e.g., indicate whether the set of coefficients of first CSI quantity are predictable by the network access node 300 or not. Based on the predictability metric, the client device 100 may then determine whether the set of coefficients of the first CSI quantity is important to report to the network access node 300 or can be omitted without any loss of information. In embodiments, the client device 100 may further determine to not transmit the CSI report to the network access node 300. For example, if the predictability metric indicates that the set of coefficients of the first CSI quantity do not provide any new or valuable information, the client device 100 may omit transmitting the CSI report to the network access node 300. In this way, the client device 100 may omit or skip transmitting a scheduled CSI report.

In embodiments, the client device 100 may divide the coefficients of the first CSI quantity into subsets and give each subset of coefficients a priority level based on the predictability metric. The subset of coefficients of the first CSI quantity may then be included in the CSI report based on their respective priority level. When there are not enough uplink resources to report all the coefficients of the first CSI quantity, the client device 100 may then determine to omit or drop subsets of coefficients of the first CSI quantity with low priority and only include subsets of coefficients of the first CSI quantity with high priority in the CSI report. The client device 100 may hence determine to omit or drop parts of the first CSI quantity from the CSI report based on the predictability metric.

Furthermore, the client device 100 may change a reporting format of the subset of coefficients of first CSI quantity in the CSI report, e.g., report the subset of coefficients of the first CSI quantity in a format different to a configured reporting format. The first CSI quantity may be configured to be reported in a first reporting format and the subset of coefficients of the first CSI quantity may be indicated in the CSI report in a second reporting format different from the first reporting format. The second reporting format may be associated with a smaller bit width than the first reporting format and hence reduce the size of the CSI report indicating the subset of coefficients of the first CSI quantity in the second reporting format.

The client device 100 may use a criterion for the predictability metric when determining which coefficients of the first CSI quantity to include in the CSI report, i.e., when determining the subset of coefficients of the first CSI quantity to be indicated in the CSI report. Thus, the CSI report may indicate the subset of coefficients of the first CSI quantity when the predictability metric fulfills a criterion. The client device 100 may further transmit the CSI report to the network access node 300 when the predictability metric fulfills a criterion. The criterion for the predictability metric may be a threshold value for the predictability metric or similar. For example, a threshold value associated with the mentioned distance, correlation, performance or divergence. The client device 100 may e.g., determine to include the set of coefficients of the first CSI quantity in the CSI report when a distance between the set of coefficients of the first CSI quantity and the set of coefficients of the second CSI quantity is above a threshold value.

The criterion for the predictability metric may be pre-configured in the client device 100 or obtained from the network access node 300, e.g., in the second control signal 520. In the latter case, the client device 100 may determine which coefficients of the first CSI quantity to include in the CSI report and/or whether to transmit the CSI report based on an evaluation of the criterion for the predictability metric indicated in the second control signal 520.

The client device 100 may in addition to the predictability metric consider the amount of uplink resources available for CSI reporting, as well as the size of the CSI report and/or the subset of coefficients, when determining the CSI report. The client device 100 may hence in embodiments determine the CSI report and/or transmit the CSI report to the network access node 300 based on one or more of an uplink resource for the CSI report, a bit width of the CSI report and a bit width of the subset of coefficients of the first CSI quantity. In this way, the CSI report may be adapted to the available uplink resource for the CSI report such that the coefficients of the first CSI quantity providing the highest value to the network access node 300 may be included in the CSI report.

When the client device 100 has determined the subset of coefficients of the first CSI quantity to indicate in the CSI report, the client device 100 transmits the CSI report to the network access node 300, as shown in Fig. 6. The network access node 300 receive the CSI report from the client device 100 and hence the subset of coefficients of the first CSI quantity for the client device 100 indicated in the CSI report. As the subset of coefficients of the first CSI quantity has been determined from the set of coefficients of the first CSI quantity based on the predictability metric for the set of coefficients of the first CSI quantity, the subset of coefficients of the first CSI quantity provides valuable CSI information to the network access node 300. The network access node 300 may use the subset of coefficients of the first CSI quantity to e.g., train an ML model, interpolate or extrapolate in order to predict the rest of the coefficients of the first CSI quantity, determine precoders or beams for downlink transmission or reception, as input to a ML model that predicts the rest of the coefficients of the first CSI quantity or future values for these coefficients.

The client device 100 may further transmit a first control signal 510 to the network access node 300, as shown in optional step III in Fig. 6. The first control signal 510 indicates that the CSI report is based on the predictability metric and/or a format of the CSI report. With the first control signal 510, the client device 100 may inform the network access node 300 that the CSI report has been determined and/or transmitted based on the predictability metric, i.e., that parts of the first CSI quantity may have been removed. The client device 100 may further use the first control signal 510 to inform the network access node 300 of the format of the CSI report such that the CSI report can be properly decoded at the network access node 300.

The network access node 300 receives the first control signal 510 from the client device 100 and hence the indication that the CSI report is based on a predictability metric for the set of coefficients of the first CSI quantity and/or a reporting format for the CSI report. The network access node 300 may use the information obtained from the first control signal 510 when processing the CSI report received from the client device 100. The network access node 300 may e.g., use the indicated reporting format for the CSI report when decoding the CSI report. In this way, the network access node 300 may be able to decode the subset of coefficients of the first CSI quantity even if the reporting format of the subset of coefficients of the first CSI quantity has been changed from a configured reporting format, i.e., a format the network access node 300 is expecting to receive coefficients of the first CSI quantity in.

Fig. 8 shows physical resource usage for CSI reporting according to an embodiment of the invention. By using the predictability metric, a more effective usage of physical resources can be achieved. The client device 100 can reduce the size of its CSI reports without losing valuable information. With reference to Fig. 8, the client device 100 is configured to periodically receive CSI reference signals CSI-RS from the network access node 300 and transmit one or more CSI reports in response. The size of the CSI report boxes in Fig. 8 indicates the size of the CSI report(s) transmitted by the client device 100. At a first time instance Tl, the client device 100 does not omit any information and sends the full CSI report as configured. At a second T2 and fourth time instance T4, the client device 100 omits parts of the CSI report based on the predictability metric and only indicates a subset of coefficients of the first CSI quantity in the CSI report. The CSI reports transmitted at the second T2 and the fourth time instance T4 are hence smaller than the CSI report transmitted at the first time instance Tl, as shown in Fig. 8. At a third time instance T3, no CSI report is transmitted in the schedule uplink resource. The client device 100 may e.g., have determined that the coefficients of the first CSI quantity in the CSI report are all predictable by the network access node 300 and hence determined to not transmit the CSI report. When CSI reporting is performed based on the predictability metric according to the invention, physical resources may hence be saved.

The client device herein may be denoted as a user device, a user equipment (UE), a mobile station, an internet of things (loT) device, a sensor device, a wireless terminal and/or a mobile terminal, and is enabled to communicate wirelessly in a wireless communication system, sometimes also referred to as a cellular radio system. The UEs may further be referred to as mobile telephones, cellular telephones, computer tablets or laptops with wireless capability. The UEs in this context may be, for example, portable, pocket-storable, hand-held, computer- comprised, or vehicle-mounted mobile devices, enabled to communicate voice and/or data, via a radio access network (RAN), with another communication entity, such as another receiver or a server. The UE may further be a station, which is any device that contains an IEEE 802.11- conformant media access control (MAC) and physical layer (PHY) interface to the wireless medium (WM). The UE may be configured for communication in 3GPP related long term evolution (LTE), LTE-advanced, fifth generation (5G) wireless systems, such as new radio (NR), and their evolutions, as well as in IEEE related Wi-Fi, worldwide interoperability for microwave access (WiMAX) and their evolutions.

The network access node herein may also be denoted as a radio network access node, an access network access node, an access point (AP), or a base station (BS), e.g., a radio base station (RBS), which in some networks may be referred to as transmitter, “gNB”, “gNodeB”, “eNB”, “eNodeB”, “NodeB” or “B node”, depending on the standard, technology and terminology used. The radio network access nodes may be of different classes or types such as e.g., macro eNodeB, home eNodeB or pico base station, based on transmission power and thereby the cell size. The radio network access node may further be a station, which is any device that contains an IEEE 802.11-conformant MAC and PHY interface to the WM. The radio network access node may be configured for communication in 3GPP related LTE, LTE-advanced, 5G wireless systems, such as NR and their evolutions, as well as in IEEE related Wi-Fi, WiMAX and their evolutions.

Furthermore, any method according to embodiments of the invention may be implemented in a computer program, having code means, which when run by processing means causes the processing means to execute the steps of the method. The computer program is included in a computer readable medium of a computer program product. The computer readable medium may comprise essentially any memory, such as previously mentioned a ROM, a PROM, an EPROM, a flash memory, an EEPROM, or a hard disk drive.

Moreover, it should be realized that the client device and the network access node comprise the necessary communication capabilities in the form of e.g., functions, means, units, elements, etc., for performing or implementing embodiments of the invention. Examples of other such means, units, elements and functions are: processors, memory, buffers, control logic, encoders, decoders, rate matchers, de-rate matchers, mapping units, multipliers, decision units, selecting units, switches, interleavers, de-interleavers, modulators, demodulators, inputs, outputs, antennas, amplifiers, receiver units, transmitter units, DSPs, TCM encoder, TCM decoder, power supply units, power feeders, communication interfaces, communication protocols, etc. which are suitably arranged together for performing the solution.

Therefore, the processor(s) of the client device and the network access node may comprise, e.g., one or more instances of a CPU, a processing unit, a processing circuit, a processor, an ASIC, a microprocessor, or other processing logic that may interpret and execute instructions. The expression “processor” may thus represent a processing circuitry comprising a plurality of processing circuits, such as e.g., any, some or all of the ones mentioned above. The processing circuitry may further perform data processing functions for inputting, outputting, and processing of data comprising data buffering and device control functions, such as call processing control, user interface control, or the like. Finally, it should be understood that the invention is not limited to the embodiments described above, but also relates to and incorporates all embodiments within the scope of the appended independent claims.

Claims

1. A client device (100) for a communication system (500), the client device (100) being configured to: determine a predictability metric for a set of coefficients of a first channel state information, CSI, quantity for the client device (100) based on the first CSI quantity and a second CSI quantity; and transmit a CSI report to a network access node (300), the CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

2. The client device (100) according to claim 1, wherein the predictability metric is based on a distance between the set of coefficients of the first CSI quantity and a set of coefficients of the second CSI quantity.

3. The client device (100) according to claim 1, wherein the predictability metric is based on a correlation between the set of coefficients of the first CSI quantity and a set of coefficients of the second CSI quantity.

4. The client device (100) according to any one of claim 1 to 3, wherein the predictability metric is based on a training performance and/or an inference performance for a machine learning model associated with the first CSI quantity and/or the second CSI quantity.

5. The client device (100) according to any one of claim 1 to 3, wherein the predictability metric is based on a divergence between an estimated distribution based on the first CSI quantity and a reference distribution based on the second CSI quantity.

6. The client device (100) according to any one of the preceding claims, wherein the first CSI quantity and the second CSI quantity are based on measurements of reference signals transmitted at a first time instance; or the first CSI quantity is based on measurements of reference signals transmitted at the first time instance and the second CSI quantity is based on measurements of reference signals transmitted at a second time instance; or the first CSI quantity is based on measurements of reference signals transmitted at the first time instance and the second CSI quantity is based on reference measurements.

7. The client device (100) according to any one of the preceding claims, wherein the CSI report indicates the subset of coefficients of the first CSI quantity when the predictability metric fulfills a criterion.

8. The client device (100) according to any one of the preceding claims, configured to: transmit the CSI report to the network access node (300) based on one or more of an uplink resource for the CSI report, a bit width of the CSI report and a bit width of the subset of coefficients of the first CSI quantity.

9. The client device (100) according to any one of the preceding claims, configured to: transmit a first control signal (510) to the network access node (300), the first control signal (510) indicating that the CSI report is based on the predictability metric and/or a reporting format for the CSI report.

10. The client device (100) according to any one of the preceding claims, configured to: receive a second control signal (520) from the network access node (300), the second control signal (520) indicating a measurement configuration for the predictability metric and/or a format of the predictability metric; determine the predictability metric based on the second control signal (520).

11. A network access node (300) for a communication system (500), the network access node (300) being configured to: receive a CSI report from a client device (100), the CSI report indicating a subset of coefficients of a first CSI quantity for the client device (100), the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity.

12. The network access node (300) according to claim 11, configured to: receive a first control signal (510) from the client device (100), the first control signal (510) indicating that the CSI report is based on a predictability metric for the set of coefficients of the first CSI quantity and/or a reporting format for the CSI report.

13. The network access node (300) according to claim 11 or 12, configured to: transmit a second control signal (520) to the client device (100), the second control signal (520) indicating a measurement configuration for the predictability metric and/or a format of the predictability metric.

14. A method (200) for a client device (100), the method (200) comprising: determining (202) a predictability metric for a set of coefficients of a first channel state information, CSI, quantity for the client device (100) based on the first CSI quantity and a second CSI quantity; and transmitting (204) a CSI report to a network access node (300), the CSI report indicating a subset of coefficients of the first CSI quantity determined from the set of coefficients of the first CSI quantity based on the predictability metric.

15. A method (400) for a network access node (300), the method (400) comprising: receiving (402) a CSI report from a client device (100), the CSI report indicating a subset of coefficients of a first CSI quantity for the client device (100), the subset of coefficients of the first CSI quantity being determined from a set of coefficients of the first CSI quantity based on a predictability metric for the set of coefficients of the first CSI quantity.

16. A computer program with a program code for performing a method according to claim 14 or 15 when the computer program runs on a computer.