Apparatus, method and computer program

Abstract

There is provided an apparatus (400) for channel state information feedback in a cellular communication network, the apparatus comprising: means for determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, the channel state information feedback size threshold defines the maximum bit number for channel state information feedback, and the optimal pair is optimal for the user equipment; means for assigning, to the user equipment, a pair comprising a channel state information reference signal pattern and a channel state information feedback size threshold defining the maximum number of bits for channel state information feedback, based on at least one of cell load information and cell performance information and the determined optimal pair; and means for sending information indicating the assigned pair to the user equipment.

Description

Technical Field

[0001] The present application relates to a method, an apparatus and a computer program, and particularly but not exclusively to channel state information feedback. Background Art

[0002] A communication system can be viewed as a facility for realizing a communication session between two or more entities such as a user terminal, a base station and / or other nodes by providing a carrier between various entities involved in a communication path. For example, a communication system can be provided by means of a communication network and one or more compatible communication devices. A communication session may include, for example, data communications for carrying communications such as voice, video, email (email), text messages, multimedia and / or content data. Non-limiting examples of the services provided include two-way or multi-way calls, data communications or multimedia services, and access to data network systems such as the Internet. In a wireless communication system, at least a portion of a communication session between at least two stations occurs over a wireless link. Examples of wireless systems include public land mobile networks (PLMNs), satellite-based communication systems, and different wireless local area networks, such as wireless local area networks (WLANs). Some wireless systems may be divided into cells, and are therefore often referred to as cellular systems.

[0003] A user may access the communication system by means of an appropriate communication device or terminal. The communication device of a user may be referred to as user equipment (UE) or user device.

[0004] A communication system and associated equipment typically operate according to a given standard or specification that states what the various entities associated with the system are permitted to do and how it should be achieved. The communication protocols and / or parameters used for the connection are also typically defined. An example of a communication system is UTRAN (3G radio). Other examples of communication systems are the Long Term Evolution (LTE) of the Universal Mobile Telecommunications System (UMTS) radio access technology and the so-called 5G or New Radio (NR) networks. NR is being standardized by the 3rd Generation Partnership Project (3GPP). Other examples of communication systems include Advanced 5G (NR Release 18 and above) and 6G.

[0005] It is known that modern communication systems are characterized by channel state information (CSI) feedback overhead. CSI feedback enhancement (such as CSI feedback compression or CSI overhead reduction) is used to improve the performance of CSI feedback overhead. Summary of the invention

[0006] According to a first aspect, an apparatus for channel state information feedback in a cellular communication network is disclosed, comprising: a component for determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; a component for assigning a pair comprising a channel state information reference signal pattern and a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and a component for sending information indicating the assigned pair to the user equipment.

[0007] The optimal pair refers to a channel state information reference signal (CSI-RS) pattern and a channel state information (CSI) feedback size threshold pair that together can achieve a specific CSI recovery performance (eg, mean square error (MSE)) that the UE expects to achieve.

[0008] According to some examples, the apparatus further includes means for receiving channel state information feedback bits from the user equipment based on the assigned pairs.

[0009] According to some examples, determining the optimal pair includes: determining multiple pairs including a channel state information reference signal pattern and a channel state information feedback size threshold based on an estimation or measurement of radio channel characteristics; and selecting the optimal pair from among the determined multiple pairs.

[0010] According to some examples, the plurality of pairs is determined based on using a default channel state information feedback size threshold and a channel state information reference signal pattern pair.

[0011] According to some examples, the components for determining the optimal pair include: determining an estimated channel recovery mean square error of the feedback recovery method associated with each of the multiple pairs; determining an estimated overhead associated with each of the multiple pairs; and selecting the optimal pair from the multiple pairs based on at least one of: the determined channel recovery mean square error; and the determined overhead.

[0012] According to some examples, the overhead associated with each of the plurality of pairs includes at least one of: channel state information feedback overhead; and channel state information reference signal pilot resource element overhead.

[0013] According to some examples, the apparatus further includes means for determining, using at least one neural network, an estimated channel recovery mean square error and overhead associated with each of the plurality of pairs.

[0014] According to some examples, the at least one neural network is a convolutional neural network.

[0015] According to some examples, the apparatus further comprises means for training and testing the at least one neural network using a channel, wherein the channel is obtained by the apparatus using uplink-downlink reciprocity.

[0016] According to some examples, the apparatus further comprises means for constructing a plurality of neural networks, wherein each of the plurality of neural networks is used to determine an estimated channel recovery mean square error and an estimated overhead associated with one of the plurality of pairs.

[0017] According to some examples, the apparatus further includes means for training and testing multiple neural networks in parallel.

[0018] According to some examples, the device includes a component for determining a feedback recovery method and sending information about the feedback recovery method to the user equipment, wherein the feedback recovery method includes information indicating at least one of: how the device recovers channel state information feedback bits; and how the device reconstructs a channel.

[0019] According to some examples, the apparatus further comprises a component for constructing a separate machine learning model on top of the at least one neural network, wherein the machine learning model learns the structure of the at least one neural network.

[0020] According to some examples, the apparatus includes an access node.

[0021] According to a second aspect, a device is provided, comprising at least one processor; and at least one memory storing instructions, which, when executed by the at least one processor, causes the device to at least perform: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal pattern, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assigning to the user equipment a pair including a channel state information reference signal pattern and a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback based on at least one of cell load information and cell performance information and the determined optimal pair; and sending information indicating the assigned pair to the user equipment.

[0022] According to a third aspect, there is provided an apparatus for channel state information feedback in a cellular communication network, the apparatus comprising: a component for receiving information indicating assigned pairs from an access node, the assigned pairs comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and a component for determining a channel state information feedback bit based on the assigned pairs.

[0023] According to some examples, the apparatus further includes means for determining channel state information feedback bits using at least one neural network.

[0024] According to some examples, the apparatus further includes means for sending the determined channel state information feedback bits to the access node.

[0025] According to some examples, the apparatus includes user equipment.

[0026] According to a fourth aspect, a device is provided, comprising at least one processor; and at least one memory storing instructions, which, when executed by the at least one processor, causes the device to at least perform: receiving information indicating an assigned pair from an access node, the assigned pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal pattern; and determining a channel state information feedback bit based on the assigned pair.

[0027] According to a fifth aspect, a device for channel state information feedback in a cellular communication network is provided, the device comprising: a component for determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal pattern, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the device; a component for sending information indicating the determined optimal pair to an access node; a component for receiving information indicating an assigned pair from the access node, the assigned pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal pattern; and a component for determining a channel state information feedback bit based on the assigned pair.

[0028] According to some examples, the apparatus further includes means for determining channel state information feedback bits using at least one neural network.

[0029] According to some examples, the apparatus further includes means for sending the determined channel state information feedback bits to the access node.

[0030] According to some examples, determining the optimal pair includes: determining multiple pairs including a channel state information reference signal pattern and a channel state information feedback size threshold based on estimation or measurement of radio channel characteristics; and selecting the optimal pair from among the determined multiple pairs.

[0031] According to some examples, the plurality of pairs is determined based on using a default channel state information feedback size threshold and a channel state information reference signal pattern pair.

[0032] According to some examples, the components for determining the optimal pair include: determining an estimated channel recovery mean square error of the feedback recovery method associated with each of the multiple pairs; determining an estimated overhead associated with each of the multiple pairs; and selecting the optimal pair from the multiple pairs based on at least one of: the determined channel recovery mean square error; and the determined overhead.

[0033] According to some examples, the overhead associated with each of the plurality of pairs includes at least one of: channel state information feedback overhead; and channel state information reference signal pilot resource element overhead.

[0034] According to some examples, the apparatus further includes means for determining, using at least one neural network, an estimated channel recovery mean square error and overhead associated with each of the plurality of pairs.

[0035] According to some examples, the at least one neural network is a convolutional neural network.

[0036] According to some examples, the apparatus further comprises means for training and testing the at least one neural network using the channel determined by the data-aided channel estimation.

[0037] According to some examples, the apparatus further comprises means for constructing a plurality of neural networks, wherein each of the plurality of neural networks is used to determine an estimated channel recovery mean square error and an estimated overhead associated with one of the plurality of pairs.

[0038] According to some examples, the apparatus further includes means for training and testing multiple neural networks in parallel.

[0039] According to some examples, the apparatus further comprises a component for receiving information on a feedback recovery method from an access node, wherein the information on the feedback recovery method comprises information indicating at least one of: how the access node recovers channel state information feedback bits; and how the access node reconstructs a channel.

[0040] According to some examples, the apparatus further comprises a component for constructing a separate machine learning model on top of the at least one neural network, wherein the machine learning model learns the structure of the at least one neural network.

[0041] According to some examples, the apparatus includes user equipment.

[0042] According to a sixth aspect, a device is provided, comprising at least one processor; and at least one memory storing instructions, which, when executed by the at least one processor, causes the device to at least perform: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal pattern, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the device; transmitting information indicating the determined optimal pair to an access node; receiving information indicating an assigned pair from the access node, the assigned pair including a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pair.

[0043] According to a seventh aspect, a device for channel state information feedback in a cellular communication network is provided, the device comprising: a component for receiving an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; a component for assigning a pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and a component for sending information indicating the assigned pair to the user equipment.

[0044] According to some examples, the apparatus further includes means for receiving channel state information feedback bits from a user equipment based on the assigned pairs.

[0045] According to some examples, the apparatus includes an access node.

[0046] According to an eighth aspect, a device is provided, comprising at least one processor; and at least one memory storing instructions, which, when executed by the at least one processor, causes the device to at least perform: receiving an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assigning a pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and transmitting information indicating the assigned pair to the user equipment.

[0047] According to a ninth aspect, a method is provided, comprising: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assigning a pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and sending information indicating the assigned pair to the user equipment.

[0048] According to some examples, the method further includes receiving channel state information feedback bits from the user equipment based on the assigned pairs.

[0049] According to some examples, determining the optimal pair includes: determining multiple pairs including a channel state information reference signal pattern and a channel state information feedback size threshold based on estimation or measurement of radio channel characteristics; and selecting the optimal pair from among the determined multiple pairs.

[0050] According to some examples, the plurality of pairs is determined based on using a default channel state information feedback size threshold and a channel state information reference signal pattern pair.

[0051] According to some examples, determining the optimal pair includes: determining an estimated channel recovery mean square error of the feedback recovery method associated with each of the multiple pairs; determining an estimated overhead associated with each of the multiple pairs; and selecting the optimal pair from the multiple pairs based on at least one of: the determined channel recovery mean square error; and the determined overhead.

[0052] According to some examples, the overhead associated with each of the plurality of pairs includes at least one of: channel state information feedback overhead; and channel state information reference signal pilot resource element overhead.

[0053] According to some examples, the method further includes determining, using at least one neural network, an estimated channel recovery mean square error and overhead associated with each of the plurality of pairs.

[0054] According to some examples, the at least one neural network is a convolutional neural network.

[0055] According to some examples, the method further includes: training and testing the at least one neural network using a channel, wherein the channel is obtained by the device using uplink-downlink reciprocity.

[0056] According to some examples, the method further includes constructing a plurality of neural networks, wherein each neural network of the plurality of neural networks is used to determine an estimated channel recovery mean square error and an estimated overhead associated with one of the plurality of pairs.

[0057] According to some examples, the method also includes training and testing the plurality of neural networks in parallel.

[0058] According to some examples, the method includes: determining the feedback recovery method and sending information of the feedback recovery method to the user equipment, wherein the feedback recovery method includes information indicating at least one of the following: how the access node recovers the channel state information feedback bit; and how the access node reconstructs the channel.

[0059] According to some examples, the method also includes constructing a separate machine learning model on top of the at least one neural network, wherein the machine learning model learns the structure of the at least one neural network.

[0060] According to some examples, the method is performed by an access node.

[0061] According to the tenth aspect, a method is provided, comprising: receiving information indicating an assigned pair from an access node, the assigned pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal pattern; and determining a channel state information feedback bit based on the assigned pair.

[0062] According to some examples, the method further includes determining, using at least one neural network, a channel state information feedback bit.

[0063] According to some examples, the method further includes sending the determined channel state information feedback bits to the access node.

[0064] According to some examples, the method is performed by a user device.

[0065] According to an eleventh aspect, a method is provided, comprising: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; sending information indicating the determined optimal pair to an access node; receiving information indicating an assigned pair from the access node, wherein the assigned pair includes a channel state information feedback size threshold and a channel state information reference signal mode defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pair.

[0066] According to some examples, the method further includes determining, using at least one neural network, a channel state information feedback bit.

[0067] According to some examples, the method further includes sending the determined channel state information feedback bits to the access node.

[0068] According to some examples, determining the optimal pair includes: determining multiple pairs including a channel state information reference signal pattern and a channel state information feedback size threshold based on estimation or measurement of radio channel characteristics; and selecting the optimal pair from among the determined multiple pairs.

[0069] According to some examples, the plurality of pairs is determined based on using a default channel state information feedback size threshold and a channel state information reference signal pattern pair.

[0070] According to some examples, determining the optimal pair includes: determining an estimated channel recovery mean square error of the feedback recovery method associated with each of the multiple pairs; determining an estimated overhead associated with each of the multiple pairs; and selecting the optimal pair from the multiple pairs based on at least one of: the determined channel recovery mean square error; and the determined overhead.

[0071] According to some examples, the overhead associated with each of the plurality of pairs includes at least one of: channel state information feedback overhead; and channel state information reference signal pilot resource element overhead.

[0072] According to some examples, the method further includes determining, using at least one neural network, an estimated channel recovery mean square error and overhead associated with each of the plurality of pairs.

[0073] According to some examples, the at least one neural network is a convolutional neural network.

[0074] According to some examples, the method further includes training and testing the at least one neural network using the channel determined by the data-aided channel estimation.

[0075] According to some examples, the method further includes constructing a plurality of neural networks, wherein each neural network of the plurality of neural networks is used to determine an estimated channel recovery mean square error and an estimated overhead associated with one of the plurality of pairs.

[0076] According to some examples, the method also includes training and testing the plurality of neural networks in parallel.

[0077] According to some examples, the method also includes receiving information on a feedback recovery method from the access node, wherein the information on the feedback recovery method includes information indicating at least one of: how the access node recovers channel state information feedback bits; and how the access node reconstructs the channel.

[0078] According to some examples, the method also includes constructing a separate machine learning model on top of the at least one neural network, wherein the machine learning model learns the structure of the at least one neural network.

[0079] According to some examples, the method is performed by a user device.

[0080] According to the twelfth aspect, a method is provided, comprising: receiving an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assigning a pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and sending information indicating the assigned pair to the user equipment.

[0081] According to some examples, the method further includes receiving channel state information feedback bits from the user equipment based on the assigned pairs.

[0082] According to some examples, the method is performed by an access node.

[0083] According to the thirteenth aspect, a computer program is provided, comprising instructions for causing an apparatus to perform at least the following operations: determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assigning a pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and sending information indicating the assigned pair to the user equipment.

[0084] According to some examples, the computer program includes instructions for causing the apparatus to: receive channel state information feedback bits from a user equipment based on the assigned pairs.

[0085] According to some examples, determining the optimal pair includes: determining multiple pairs including a channel state information reference signal pattern and a channel state information feedback size threshold based on an estimation or measurement of radio channel characteristics; and selecting the optimal pair from among the determined multiple pairs.

[0086] According to some examples, the plurality of pairs is determined based on using a default channel state information feedback size threshold and a channel state information reference signal pattern pair.

[0087] According to some examples, determining the optimal pair includes: determining an estimated channel recovery mean square error of the feedback recovery method associated with each of the multiple pairs; determining an estimated overhead associated with each of the multiple pairs; and selecting the optimal pair from the multiple pairs based on at least one of: the determined channel recovery mean square error; and the determined overhead.

[0088] According to some examples, the overhead associated with each of the plurality of pairs includes at least one of: channel state information feedback overhead; and channel state information reference signal pilot resource element overhead.

[0089] According to some examples, the computer program includes instructions for causing the apparatus to determine, using at least one neural network, an estimated channel recovery mean square error and overhead associated with each of the plurality of pairs.

[0090] According to some examples, the at least one neural network is a convolutional neural network.

[0091] According to some examples, the computer program includes instructions for causing the apparatus to perform the following operations: training and testing the at least one neural network using a channel, wherein the channel is obtained by the apparatus using uplink-downlink reciprocity.

[0092] According to some examples, the computer program includes instructions for causing the apparatus to perform the following operations: construct a plurality of neural networks, wherein each neural network of the plurality of neural networks is used to determine an estimated channel recovery mean square error and an estimated overhead associated with one of the plurality of pairs.

[0093] According to some examples, the computer program includes instructions for causing the apparatus to: train and test the plurality of neural networks in parallel.

[0094] According to some examples, the computer program includes instructions for causing the device to perform the following operations: determine the feedback recovery method and transmit information about the feedback recovery method to the user equipment, wherein the feedback recovery method includes information indicating at least one of the following: how the device recovers channel state information feedback bits; and how the device reconstructs the channel.

[0095] According to some examples, the computer program includes instructions for causing the apparatus to perform the following operations: construct a separate machine learning model on top of the at least one neural network, wherein the machine learning model learns the structure of the at least one neural network.

[0096] According to some examples, the apparatus includes an access node.

[0097] According to a fourteenth aspect, a computer program is provided, comprising instructions for causing a device to perform at least the following operations: receiving information indicating an assigned pair from an access node, the assigned pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pair.

[0098] According to some examples, the computer program includes instructions for causing the apparatus to perform the following operations: determine the channel state information feedback bit using at least one neural network.

[0099] According to some examples, the computer program comprises instructions for causing the apparatus to: transmit the determined channel state information feedback bit to the access node.

[0100] According to some examples, the apparatus includes user equipment.

[0101] According to the fifteenth aspect, a computer program is provided, comprising instructions for causing a device to perform at least the following operations: determine an optimal pair including a channel state information feedback size threshold and a channel state information reference signal pattern, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the device; send information indicating the determined optimal pair to an access node; receive information indicating an assigned pair from the access node, wherein the assigned pair includes a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determine a channel state information feedback bit based on the assigned pair.

[0102] According to some examples, the computer program includes instructions for causing the apparatus to perform the following operations: determine channel state information feedback bits using at least one neural network.

[0103] According to some examples, the computer program comprises instructions for causing the apparatus to perform the following operations: sending the determined channel state information feedback bit to the access node.

[0104] According to some examples, determining the optimal pair includes: determining multiple pairs including a channel state information reference signal pattern and a channel state information feedback size threshold based on an estimation or measurement of radio channel characteristics; and selecting the optimal pair from among the determined multiple pairs.

[0105] According to some examples, the plurality of pairs is determined based on using a default channel state information feedback size threshold and a channel state information reference signal pattern pair.

[0106] According to some examples, determining the optimal pair includes: determining an estimated channel recovery mean square error of the feedback recovery method associated with each of the multiple pairs; determining an estimated overhead associated with each of the multiple pairs; and selecting the optimal pair from the multiple pairs based on at least one of: the determined channel recovery mean square error; and the determined overhead.

[0107] According to some examples, the overhead associated with each of the plurality of pairs includes at least one of: channel state information feedback overhead; and channel state information reference signal pilot resource element overhead.

[0108] According to some examples, the computer program includes instructions for causing the apparatus to determine, using at least one neural network, an estimated channel recovery mean square error and overhead associated with each of the plurality of pairs.

[0109] According to some examples, the at least one neural network is a convolutional neural network.

[0110] According to some examples, the computer program includes instructions for causing the apparatus to train and test the at least one neural network using the channel determined by the data-aided channel estimation.

[0111] According to some examples, the computer program includes instructions for causing the apparatus to perform the following operations: construct a plurality of neural networks, wherein each neural network of the plurality of neural networks is used to determine an estimated channel recovery mean square error and an estimated overhead associated with one of the plurality of pairs.

[0112] According to some examples, the computer program includes instructions for causing the apparatus to: train and test the plurality of neural networks in parallel.

[0113] According to some examples, the computer program includes instructions for causing the device to perform the following operations: receiving information about the feedback recovery method from the access node, wherein the information about the feedback recovery method includes information indicating at least one of the following: how the access node recovers channel state information feedback bits; and how the access node reconstructs the channel.

[0114] According to some examples, the computer program includes instructions for causing the apparatus to perform the following operations: construct a separate machine learning model on top of the at least one neural network, wherein the machine learning model learns the structure of the at least one neural network.

[0115] According to some examples, the apparatus includes user equipment.

[0116] According to the sixteenth aspect, a computer program is provided, comprising instructions for causing an apparatus to perform at least the following operations: receiving an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assigning a pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and transmitting information indicating the assigned pair to the user equipment.

[0117] According to some examples, the computer program includes instructions for causing the apparatus to: receive channel state information feedback bits from the user equipment based on the assigned pairs.

[0118] According to some examples, the apparatus includes an access node.

[0119] According to the seventeenth aspect, a computer program is provided, comprising instructions stored thereon for performing at least the following operations: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assigning a pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and transmitting information indicating the assigned pair to the user equipment.

[0120] According to the eighteenth aspect, a computer program is provided, comprising instructions stored thereon for performing at least the following operations: receiving information indicating an assigned pair from an access node, the assigned pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pair.

[0121] According to the nineteenth aspect, a computer program is provided, comprising instructions stored thereon for performing at least the following operations: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; sending information indicating the determined optimal pair to an access node; receiving information indicating an assigned pair from the access node, wherein the assigned pair includes a channel state information feedback size threshold and a channel state information reference signal mode defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pair.

[0122] According to the twentieth aspect, a computer program is provided, comprising instructions stored thereon for performing at least the following operations: receiving an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assigning a pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and sending information indicating the assigned pair to the user equipment.

[0123] According to the twenty-first aspect, a non-transitory computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: determine an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assign a pair comprising a channel state information feedback size threshold and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback; and transmit information indicating the assigned pair to the user equipment.

[0124] According to the twenty-second aspect, a non-volatile computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: receiving information indicating assigned pairs from an access node, the assigned pairs comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pairs.

[0125] According to the twenty-third aspect, a non-volatile computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: determine an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the apparatus; send information indicating the determined optimal pair to an access node; receive information indicating an assigned pair from the access node, wherein the assigned pair comprises a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determine a channel state information feedback bit based on the assigned pair.

[0126] According to the twenty-fourth aspect, a non-transitory computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: receive an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assign a pair comprising a channel state information feedback size threshold and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback; and send information indicating the assigned pair to the user equipment.

[0127] According to the twenty-fifth aspect, a non-volatile computer-readable medium is provided, comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assigning a pair comprising a channel state information feedback size threshold and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback; and sending information indicating the assigned pair to the user equipment.

[0128] According to the twenty-sixth aspect, a non-volatile computer-readable medium is provided, comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: receiving information indicating assigned pairs from an access node, the assigned pairs comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pairs.

[0129] According to the twenty-seventh aspect, there is provided a non-volatile computer-readable medium comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; sending information indicating the determined optimal pair to an access node; receiving information indicating an assigned pair from the access node, the assigned pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pair.

[0130] According to the twenty-eighth aspect, a non-volatile computer-readable medium is provided, comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: receiving an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assigning to the user equipment a pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode based on at least one of cell load information and cell performance information and the determined optimal pair; and sending information indicating the assigned pair to the user equipment.

[0131] According to the twenty-ninth aspect, a computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: determine an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assign a pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and transmit information indicating the assigned pair to the user equipment.

[0132] According to the thirtieth aspect, there is provided a computer-readable medium comprising program instructions, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: receiving information indicating assigned pairs from an access node, the assigned pairs comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pairs.

[0133] According to the thirty-first aspect, a computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: determine an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the apparatus; send information indicating the determined optimal pair to an access node; receive information indicating an assigned pair from the access node, wherein the assigned pair comprises a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determine a channel state information feedback bit based on the assigned pair.

[0134] According to the thirty-second aspect, a computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following operations: receive an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assign a pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair; and send information indicating the assigned pair to the user equipment.

[0135] According to the thirty-third aspect, there is provided a computer-readable medium comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; assigning a pair comprising a channel state information feedback size threshold and a channel state information reference signal mode to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback; and sending information indicating the assigned pair to the user equipment.

[0136] According to a thirty-fourth aspect, there is provided a computer-readable medium comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: receiving information indicating assigned pairs from an access node, the assigned pairs comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pairs.

[0137] According to the thirty-fifth aspect, there is provided a computer-readable medium comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; sending information indicating the determined optimal pair to an access node; receiving information indicating an assigned pair from the access node, the assigned pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern defining a maximum number of bits for channel state information feedback; and determining a channel state information feedback bit based on the assigned pair.

[0138] According to the thirty-sixth aspect, there is provided a computer-readable medium comprising program instructions stored thereon, the program instructions being used to perform at least the following operations: receiving an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; assigning to the user equipment a pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode based on at least one of cell load information and cell performance information and the determined optimal pair; and transmitting information indicating the assigned pair to the user equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0139] Embodiments will now be described, by way of example only, with reference to the accompanying drawings, in which:

[0140] Figure 1 shows a representation of a network system according to some example embodiments;

[0141] Figure 2 is a signaling diagram according to an example embodiment.

[0142] Figure 3 is a signaling diagram according to an example embodiment.

[0143] Figure 4 is a signaling diagram according to an example embodiment.

[0144] Figure 5 is a signaling diagram according to an example embodiment.

[0145] Figure 6 shows a schematic representation of a control device according to some example embodiments;

[0146] Figure 7 shows a schematic representation of an apparatus according to some example embodiments;

[0147] Figure 8 is a flow chart of a method according to an example embodiment;

[0148] Fig. 9 is a flow chart of a method according to an example embodiment;

[0149] Fig.10 is a flow chart of a method according to an example embodiment;

[0150] Fig.11 is a flow chart of a method according to an example embodiment;

[0151] Fig.12 A schematic representation of a non-volatile storage medium storing instructions that, when executed by a processor, enable the processor to perform one or more steps of the method of some embodiments is shown. DETAILED DESCRIPTION

[0152] In the following, certain embodiments are explained with reference to a mobile communication device capable of communicating via a wireless cellular system and a mobile communication system serving such a mobile communication device. Figure 1 Certain general principles of wireless communication systems, their access systems and mobile communication devices are briefly explained to help understand the technology underlying the described examples.

[0153] Figure 1 A schematic representation of a 5G system (5GS) is shown provided by way of example of a communication system. The 5GS may include a terminal or user equipment (UE) 100, a 5G radio access network (5GRAN) 108 or a next generation radio access network (NG-RAN), a 5G core network (5GC) 110, one or more application functions (AFs) 106, and one or more data networks.

[0154] The 5G-RAN may include one or more gNodeB (gNB) distributed unit functions or one or more gNodeB (gNB) connected to one or more gNodeB (gNB) centralized unit functions.

[0155] 5GC may include the following entities: network slice selection function (NSSF); network exposure function 112; network repository function (NRF); policy control function (PCF); unified data management (UDM) 104; application function (AF) 106; authentication server function (AUSF) 114; access and mobility management function (AMF) 104; session management function (SMF) 116; user plane function 118; and data network 120. Figure 1 Also shown are various interfaces (N1, N2, etc.) that may be implemented between the various elements of the system.

[0156] Note that although some embodiments are described with respect to 5G networks, similar principles may be applied with respect to other networks and communication systems, such as 5G Advanced networks or 6G networks.

[0157] Type IICSI feedback overhead is an issue in networks and communication systems. CSI feedback compression and CSI overhead reduction are discussed in the RAN1 MIMO session. In some examples, to achieve good performance for Type IICSI feedback, hundreds of feedback bits may be sent, which uses uplink (UL) data bandwidth.

[0158] CSI feedback overhead reduction is included in 3GPP RAN1 AI / ML study item RP-213599 as a use case for performance, complexity and potential specification impact. CSI feedback enhancements such as overhead reduction, improved accuracy and prediction are described.

[0159] In some examples, to improve CSI feedback, the feedback bits can be reduced while keeping the CSI feedback performance the same, or the feedback bits can be kept the same while improving the CSI feedback performance. Both approaches can improve the tradeoff between CSI feedback overhead and CSI feedback performance. In some examples, the CSI feedback performance can be measured as the channel recovery mean square error (MSE) at the gNB.

[0160] Figure 2 FIG. 1 shows a conventional CSI feedback process modeled in two steps. In the conventional CSI feedback process, information can be transmitted between the UE and the BS. In some examples, information can be transmitted between the UE and the BS via the channel h UL and h DL Information is transferred between the BS and the UE. The CSI feedback process enables DL reconstruction at the BS. The process includes a channel estimation module and a CSI encoder module that process information at the UE, and a CSI decoder module and a CSI mode determination module that process information at the BS. These modules all involve high-complexity calculations. In some examples, report type IICSI feedback requires high-complexity calculations. Report type IICSI feedback may require calculation of eigendecomposition of multi-antenna channels. It is known that eigendecomposition is highly complex and resource-consuming.

[0161] In addition to the high complexity, the conventional two-step CSI feedback process may result in information loss according to the data processing theorem, and as such, the CSI feedback performance may be degraded. The data processing theorem states that data processing destroys information. At the UE, the channel estimation module may decompress the channel on a full transmission time interval physical resource block (TTI-PRB) resource grid with 12×14 resource elements (REs) from the channel on the pilot REs. The channel on the pilot REs may be based on the configured CSI-RS pattern. Interpolation may be used to recover the channel on the full TTI-PRB resource grid. The recovered channel is Figure 2 Medium , which represents the estimated DL channel at the UE. The CSI encoder module can compress the recovered channel on the full TTI-PRB resource grid into the CSI feedback container. The CSI encoder module compresses the recovered channel on the full TTI-PRB resource grid using eigendecomposition and calculates eigenvectors to simulate CSI feedback. For the traditional two-step CSI feedback process, the number of feedback bits can be implicitly configured with the number of feedback beams and the corresponding in-phase quantization bit width. At the BS, the CSI decoder module can decode the information from the CSI encoder module to produce a reconstructed DL channel. However, the two-step decompression-compression process may result in information loss compared to the one-step process.For example, the one-step process may directly perform processing from the channel on the pilot REs to the feedback bits.

[0162] Therefore, the conventional two-step CSI feedback process may be improved in view of the reduction in CSI feedback performance and computational complexity. For example, the CSI feedback performance may be improved by improving the channel recovery performance using a fixed number of feedback bits.

[0163] Embodiments of the present invention provide a new CSI feedback process. Specifically, a direct quantization CSI feedback process is provided. In some examples, the direct quantization CSI feedback process includes direct quantization of a channel on an RE pilot using a convolutional neural network (CNN). In some examples, the direct quantization CSI feedback process includes outputting feedback of quantized feedback bits.

[0164] Figure 3 Represents the ground truth CSI feedback E, the traditional CSI feedback E' and the proposed direct quantized CSI feedback This example should be understood as an example showing how direct quantized CSI feedback compares with true value CSI feedback and / or traditional CSI feedback process. Figure 3Simplified steps of the true value CSI feedback process S301, the traditional two-step CSI feedback process S302 and the direct quantization CSI feedback process S303 are shown.

[0165] S301 shows the true CSI feedback process. In this example, eigendecomposition is applied to the true channel on the full TTI-PRB resource grid 300. Eigendecomposition can be used to derive an eigenvector of the channel 300. The eigenvector can be used as a training and / or test label for the true CSI feedback E. In some examples, if the channel will be measured optimally without any error or loss, it can be understood that the true CSI feedback E represents the channel.

[0166] S302 shows a traditional two-step CSI feedback process. In some examples, traditional CSI feedback E' can be understood to represent CSI feedback obtained using a two-step CSI feedback process (e.g., a type IICSI feedback process). In this example, based on the configured CSI-RS pattern, the true value channel on the full TTI-PRB resource grid 300 is sampled using pilot RE 302. Interpolation can be used to restore the channel 300 to obtain the restored channel 304. As described above, this can correspond to a decompression step performed by a channel estimation module. The restored channel 304 is fed to the eigendecomposition module 306, and the eigenvector is calculated to simulate the CSI feedback bit E'. As described above, this can correspond to a compression step performed by a CSI encoder module.

[0167] S303 shows the direct quantization CSI feedback process. In some examples, the CSI feedback bits It can be understood as the CSI feedback bits obtained using the methods and processes described herein. For example, it can refer to directly quantizing the CSI feedback bits. Instead of using the two steps of channel estimation and eigendecomposition, these two steps are replaced by CNN 308. In some examples, CNN 308 is used to calculate the CSI feedback bits In some examples, CNN 308 is used to quantize the channel on RE pilot 302 to generate CSI feedback bits in one step. In some examples, the behavior of CNN 308 can be similar to traversing a direct quantization function. In some examples, the channel on RE pilot 302 can be based on the configured CSI-RS pattern. In some examples, the CSI feedback bits The number is based on a configured CSI feedback size threshold. In some examples, the CSI feedback size threshold defines a maximum number of bits for CSI feedback.

[0168] In some examples, the following model is used to test and train CNN 308:

[0169] Input = RE pilot

[0170] Output = CSI feedback bits

[0171] Label = CSI feedback bit E

[0172] Loss = E and MSE between

[0173] An example CNN implementation for the direct quantization CSI feedback process is shown below:

[0174] #Define the input placeholder as a tensor with shape input_shape. Think of this as your input image

[0175] X_input=Input(input_shape)

[0176] X=Conv3D(2,6, strides=(1,1,1), name='conv0', padding="same", use_bias=True)(X_input)#(None,12,14,8,2)

[0177] X=Conv3D(2,(4,7,8), strides=(3,7,1), name='conv0', padding="same", use_bias=True)(X_input)#(None,4,2,8,2)

[0178] X=Dense(2,activation='sigmoid',name='fc',use_bias=True)(X)#(None,4,2,8,2)

[0179] nam = 'directEigenMapping'

[0180] #Create the model. This creates your Keras model instance, which you will use to train / test the model.

[0181] model=Model(inputs=X_input,output=X,name=nam)

[0182] return model

[0183] In some examples, the direct quantization CSI feedback process is compared with the conventional two-step CSI feedback process using the following model:

[0184] Input = RE pilot

[0185] ·Compare = (E and MSE between E and E'

[0186] An example simulation MSE comparison between the direct quantization CSI feedback process and the traditional two-step CSI feedback process is shown below:

[0187] This is the end of the assistance and the MSE for traditional methods can be done here.

[0188] Period 1 / 10

[0189] 408 / 408[============] - 16s 38ms / step - loss: 0.1212 - mse: 0.1212 epoch 2 / 10

[0190] 408 / 408[============] -16s 39ms / step -loss:0.0657 -mse:0.0657 epoch 3 / 10

[0191] 408 / 408[============]-16s 40ms / step-loss: 0.0625-mse: 0.0624 epoch 4 / 10

[0192] 408 / 408[============] -16s 39ms / step -loss:0.0617 -mse:0.0617 epoch 5 / 10

[0193] 408 / 408[============]-16s 40ms / step-loss: 0.0613-mse: 0.0613 epoch 6 / 10

[0194] 408 / 408[============] - 16s 39ms / step - loss: 0.0611 - mse: 0.0611 epoch 7 / 10

[0195] 408 / 408[============]-16s 40ms / step-loss: 0.0610-mse: 0.0610 epoch 8 / 10

[0196] 408 / 408[============] - 16s 38ms / step - loss: 0.0609 - mse: 0.0609 epoch 9 / 10

[0197] 408 / 408[============] - 16s 39ms / step - loss: 0.0608 - mse: 0.0608 epoch 10 / 10

[0198] 408 / 408[============]-16s 39ms / step-loss: 0.0607-mse: 0.0607The mse of the direct quantization method is 0.064899646

[0199] The time consumption of the direct quantization method is 0.328125

[0200] The mse of the traditional method is 0.11524421243078596

[0201] The time consumption of the traditional method of eigendecomposition is 1.09375

[0202] From the above simulation comparison, it can be seen that the MSE for the direct quantization CSI feedback process (MSE = 0.064899646) is much lower than the traditional two-step CSI feedback process (MSE = 0.11524421243078596). This result is consistent with the data processing theorem, where more processing leads to greater information loss. In some examples, once the CNN is trained, the CNN may only require sequential addition and multiplication operations to infer the results of the trained CNN. In contrast, the traditional two-step CSI process requires complex eigendecomposition, which involves iterative operations. In some examples, the computational complexity of the direct quantization CSI feedback process in the inference stage is much lower than the traditional process based on eigendecomposition. Therefore, in some examples, the direct quantization CSI feedback process reduces the complexity of the CSI feedback calculation.

[0203] In order to measure the time consumption of each CSI feedback process, the python built-in function of CPU (Intel(R) Core(TM) i5-8350U CPU@1.70GHz) has been used. As shown in the above simulation comparison, the proposed direct quantization CSI feedback process takes less time than the traditional two-step CSI feedback process. In addition, in the above simulation, only the time consumption for eigendecomposition in the traditional two-step CSI feedback process is calculated (the time consumption for channel estimation is ignored).

[0204] The key parameters used for dataset generation are listed below:

[0205] Channel: 408 subcarriers x 200TTI for 8x 1 channels

[0206] delayProfiles={'CDL-A','CDL-B','CDL-C','CDL-D','CDL-E'};

[0207] channel.DelayProfile=string(delayProfiles(randi([1numel(delayProfiles)])));

[0208] channel.DelaySpread=randi([1 300])*1e-9;

[0209] channel.MaximumDopplerShift=randi([5 400]);

[0210] In the above simulations, there are several simplifications in the modeling of the CSI feedback process. For example, for the traditional two-step CSI feedback process and the proposed direct quantization CSI feedback process (for training and testing), the RE pilot is noise-free. This simplification should alleviate the performance gap between the two processes; however, it does not compromise the fairness of the comparison. In another example, the CSI encoder module in the traditional two-step CSI feedback process is modeled as eigendecomposition, and the acquired eigenvectors are assumed to be CSI feedback information. Based on previous simulations and corresponding observations of the traditional two-step CSI feedback process, the correct feedback eigenbeam contributes to more than 80% of the DL MIMO performance. Therefore, the use of eigenvectors should be sufficient to reflect the performance gap. In another example, there is usually a quantization step size for the eigenvector before the CSI feedback should be signaled in the form of "bits". However, this step is omitted in the above model because each true value CSI feedback process, the traditional two-step CSI feedback process and the direct quantization CSI feedback process share the same quantization process, and therefore the omission does not affect the results.

[0211] In the current standard, the CSI-RS pattern and the CSI feedback bit width are configured independently. In the traditional two-step CSI feedback process, the intermediate variable (i.e., the recovered channel) separates the decompression (performed by the channel estimation module) and the compression (performed by the CSI encoder module). For the direct quantization CSI feedback process, there are no intermediate variables to separate the two steps of decompression and compression. Instead, a one-step process is applied, and the decompression and compression steps are related. In some examples, the CSI-RS pattern (i.e., the number of CSI-RS REs and their distribution in the TTI-PRB resource grid) and the CSI feedback size threshold (i.e., the maximum number of feedback bits for CSI feedback) can be determined and / or configured together. In some examples, the CSI feedback size threshold may be referred to as the CSI feedback container size.

[0212] In some examples, a signaling update may be required to perform direct quantized CSI feedback.

[0213] In some examples, UE capability indication may be required. The UE may indicate to the gNB during the RRC establishment phase whether the UE supports direct quantization CSI feedback. This may be indicated by a 1-bit indication: when indicating "1", the UE supports the new direct quantization CSI feedback process; when indicating "0", the UE supports the traditional CSI feedback process.

[0214] In some examples, CSI feedback configuration / reconfiguration may be required. CSI feedback configuration / reconfiguration may include at least one of the following: CSI feedback process selection; CSI feedback recovery method selection; CSI-RS pattern and CSI feedback size threshold configuration, defining the maximum number of feedback bits for CSI feedback.

[0215] In some examples, the gNB indicates to the UE which CSI feedback process is selected. For example, when the UE indicates to the gNB during the RRC establishment phase that it supports the direct quantization CSI feedback process, the gNB needs to indicate to the UE which CSI feedback process has been applied. For example, the gNB needs to indicate to the UE whether the traditional two-step CSI feedback process or the direct quantization CSI feedback process has been applied. In some examples, this can be indicated by a 1-bit indication: when indicating "1", the UE can apply the direct quantization CSI feedback process; when indicating "0", the UE can apply the traditional two-step CSI feedback process.

[0216] In some examples, the gNB indicates to the UE which CSI feedback recovery method is selected. For example, during the RRC establishment process, the gNB may indicate to the UE how it recovers the CSI feedback bits and how it reconstructs the channel. In some examples, several predefined methods may be specified. The gNB may select a predefined method and indicate it to the UE. In some examples, this may be indicated by implementing a new field in the RRC message. In some examples, the default CSI feedback recovery method may be specified as the traditional two-step CSI feedback recovery method. In some examples, no new CSI feedback recovery method needs to be defined. The traditional two-step CSI feedback recovery method works with the direct quantization CSI feedback process. In some examples, the direct quantization CSI feedback process is a machine learning (ML) based method and will adapt the CNN to any predefined recovery method through a training process. In some examples, the simplified recovery method may reduce the computational complexity while maintaining channel recovery performance.

[0217] In some examples, the gNB indicates the configured CSI-RS pattern and CSI feedback size threshold to the UE. In some examples, the CSI feedback size threshold is the maximum number of feedback bits for CSI feedback. In the conventional two-step CSI feedback process, the number of feedback bits is implicitly configured with the number of feedback beams and the corresponding in-phase quantization bit width. In contrast, for the direct quantization CSI feedback process, only one number of total feedback bits may be configured. The gNB may indicate the CSI feedback size threshold (L) as well as the CSI-RS pattern (P). In other words, the gNB may indicate the (L, P) pair to the UE. The (L, P) pair may be determined based on the selected feedback recovery method and the desired channel recovery performance (MSE). In some examples, joint encoding of the CSI-RS pattern (P) and the CSI feedback size threshold (L) may be required. For example, for one CSI feedback recovery method, several (L, P) pairs may be defined by joint encoding to achieve different channel recovery MSEs while saving signaling overhead. In some examples, this may be achieved by using a table indexing the (L, P) pairs. For example, the gNB may indicate to the UE one of the indices that indicate a particular (L, P) pair. In some examples, a default configuration of (L, P) pairs may be calculated using a conventional two-step CSI feedback process with a maximum possible CSI feedback size threshold and a CSI-RS pattern with a maximum possible number of pilot REs. In some examples, using the maximum possible CSI feedback size threshold and a CSI-RS pattern with a maximum possible number of pilot REs may improve CSI feedback performance at the start of a direct quantization CSI feedback process.

[0218] In some examples, UE-assisted direct quantization CSI feedback configuration may be required. For example, as described above, the UE may determine several optional (L, P) pairs based on the configured CSI feedback recovery method indicated to the UE during the gNB CSI feedback configuration / reconfiguration process, and indicate the optimal (L, P) pair to the gNB. In some examples, once the CSI feedback recovery method is indicated to the UE during the RRC establishment phase, the UE may train at least one CNN to derive the optimal CSI-RS pattern and CSI feedback size threshold pair, which together can achieve the specific CSI recovery performance (MSE) (based on the CSI recovery method) that the UE expects to achieve. In some examples, the UE may store the optimal (L, P) pair from any previous session. In some examples, the UE may store the configured feedback recovery method from any previous session. In some examples, during the RRC establishment process, the UE may indicate the optimal (L, P) pair and feedback recovery method from a previous session to assist the gNB in ​​configuring the CSI feedback process.

[0219] In some examples, the gNB can use UL-DL reciprocity to derive the (L, P) pair. For example, the UE can use data-assisted channel estimation to determine the channels on the full TTI-PRB resource grid (i.e., training data). The gNB can obtain the training data by exploiting UL-DL channel reciprocity. UL-DL channel reciprocity enables the gNB to acquire the UL channel and determine that the DL channel is the same and / or similar to the UL channel. In some examples, if training data for the channels on the full TTI-PRB resource grid is available and the CSI feedback recovery method is known, both the UE and the gNB can derive the optimal (L, P) pair. Deriving the optimal (L, P) pair is described in more detail below.

[0220] Figure 4 An example signaling process for directly quantized CSI feedback is shown. In this example, the UE derives the optimal (L, P). However, as described above, the gNB can also derive the optimal (L, P) pair.

[0221] In this example, at S401, UE 402 sends a capability indication to gNB 400. In some examples, the capability indication is sent to gNB 400 during the RRC establishment phase. In some examples, the capability indication includes an indication indicating whether UE 402 supports direct quantized CSI feedback. In some examples, the capability indication includes a suggested (L, P) pair from a previous session. In some examples, the capability indication includes information of a configured CSI feedback recovery method from a previous session.

[0222] At S402, gNB 400 sends a CSI feedback configuration message to UE 402. In some examples, the CSI feedback configuration message includes an indication indicating whether the gNB 400 has selected a direct quantization CSI feedback process. In some examples, the CSI feedback configuration message includes information of a selected CSI feedback recovery method. For example, gNB 400 may select a conventional two-step CSI feedback recovery method. In some examples, the information of the selected CSI feedback recovery method indicates how gNB 400 will recover CSI feedback bits and / or how gNB 400 will reconstruct a channel. In some examples, the CSI feedback configuration message includes an (L, P) pair. For example, gNB 400 determines the (L, P) pair to be sent to UE 402 after considering other UEs in the same cell. In some examples, gNB 400 determines the (L, P) pair to be transmitted to UE 402 after considering cell load and cell performance. For example, gNB 400 may send a (L, P) pair to UE 402 that may improve the performance of the cell.

[0223] In S403, gNB 400 and UE 402 perform DL transmission and CSI feedback.

[0224] At S404, UE 402 performs data-assisted channel estimation to determine a channel on the full TTI-PRB resource grid. In some examples, the channel on the full TTI-PRB resource grid may be training data. Data-assisted channel estimation is described in more detail below.

[0225] At S405, UE 402 derives an optimal (L, P) pair. In some examples, UE 402 trains at least one CNN to derive an optimal (L, P) pair. In some examples, UE 402 determines CNN network parameters. For example, CNN network parameters may include input pilot REs (which are input to the CNN) and output CSI feedback bits (which are output by the CNN). Derivation of the optimal (L, P) pair is described in further detail below.

[0226] At S406, UE 402 sends a triggered RRC reconfiguration message to gNB 400. In some examples, the triggered RRC reconfiguration message includes the derived optimal (L, P) pair. In some examples, the triggered RRC reconfiguration message includes CSI feedback based on the derived optimal (L, P) pair.

[0227] At S407, gNB 400 sends a CSI feedback reconfiguration message to UE 402. In some examples, the CSI feedback reconfiguration message is a reconfiguration message for direct quantization feedback. In some examples, the CSI feedback reconfiguration message includes information of a selected CSI feedback recovery method. For example, gNB 400 may select a conventional two-step CSI feedback recovery method. In some examples, the information of the selected CSI feedback recovery method indicates how gNB 400 will recover CSI feedback bits and / or how gNB 400 will reconstruct a channel. In some examples, the CSI feedback reconfiguration message includes an (L, P) pair. For example, after gNB 400 receives the optimal (L, P) pair from UE 420, gNB 400 may use the optimal (L, P) pair to determine the (L, P) pair to be used by UE 420 to determine the CSI feedback bits. In some examples, gNB 400 will assign an (L, P) pair that improves the performance of the entire cell. In some examples, gNB 400 assigns a (L, P) pair to UE 402 based on at least one of: a received optimal (L, P) pair; cell load information; and cell performance information. In some examples, the optimal (L, P) pair determined by UE 402 is optimal for UE 402. In some examples, UE 420 will use the (L, P) pair indicated by gNB 400 to determine CSI feedback bits and send the CSI feedback bits to gNB 400. In some examples, UE 402 may send CSI feedback bits to gNB 400 based on a CSI feedback threshold size associated with the assigned (L, P) pair received from gNB 400.

[0228] As described in S404 above, the UE performs data-assisted channel estimation to determine the true channel on the full TTI-PRB resource grid. In some examples, the channel on the full TTI-PRB resource grid may be training data. In some examples, data-assisted channel estimation may improve channel estimation performance. In some examples, data-assisted channel estimation may include decoding a transport block. In some examples, when a transport block is successfully decoded, the decoded data RE may be considered as a pilot RE. Using the decoded data RE and reference signal RE, a channel frequency response may be obtained for the entire transport block, thereby obtaining a channel on the full TTI-PRB resource grid.

[0229] As described in S405 above, the UE may determine an optimal (L, P) pair. In some examples, the optimal (L, P) pair may be communicated between the gNB and the UE. In some examples, determining the optimal (L, P) pair and sending it to the gNB facilitates gNB scheduling. In some examples, the gNB may derive the optimal (L, P) pair. In some examples, determining the optimal (L, P) pair facilitates gNB scheduling design.

[0230] In some examples, the UE may determine the (L, P) pair based on data-assisted channel estimation performed by the UE. In some examples, the gNB may determine the (L, P) pair using UL-DL channel reciprocity. In some examples, ML model training may be used to determine the (L, P) pair. In some examples, the UE may determine the optimal (L, P) pair based on data-assisted channel estimation performed by the UE. In some examples, the gNB may determine the optimal (L, P) pair based on UL-DL channel reciprocity. For example, the ML model training may include determining a true channel on the full TTI-PRB resource grid based on the data-assisted channel estimation, and selecting the optimal (L, P) pair by comparing the channel recovery MSEs of multiple (L, P) pairs. In some examples, the gNB obtains the true channel on the full TTI-PRB resource grid by utilizing UL-DL channel reciprocity. In some examples, the channel recovery MSE of the (L, P) pair may be obtained by comparing the channel estimate obtained using the (L, P) pair with the true channel on the full TTI-PRB resource grid. In some examples, the results of the measured true value channel on the full resource grid can be used as training data for the ML model to determine the optimal (L, P) pair. In some examples, the UE and / or gNB can utilize the ML model for CSI feedback acquisition and the training data to determine multiple CSI feedbacks, wherein each of the determined multiple CSI feedbacks is associated with a corresponding (L, P) pair.

[0231] Figure 5 Determination of an optimal (L, P) pair using data-assisted channel estimation and ML model training is shown. In some examples, the optimal (L, P) pair is determined by the UE. In some examples, the optimal (L, P) pair is determined by the gNB. In some examples, the optimal (L, P) pair is optimal for the UE.

[0232] exist Figure 5In some examples, the channel grid 500 of the true value measurement can be understood as training data. In some examples, the training data is determined at the UE using data-assisted channel estimation. In some examples, the training data is obtained by the gNB using UL-DL channel reciprocity. In some examples, the true value channel can be measured on the full channel grid. In some examples, the true value channel can be measured on the resource. For example, the true value channel can be measured on the RE. In some examples, the resource may include (multiple) CSI-RS indicated by the CSI-RS pattern and some other resources (such as, for example, UL sounding reference signal). In some examples, when the gNB obtains the training data using UL-DL channel reciprocity, UL SRS resources can be used. In some examples, when the true value channel is measured on the resource, not all resources of the channel can be measured. In some examples, multiple (L, P) pairs can be determined at the UE and / or gNB based on an estimate or measurement of radio channel characteristics. In some examples, the radio channel can refer to a channel used to send CSI-RS according to the CSI-RS pattern. For example, for a specific CSI-RS pattern with specific channel characteristics (or radio conditions), a specific number of CSI feedback bits may be required. In some examples, a CSI feedback size threshold may be determined such that the certain number of CSI feedback bits may be adapted according to the CSI feedback size threshold used to transmit the CSI feedback. In some examples, a default (L, P) pair may be used to determine multiple (L, P) pairs. In some examples, the gNB may send the default (L, P) pair to the UE. In some examples, the default (L, P) pair is a default traditional two-step CSI feedback (L, P) pair. In some examples, multiple (L, P) pairs may be determined by following the default (L, P) pair and using data-assisted channel estimation. In some examples, the gNB selects and / or determines a CSI feedback recovery method. For example, the gNB may select a traditional two-step CSI feedback recovery method. In some examples, information of the selected CSI feedback recovery method indicates how the gNB will recover the CSI feedback bits and / or how the gNB will reconstruct the channel. In some examples, the gNB transmits information of the feedback recovery method to the UE. In some examples, when the UE determines the optimal (L, P) pair, the gNB may transmit information of the feedback recovery method to the UE.

[0233] At S501, the UE and / or gNB may use a plurality of different CSI-RS patterns (e.g., P1, P2, P3) to obtain different patterns (e.g., 502, 504, 506) from the measured true channel grid 500. For example, when the true channel grid 500 has been measured, different REs of the measured channel grid 500 may be obtained. In some examples, each CSI-RS pattern may be associated with a corresponding CSI feedback size threshold (e.g., L1, L2, L3). The CSI feedback size threshold and the CSI-RS pattern may form a pair (e.g., L1, P1; L2, P2; L3, P3).

[0234] At S502, the UE and / or the gNB may train the ML models 508, 510, 512 using training data. For example, the UE and / or the gNB may input resources obtained based on patterns 502, 504, 506 associated with multiple (L, P) pairs into the ML models 508, 510, 512. In some examples, CSI feedback may be obtained as an output. In some examples, the ML models 508, 510, 512 may be CNNs. In some examples, based on different patterns 502, 504, 506 associated with each of the multiple (L, P) pairs, the CSI feedback obtained using the measured channel (i.e., the measured resources) may vary. In some examples, each CSI feedback associated with each (L, P) pair may be compared with the CSI feedback associated with the measured true value channel grid 500. In some examples, each of the multiple (L, P) pairs associated with each pattern 502, 504, 506 may provide a different channel recovery MSE. For example, as Figure 5As shown in , the (L1, P1) pair can be associated with MSE1, the (L2, P2) pair can be associated with MSE2, and the (L3, P3) pair can be associated with MSE3. In some examples, each of the multiple (L, P) pairs associated with each mode 502, 504, 506 can provide a different overhead. In some examples, the MSE and overhead associated with each of the multiple (L, P) pairs are determined by the ML model 508. In some examples, the optimal (L, P) pair is selected from the multiple (L, P) pairs. In some examples, the optimal (L, P) pair is selected based on at least one of the following: the MSE associated with each (L, P) pair; and the overhead associated with each (L, P) pair. For example, the (L, P) pair associated with the lowest MSE is selected. In some examples, the (L, P) pair associated with the MSE that meets the MSE requirements of the feedback recovery method is selected. In some examples, a (L, P) pair associated with an MSE that satisfies the MSE requirement of the feedback recovery process and produces minimal overhead is selected. In some examples, the minimal overhead includes at least one of: minimal CSI feedback overhead; minimal CSI-RS RE overhead. In some examples, there are additional sources of overhead associated with each (L, P) pair. In some examples, a (L, P) pair associated with minimal overhead is selected. In some examples, a (L, P) pair associated with an optimal tradeoff between MSE and overhead is selected.

[0235] In some examples, an optimal (L, P) pair may be determined using a trial and selection method. For example, after the gNB has determined and / or selected a CSI feedback recovery method, the gNB may determine multiple (L, P) pairs. In some examples, after information about the CSI feedback recovery method has been signaled from the gNB to the UE, the UE may determine multiple (L, P) pairs. In some examples, the gNB and / or the UE may determine multiple (L, P) pairs based on an estimate or measurement of radio channel characteristics. In some examples, multiple (L, P) pairs may be determined using a default (L, P) pair. In some examples, the gNB may send a default (L, P) pair to the UE. In some examples, the default (L, P) pair is a default traditional two-step CSI feedback (L, P) pair. In some examples, multiple (L, P) pairs may be determined by following the default (L, P) pair and using data-assisted channel estimation.

[0236] In some examples, the UE and / or the gNB constructs an ML model for each of a plurality of (L, P) pairs. In some examples, the ML model may be a neural network (NN). In some examples, the ML model may be a CNN. In some examples, each ML model is trained and tested. In some examples, the training and testing of each ML model may be performed in parallel. In some examples, each ML model is trained and tested using data obtained from data-assisted channel estimation. In some examples, the training data is determined at the UE using data-assisted channel estimation. In some examples, the training data is obtained by the gNB using UL-DL reciprocity. For example, CSI-RS patterns associated with a plurality of (L, P) pairs may be used to obtain different patterns from a measured true channel grid. In some examples, resources based on different patterns associated with each (L, P) pair may be input into each corresponding constructed ML model. Each ML model constructed for each (L, P) pair may output CSI feedback. In some examples, each ML model constructed for each (L, P) pair may determine an estimated channel recovery MSE for a feedback recovery method associated with each (L, P) pair. For example, the ML model may compare each output CSI feedback with the CSI feedback associated with the true channel grid to determine an estimated channel recovery MSE associated with each (L, P) pair. In some examples, each ML model determines an estimated overhead associated with each (L, P) pair. In some examples, the gNB and / or UE determines and / or selects an optimal (L, P) pair from each of a plurality of (L, P) pairs. In some examples, the optimal pair is optimal for the UE. In some examples, the optimal (L, P) pair is selected based on at least one of: a channel recovery MSE associated with each (L, P) pair; and an overhead associated with each (L, P) pair. For example, a (L, P) pair associated with the lowest MSE is selected. In some examples, a (L, P) pair associated with an MSE that satisfies the MSE requirements of the feedback recovery method is selected. In some examples, a (L, P) pair associated with an MSE that satisfies the MSE requirements of the feedback recovery process and produces the minimum overhead is selected. In some examples, the minimum overhead includes at least one of: minimum CSI feedback overhead; minimum CSI-RS RE overhead. In some examples, there is an additional overhead source associated with each (L, P) pair. In some examples, the (L, P) pair associated with the minimum overhead is selected. In some examples, the (L, P) pair associated with the optimal tradeoff between MSE and overhead is selected.

[0237] In some examples, when the UE determines the optimal (L, P) pair, the UE may send the selected (L, P) pair to the gNB. In some examples, the UE may send the optimal (L, P) pair to the gNB via RRC signaling. In some examples, the UE may send the selected (L, P) pair to the gNB via an RRC reconfiguration message. In some examples, the gNB assigns the (L, P) pair to the UE based on at least one of the following: the received optimal (L, P) pair; cell load information; and cell performance. In some examples, the gNB sends the assigned pair to the UE. In some examples, the assigned pair may be the determined optimal (L, P) pair. In some examples, the UE determines and / or calculates CSI feedback bits based on the assigned pair. In some examples, the UE determines the CSI feedback bits using at least one ML model. In some examples, the ML model may be a NN. In some examples, the ML model may be a CNN. For example, the ML model is used to quantize the channel on the RE pilot to generate CSI feedback bits. In some examples, the channel on the RE pilot may be based on a CSI-RS pattern (P) associated with an assigned (L, P) pair indicated by the gNB. In some examples, the number of CSI feedback bits is based on a CSI feedback size threshold (L) associated with an assigned (L, P) pair indicated by the gNB. In some examples, the UE may send the determined CSI feedback bits to the gNB. In some examples, the UE may send the CSI feedback bits to the gNB based on the CSI feedback size threshold associated with the assigned (L, P) pair indicated by the gNB.

[0238] In some examples, when the gNB determines the optimal (L, P) pair, the gNB assigns the (L, P) pair based on at least one of: the determined optimal (L, P) pair; cell load; and cell performance. In some examples, the gNB sends the assigned pair to the UE. In some examples, the assigned pair may be the determined optimal (L, P) pair. In some examples, the UE determines CSI feedback bits based on the assigned pair. In some examples, the UE determines and / or calculates the CSI feedback bits using at least one ML model. In some examples, the UE determines the CSI feedback bits using at least one ML model. In some examples, the ML model may be a NN. In some examples, the ML model may be a CNN. For example, the ML model is used to quantize a channel on the RE pilot to generate the CSI feedback bits. In some examples, the channel on the RE pilot may be based on a CSI-RS pattern (P) associated with the assigned (L, P) pair indicated by the gNB. In some examples, the number of CSI feedback bits is based on a CSI feedback size threshold (L) associated with the assigned (L, P) pair indicated by the gNB. In some examples, the UE may send the determined CSI feedback bits to the gNB. In some examples, the UE may send the CSI feedback bits to the gNB based on a CSI feedback size threshold associated with the assigned (L, P) pair indicated by the gNB.

[0239] Sending the assigned pair to the UE may be beneficial because both the UE and the gNB may be aware of the (L, P) pair used to determine the CSI feedback bits during the direct quantization CSI feedback process. In some examples, the UE may not indicate the optimal (L, P) pair to the gNB if the optimal (L, P) pair is determined at the gNB.

[0240] In some examples, using an ML mode (e.g., NN or CNN) for directly quantizing CSI feedback can be simple and can have only a few layers, so the training and testing of the NN and / or CNN can be completed quickly. In some examples, if a CSI feedback recovery method other than the traditional two-step feedback CSI recovery method is defined and configured, the trial and selection method can be applied. For example, any CSI feedback recovery method can work in the trial and selection method. In some examples, the CSI feedback recovery method is predefined and communicated between the gNB and the UE. Determination of the optimal (L, P) pair can be performed at the UE and the gNB. For example, if channel data (i.e., training data) is determined (e.g., by data-assisted channel estimation), the NN and / or CNN can be trained.

[0241] In other examples, a meta-learning approach can be used to determine the optimal (L, P) pair. For example, through meta-learning, a separate ML model can be constructed on top of the constructed NN and / or CNN. In some examples, the ML model can use reinforcement learning to learn the structure of the NN and / or CNN. For example, the ML model can learn at least one of the following: the input layer structure of the NN and / or CNN (i.e., the CSI-RS pattern); the output layer structure of the NN and / or CNN (i.e., the CSI feedback size threshold); and any intermediate layers of the NN and / or CNN. In some examples, the ML model can learn the structure of a NN and / or CNN that has a fixed inner layer type but has a flexible size of the layer (e.g., the number of neurons in the layer, which can be mapped to a CSI-RS pattern or a feedback size threshold). In some examples, the structure of the learned NN and / or CNN infers the MSE and / or overhead associated with the input layer and the output layer.

[0242] The proposed direct quantization CSI feedback process includes advantages over other CSI feedback processes (e.g., conventional two-step CSI feedback process or true value CSI feedback process). The direct quantization CSI feedback process may provide better CSI feedback performance with less channel recovery MSE and / or less overhead at the gNB, including CSI-RS overhead and CSI feedback overhead. The direct quantization CSI feedback process requires less time and / or less power consumption. In addition, the direct quantization CSI feedback process reduces implementation complexity.

[0243] Figure 6 Diagram for controlling Figure 1 600 for the functions of the 5GRAN or 5GC illustrated in FIG. The control device may include at least one random access memory (RAM) 611a, at least one read-only memory (ROM) 611b, at least one processor 612, 613, and an input / output interface 614. At least one processor 612, 613 may be coupled to the RAM 611a and the ROM 611b. At least one processor 612, 613 may be configured to execute appropriate software code 615. The software code 615 may, for example, allow one or more steps to be executed to perform one or more of the present aspects. The software code 615 may be stored in the ROM 611b. The control device 600 may be interconnected with another control device 600 that controls another function of the 5GRAN or 5GC. In some embodiments, each function of the 5GRAN or 5GC includes a control device 600. In an alternative embodiment, two or more functions of the 5GRAN or 5GC may share a control device.

[0244] Figure 7 An example of a terminal 700 is shown, such as Figure 1The terminal 700 is a terminal illustrated in FIG. The terminal 700 may be provided by any device capable of sending and receiving radio signals. Non-limiting examples include user equipment, a mobile station (MS) or a mobile device such as a mobile phone or a so-called "smart phone", a computer equipped with a wireless interface card or other wireless interface device (e.g., a USB dongle), a personal data assistant (PDA) or tablet computer equipped with wireless communication capabilities, a machine type communication (MTC) device, an Internet of Things (IoT) type communication device, or any combination of these devices, etc. The terminal 700 may provide, for example, data communication for bearer communication. The communication may be one or more of voice, electronic mail (email), text message, multimedia, data, machine data, etc.

[0245] The terminal 700 may be equipped with at least one processor 701, at least one memory ROM 702a, at least one RAM 702b and other possible components 703 for software and hardware assistance to perform the tasks it is designed to perform, including controlling access and communication to access systems and other communication devices. At least one processor 701 is coupled to RAM 702b and ROM 702a. At least one processor 701 may be configured to execute appropriate software code 708. The software code 708 may, for example, allow one or more aspects of the present invention to be performed. The software code 708 may be stored in ROM 702a.

[0246] The processor, memory and other related control means may be provided on an appropriate circuit board and / or in a chipset. This feature is indicated by reference numeral 704. The device may optionally have a user interface, such as a keyboard 705, a touch-sensitive screen or keyboard, a combination thereof, etc. Depending on the type of device, one or more of a display, a speaker and a microphone may optionally be provided.

[0247] The terminal 700 may receive signals over the air or radio interface 707 via appropriate means for reception, and may transmit signals via appropriate means for transmission of radio signals. Figure 7 In the embodiment of the present invention, a transceiver device is schematically indicated by a block 706. The transceiver device 706 may be provided, for example, by means of a radio part and an associated antenna arrangement. The antenna arrangement may be arranged internally or externally to the mobile device.

[0248] Figure 8 is a flow chart of a method according to an example. Figure 8 The flowchart is viewed from the perspective of an apparatus. For example, the apparatus may include an access node.

[0249] As shown in S801, the method includes: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment.

[0250] As shown in S802, the method includes: assigning a pair of a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to a user equipment based on at least one of cell load information and cell performance information and the determined optimal pair.

[0251] As shown in S803, the method includes: sending information indicating the assigned pair to the user equipment.

[0252] Fig. 9 is a flow chart of a method according to an example. Figure 8 The flowchart is viewed from the perspective of an apparatus. For example, the apparatus may include a user device.

[0253] As shown in S901, the method includes: receiving information indicating an assigned pair from an access node, the assigned pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal pattern.

[0254] As shown in S902, the method includes: determining a channel state information feedback bit based on the assigned pair.

[0255] Fig.10 is a flow chart of a method according to one example. Figure 8 The flowchart is viewed from the perspective of an apparatus. For example, the apparatus may include a user device.

[0256] As shown in S1001, the method includes: determining an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode, wherein the channel state information feedback size threshold defines the maximum number of bits used for channel state information feedback, wherein the optimal pair is optimal for a user equipment.

[0257] As shown in S1002, the method includes: sending information indicating the determined optimal pair to the access node.

[0258] As shown in S1003, the method includes: receiving information indicating an assigned pair from an access node, the assigned pair including a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal pattern.

[0259] As shown in S1004, the method includes: determining a channel state information feedback bit based on the assigned pair.

[0260] Fig.11 is a flow chart of a method according to an example. Figure 8 The flowchart is viewed from the perspective of an apparatus. For example, the apparatus may include an access node.

[0261] As shown in S1101, the method includes: receiving an optimal pair including a channel state information feedback size threshold and a channel state information reference signal mode from a user equipment, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment.

[0262] As shown in S1102, the method includes: assigning a pair of a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal mode to a user equipment based on at least one of cell load information and cell performance information and the determined optimal pair.

[0263] As shown in S1103, the method includes: sending information indicating the assigned pair to the user equipment.

[0264] Fig.12 Schematic representations of non-volatile storage media 1200a (e.g., a computer disk (CD) or digital versatile disk (DVD)) and 1200b (e.g., a universal serial bus (USB) memory stick) storing instructions and / or parameters 1202 that, when executed by a processor, allow the processor to perform Figure 8 , Fig. 9 , Fig.10 or Fig.11 One or more steps of a method.

[0265] It should be understood that the apparatus may include or be coupled to other units or modules used in transmission and / or reception, such as a radio part or a radio head, etc. Although the apparatus has been described as one entity, different modules and memories may be implemented in one or more physical or logical entities.

[0266] Note that while some embodiments have been described with respect to 5G networks, similar principles may be applied with respect to other networks and communication systems. Thus, while certain embodiments are described above by way of example with reference to certain example architectures for wireless networks, technologies, and standards, the embodiments may be applied to any other suitable form of communication system in addition to the communication system illustrated and described herein.

[0267] It is also noted herein that while the above describes exemplifying embodiments, there are several variations and modifications which may be made to the disclosed solution without departing from the scope of the present invention.

[0268] As used herein, "at least one of: a list of two or more elements" and "at least one of " and similar expressions, where a list of two or more elements is joined by "and" or "or", mean at least any one element, or at least any two or more elements, or at least all elements.

[0269] In general, various embodiments may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. Some aspects of the disclosure may be implemented in hardware, while other aspects may be implemented in firmware or software, which may be executed by a controller, microprocessor, or other computing device, but the disclosure is not limited thereto. Although various aspects of the disclosure may be illustrated and described as block diagrams, flow charts, or using some other graphical representations, it is well understood that, as non-limiting examples, the boxes, devices, systems, techniques, or methods described herein may be implemented in hardware, software, firmware, dedicated circuits or logic, general hardware or controllers or other computing devices, or some combination thereof.

[0270] As used in this application, the term "circuitry" may refer to one or more or all of the following:

[0271] (A) hardware circuit implementation only (e.g., implementation in analog and / or digital circuitry only) and

[0272] (b) a combination of hardware circuitry and software such as (where applicable):

[0273] (i) a combination of analog and / or digital hardware circuits and software / firmware, and

[0274] (ii) any portion of hardware processor(s) (including digital signal processor(s)) with software, software and memory(s) that work together to enable a device such as a mobile phone or server to perform various functions) and

[0275] (c) Hardware circuit(s) and / or processor(s) (such as microprocessor(s) or portion(s) of microprocessor(s)) that require software (e.g., firmware) to operate, but where the software is not required for operation, the software may not be present.

[0276] This definition of circuitry applies to all uses of the 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 only a hardware circuit or processor (or multiple processors) or a portion of a hardware circuit or processor and its (or their) accompanying software and / or firmware. For example and if applicable to a particular claim element, the term circuitry also covers a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in a server, cellular network device, or other computing or network device.

[0277] Embodiments of the present invention may be implemented by computer software that can be executed by a data processor of a mobile device (such as in a processor entity), or by hardware, or by a combination of software and hardware. Computer software or programs, also referred to as program products, include software routines, small applications and / or macros, which may be stored in any device readable data storage medium, and they include program instructions for performing specific tasks. A computer program product may include one or more computer executable components that are configured to perform each embodiment when the program is running. One or more computer executable components may be at least one software code or part thereof.

[0278] Furthermore, it should be noted at this point that any box of the logic flow in the figure may represent program steps, or interconnected logic circuits, boxes and functions, or a combination of program steps and logic circuits, boxes and functions. The software may be stored on a physical medium such as a memory chip or a memory block implemented within a processor, a magnetic medium such as a hard disk or a floppy disk, and an optical medium such as a DVD and its data variant CD. Physical media are non-transient media.

[0279] The term "non-transitory" as used herein is a restriction to the medium itself (ie, tangible, not a signal), not to the persistence of data storage (eg, RAM vs. ROM).

[0280] The memory may be of any type suitable for the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory, and removable memory. As non-limiting examples, the data processor may be of any type suitable for the local technical environment and may include one or more of the following: a general-purpose computer, a special-purpose computer, a microprocessor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an FPGA, a gate-level circuit, and a processor based on a multi-core processor architecture.

[0281] Embodiments of the present invention may be practiced in various components such as integrated circuit modules. The design of integrated circuits is generally a highly automated process. Complex and powerful software tools are available to convert a logic level design into a semiconductor circuit design ready to be etched and formed on a semiconductor substrate.

[0282] The scope of protection sought by various embodiments of the present disclosure is set out by the independent claims. Embodiments and features described in this specification that do not fall within the scope of the independent claims, if any, are to be interpreted as examples useful for understanding various embodiments of the present disclosure.

[0283] The above description has provided a complete and informative description of exemplary embodiments of the present disclosure by way of non-limiting examples. However, when read in conjunction with the accompanying drawings and the appended claims, various modifications and adaptations may become apparent to those skilled in the relevant art in view of the foregoing description. However, all of these and similar modifications of the teachings of the present disclosure will still fall within the scope of the present invention as defined by the appended claims. In fact, there are other embodiments that include a combination of one or more embodiments with any of the other embodiments previously discussed.

Claims

1. A device for channel state information feedback in a cellular communication network, the device comprising: means for determining an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for a user equipment; means for assigning a pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback; as well as Means for sending information indicative of the assigned pair to the user equipment.

2. The apparatus according to claim 1, further comprising: Parts for: Channel state information feedback bits are received from the user equipment based on the assigned pairs.

3. A device for channel state information feedback in a cellular communication network, the device comprising: means for receiving information from an access node indicating an assigned pair comprising: a channel state information reference signal pattern and a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback; and Means for determining a channel state information feedback bit based on the assigned pair.

4. The apparatus according to claim 3, further comprising: Means for determining said channel state information feedback bits using at least one neural network.

5. The device according to claim 3 or 4, further comprising: Means for sending the determined channel state information feedback bits to the access node.

6. A device for channel state information feedback in a cellular communication network, the device comprising: means for determining an optimal pair comprising a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback and a channel state information reference signal pattern, wherein the optimal pair is optimal for the apparatus; means for sending information indicating the determined optimal pair to an access node; means for receiving information from the access node indicating an assigned pair, the assigned pair comprising: a channel state information reference signal pattern and a channel state information feedback size threshold defining a maximum number of bits for channel state information feedback; as well as Means for determining a channel state information feedback bit based on the assigned pair.

7. The apparatus according to claim 6, further comprising: Means for determining said channel state information feedback bits using at least one neural network.

8. The device according to claim 6 or 7, further comprising: Means for sending the determined channel state information feedback bits to the access node.

9. A device for channel state information feedback in a cellular communication network, the device comprising: means for receiving, from a user equipment, an optimal pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern, wherein the channel state information feedback size threshold defines a maximum number of bits for channel state information feedback, wherein the optimal pair is optimal for the user equipment; means for assigning a pair comprising a channel state information feedback size threshold and a channel state information reference signal pattern to the user equipment based on at least one of cell load information and cell performance information and the determined optimal pair, the channel state information feedback size threshold defining a maximum number of bits for channel state information feedback; as well as Means for transmitting information indicative of the assigned pair to the user equipment.

10. The apparatus according to claim 9, further comprising: Means for receiving channel state information feedback bits from the user equipment based on the assigned pair.

11. The apparatus according to any one of claims 1-2 or claims 6-8, wherein determining the optimal pair comprises: determining, based on an estimate or measurement of radio channel characteristics, a plurality of pairs including a channel state information reference signal pattern and a channel state information feedback size threshold; as well as The optimal pair is selected from among the determined plurality of pairs.

12. The apparatus of claim 11, wherein the means for determining the optimal pair comprises: determining an estimated channel recovery mean square error for a feedback recovery method associated with each of the plurality of pairs; determining an estimated cost associated with each of the plurality of pairs; as well as The optimal pair is selected from among the plurality of pairs based on at least one of: the determined channel recovery mean square error; and the determined overhead.

13. The apparatus of claim 12, wherein the overhead associated with each of the plurality of pairs comprises at least one of: channel state information feedback overhead; and channel state information reference signal pilot resource element overhead.

14. The device according to claim 12 or 13, further comprising: Means for determining, using at least one neural network, an estimated channel recovery mean square error and overhead associated with each of the plurality of pairs.

15. The apparatus according to claim 14, when dependent on any one of claims 6-8, further comprising: Means for training and testing the at least one neural network using the channel determined by data-aided channel estimation.

16. The apparatus of claim 14, when dependent on claim 1 or 2, further comprising: Means for training and testing the at least one neural network using a channel, wherein the channel is obtained by the device using uplink-downlink reciprocity.

17. The device according to any one of claims 12 to 16, further comprising: Means for constructing a plurality of neural networks, wherein each neural network of the plurality of neural networks is used to determine an estimated channel recovery mean square error and an estimated overhead associated with one of the plurality of pairs.

18. The apparatus according to claim 17, further comprising: Means for training and testing the plurality of neural networks in parallel.

19. The apparatus of claim 12, when dependent on claim 1 or claim 2, comprising: A component for determining the feedback recovery method and sending information of the feedback recovery method to the user equipment, wherein the feedback recovery method includes information indicating at least one of the following: how the device recovers channel state information feedback bits; and how the device reconstructs a channel.

20. The apparatus according to claim 12, when dependent on any one of claims 6-8, further comprising: A means for receiving information about the feedback recovery method from the access node, wherein the information about the feedback recovery method includes information indicating at least one of: how the access node recovers channel state information feedback bits; and how the access node reestablishes a channel.

21. The device according to any one of claims 14 to 20, further comprising: A component for constructing a separate machine learning model on top of the at least one neural network, wherein the machine learning model learns the structure of the at least one neural network.

Images