Communication method, communication device, medium, and program product

Abstract

The invention provides a communication method, a communication device, a computer readable storage medium and a computer program product. The communication method comprises the following steps: a first communication device obtains training data, wherein the training data is used for training a dictionary for data compression; determining a group of subsets to be updated in the dictionary based on the training data; determining dictionary updating information corresponding to the group of subsets to be updated; and sending dictionary update information. In this way, efficient dictionary updating can be supported, and computing resources and wireless transmission resources consumed by dictionary updating are reduced.

Description

Communication method, communication device, medium and program product Technical Field

[0001] The present application relates to the field of communications, and more particularly, to a communication method, a communication device, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the development of wireless communication technology, higher requirements are placed on data transmission, such as higher throughput, lower latency, and higher accuracy. When the amount of data is large, compression operations need to be performed before data transmission to reduce the consumption of wireless transmission resources. In order to improve compression efficiency, a variety of data compression technologies have been proposed, such as Discrete Cosine Transform (DCT), Discrete Fourier Transform (DFT), Discrete Wavelet Transform (DWT), entropy coding, etc. If the above data compression methods are used, when the compression rate is high, the data transmission performance will deteriorate; when the compression rate is low, a large amount of transmission resources will be occupied, resulting in high transmission resource overhead. Therefore, there is still a need for improved technologies to meet the needs of higher accuracy and lower resource consumption.

[0003] Summary of the Invention

[0004] In view of this, embodiments of the present application provide a communication method, a communication device, a computer-readable storage medium, and a computer program product based on a dynamic dictionary for data compression.

[0005] In a first aspect, a method is provided. The method includes: a first communication device obtaining training data for training a dictionary for data compression; determining, based on the training data, a set of subsets in the dictionary to be updated; determining dictionary update information corresponding to the set of subsets to be updated; and transmitting the dictionary update information. This method supports efficient dictionary updates and reduces computing and wireless transmission resources consumed by dictionary updates. This improves the accuracy of dictionary compression.

[0006] In a second aspect, a method is provided. The beneficial effects of the method can be found in the description of the first aspect and are not further elaborated herein. The method includes: a second communication device receiving dictionary update information for a dictionary used for data compression from a first communication device, wherein the dictionary update information includes information indicating a subset of a set of subsets to be updated in the dictionary and an update value corresponding to the subset of the set of subsets to be updated; and updating the dictionary based on the dictionary update information.

[0007] In a third aspect, a first communication device is provided, comprising a module or unit configured to execute the method of the first aspect.

[0008] In a fourth aspect, a second communication device is provided, which includes a module or unit configured to execute the method of the second aspect.

[0009] In a fifth aspect, a first communication device is provided, comprising a processor coupled to a memory, wherein the memory stores instructions that, when executed by the processor, cause the first communication device to perform the method of the first aspect.

[0010] In a sixth aspect, a second communication device is provided, comprising a processor coupled to a memory, wherein the memory stores instructions that, when executed by the processor, cause the second communication device to perform the method of the second aspect.

[0011] In a seventh aspect, a computer-readable storage medium is provided, which stores instructions, which, when run, enable the method according to the first aspect or any implementation thereof to be executed.

[0012] In an eighth aspect, a computer-readable storage medium is provided, which stores instructions, which, when run, enable the method according to the second aspect or any implementation thereof to be executed.

[0013] In a ninth aspect, a computer program product is provided, wherein the computer program product includes instructions, which, when executed, enable the method according to the first aspect or any implementation thereof to be executed.

[0014] In a tenth aspect, a computer program product is provided, wherein the computer program product includes instructions, which, when executed, enable the method according to the second aspect or any implementation thereof to be executed.

[0015] In an eleventh aspect, a communication system is provided, comprising the first communication device according to the third or fifth aspect and the second communication device according to the fourth or seventh aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0017] FIG1 is a schematic diagram of a communication system to which embodiments of the present application may be applied;

[0018] FIG2 shows an interactive signaling diagram of a method for updating a dictionary according to some embodiments of the present application;

[0019] FIG3A is a schematic diagram showing an example implementation of a method for determining dictionary update information according to some embodiments of the present application;

[0020] FIG3B is a schematic diagram showing an example implementation of a method for updating a compression dictionary value according to some embodiments of the present application;

[0021] FIG3C is a schematic diagram illustrating an example implementation of compressing control information according to some embodiments of the present application;

[0022] FIG4 is a schematic diagram showing an example implementation of a method for selecting training data according to some embodiments of the present application;

[0023] FIG5A is a schematic diagram showing an example implementation of a method for periodically updating a dictionary according to some embodiments of the present application;

[0024] FIG5B is a schematic diagram showing an example implementation of a method for aperiodic dictionary update according to some embodiments of the present application;

[0025] FIG6A is a schematic diagram showing an example implementation of a method for periodically updating a dictionary according to some embodiments of the present application;

[0026] FIG6B is a schematic diagram showing an example implementation of a method for aperiodic dictionary update according to some embodiments of the present application;

[0027] FIG7A shows a schematic diagram of an example implementation of a method for compressing channel state information according to some embodiments of the present application;

[0028] FIG7B is a schematic diagram showing an example implementation of a dictionary compression method according to some embodiments of the present application;

[0029] 8A and 8B are schematic diagrams illustrating example implementations of bit sequence sets according to some embodiments of the present application;

[0030] 9A to 9D are schematic diagrams illustrating example implementations of information including dictionary compression parameters according to some embodiments of the present application;

[0031] FIG10 shows a schematic flowchart of a method implemented at a first communication device according to some embodiments of the present application;

[0032] FIG11 shows a schematic flow chart of a method implemented at a second communication device according to some embodiments of the present application;

[0033] FIG12 is a schematic diagram of the main components of an example device of a possible implementation method according to an embodiment of the present disclosure; and

[0034] FIG13 is a simplified block diagram of an example device for one possible implementation of an embodiment of the present disclosure. DETAILED DESCRIPTION

[0035] The following describes embodiments of the present application in more detail with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are for illustrative purposes only and are not intended to limit the scope of protection of the present application.

[0036] The present invention can be implemented in accordance with any suitable communication protocol, including but not limited to the fourth generation (4G) th generation, 4G), fifth generation (5 th generation, 5G) and the communication protocols that evolved after 5G (for example, the sixth generation (6 th generation, 6G)) and other cellular communication protocols, wireless local area network communication protocols such as Institute of Electrical and Electronics Engineers (IEEE) 802.11, and / or any other protocol currently known or developed in the future.

[0037] The technical solutions of the embodiments of the present application are applied to communication systems that follow any appropriate communication protocol, such as: Long Term Evolution (LTE) system, Frequency Division Duplex (FDD) system, Time Division Duplex (TDD) system, 5G system (e.g., NR) and communication systems evolved after 5G (e.g., 6G system), etc.

[0038] For the purpose of explanation, the following is referred to as the Third Generation Partnership Project (3 rd The embodiments of the present application are described in the context of a cellular communication system in the 3GPP (3rd Generation Partnership Project). However, it should be understood that the embodiments of the present application are not limited to this communication system, but can be applied to any communication system with similar problems, such as a wireless local area network (WLAN), a wired communication system, or other communication systems developed in the future.

[0039] The term "terminal" or "terminal device" as used in this disclosure refers to any terminal device that can perform wired or wireless communication with network devices or with each other. Terminal devices may sometimes be referred to as user equipment or UE. Terminal devices may be any type of mobile terminal, fixed terminal or portable terminal. Terminal devices may be various wireless communication devices with wireless communication capabilities. With the rise of Internet of Things (IoT) technology, more and more devices that did not previously have communication capabilities, such as but not limited to household appliances, vehicles, tools and equipment, service equipment and service facilities, have begun to obtain wireless communication capabilities by configuring wireless communication units, so that they can access wireless communication networks and accept remote control. Such devices have wireless communication capabilities because they are configured with wireless communication units, and therefore also fall into the category of wireless communication devices. As an example, the terminal device may include a mobile cellular phone, a cordless phone, a mobile terminal (MT), a mobile station, a mobile device, a wireless terminal, a handheld device, a client, a subscription station, a portable subscription station, an Internet node, a communicator, a desktop computer, a laptop computer, a notebook computer, a tablet computer, a personal communication system device, a personal navigation device, a personal digital assistant (PDA), a wireless data card, a wireless modem (Modulator demodulator, Modem), a positioning device, a radio broadcast receiver, an e-book device, a gaming device, an IoT device, a vehicle-mounted device, an aircraft, a virtual reality (VR) device, an augmented reality (AR) device, a wearable device (e.g., a smart watch), a terminal device in a 5G network or any terminal device in an evolved public land mobile network (PLMN), other devices that can be used for communication, or any combination thereof. The embodiments of the present application are not limited to this.

[0040] The term "network node" or "network device" used in this application refers to an entity or node that can be used to communicate with a terminal device, for example, an access network device. An access network device can be a device deployed in a wireless access network to provide wireless communication functions for a mobile terminal, for example, a radio access network (RAN) network device. Access network devices may include various types of base stations. Base stations are used to provide wireless access services to terminal devices. Depending on the size of the service coverage area provided, access network devices may include macro base stations providing macro cells, micro base stations for providing micro cells (Pico cells), pico base stations for providing micro cells, and femto base stations for providing femto cells. In addition, access network devices may also include various forms of satellites, relay stations, access points, remote radio units (RRUs), radio heads (RHs), remote radio heads (RRHs), transmission points (transmitting and receiving points, TRPs), transmitting points (transmitting points, TPs), etc. In systems using different wireless access technologies, the names of access network equipment may vary. For example, in LTE networks, it is called an evolved NodeB (eNB or eNodeB), in 3G networks it is called a NodeB (NB), and in 5G networks it may be called a gNodeB (gNB) or a NR NodeB (NR NB), etc. In some scenarios, access network equipment may include a Central Unit (CU) and / or a Distributed Unit (DU). The CU and DU can be located in different locations, for example: a remote DU in a high-traffic area and a CU in a central computer room. Alternatively, the CU and DU can be located in the same computer room. The CU and DU can also be different components within the same rack. Network equipment can also be devices that perform base station functions in device-to-device (D2D), vehicle-to-everything (V2X), and machine-to-machine (M2M) communications. For the convenience of description, in the subsequent embodiments of the present application, the above-mentioned devices that provide wireless communication functions for mobile terminals are collectively referred to as network devices, and the embodiments of the present application are no longer specifically limited.

[0041] With the advancement of wireless communication technology, physical-layer sensing, imaging, and artificial intelligence / machine learning (AI / ML) computing will become potential technologies and new application scenarios for future cellular and Wi-Fi communication systems. Future mobile terminals, sensors, base stations, and other devices will be able to sense and image the environment using electromagnetic signals, enabling offline or real-time modeling and analysis of the wireless transmission environment, ultimately significantly improving communication system performance. The computing power, battery capacity, and environmental range of individual devices are relatively limited. Therefore, the results of sensing, imaging, and AI / ML computing may need to be transmitted back to a remote central node (such as a base station, server, cloud computing center, or terminal device with strong computing power) for information fusion. For applications such as integrated sensing, imaging, and communication-sensing integration, the resulting signal data volume is large due to the use of broadband, multi-frequency bands, larger antenna arrays, and the acquisition of electromagnetic signals from different directions. Therefore, compression is required before wireless backhaul to reduce the consumption of wireless transmission resources. For AI / ML applications, processes such as model distribution and online training may also involve the transmission of large amounts of data, thus also requiring compression before transmission. However, there is currently a lack of effective solutions to meet the needs of higher accuracy and lower resource consumption.

[0042] In view of this, an embodiment of the present application provides a communication method based on a dynamic dictionary. In this method, a first communication device (for example, a terminal device or a network device) obtains training data for determining dictionary update information, and the dictionary is used for data compression. The first communication device determines a group of subsets to be updated in the dictionary based on the training data, and then determines the dictionary update information corresponding to the group of subsets to be updated. The first communication device sends the dictionary update information to the communication device communicating with it. In this way, the first communication device can efficiently perform dictionary update training, reduce the computing resources consumed by dictionary updates, and save the wireless transmission resources required to send dictionary update information. Thus, the accuracy of dictionary compression can be improved. As pointed out above, the embodiment of the present application can be applied to any other communication scenario without any limitation. In order to more clearly describe the embodiment of the present application, the embodiment of the present application is described with reference to Figures 1 to 11.

[0043] FIG1 illustrates a schematic diagram of a communication system 100 in which embodiments of the present application may be implemented. As shown in FIG1 , system 100 may include first to third terminal devices 110-1 to 110-3 (collectively referred to as terminal devices 110) and a network device 120. Network device 120 and terminal device 110 may communicate directly. For example, terminal device 110 may communicate with corresponding network device 120 via a wireless link.

[0044] It should be understood that the number of terminal devices and network devices shown in Figure 1 is for example only. There may be more or fewer terminal devices and network devices, and this application does not impose any restrictions on this. The terminal device (or chip in the terminal device) or network device (or chip in the network device) that performs dictionary update training to determine the dictionary update training below may be referred to as a first communication device or may include a first communication device, and the network device (or chip in the network device) or terminal device (or chip in the terminal device) that receives dictionary update information to perform dictionary update may be referred to as a second communication device or may include a second communication device. Some embodiments below describe communication between a first communication device and a second communication device, and some embodiments describe communication between a terminal device and a network device. It should be understood that these communications are not limited to occurring between a terminal device and a network device, and in some scenarios may also occur between terminal devices, between network devices, or between any two or more communication devices. For example, a communication system may include multiple network devices, and network devices may communicate directly with each other. For example, network devices may communicate with each other via a backhaul link, which may be a wired backhaul link (eg, optical fiber, copper cable) or a wireless backhaul link (eg, microwave).

[0045] FIG2 illustrates an interactive signaling diagram for a method 200 for updating a dictionary according to some embodiments of the present application. For clarity, method 200 will be described in conjunction with FIG1 . First communication device 210 may be terminal device 110 (or a chip of terminal device 110) or network device 120 (or a chip of network device 120) in FIG1 , and second communication device 220 may be network device 120 (or a chip of network device 120) or terminal device 110 (or a chip of terminal device 110) in FIG1 .

[0046] 2 , the first communication device 210 obtains ( 202 ) training data. The training data is used to train a dictionary for data compression. The term “dictionary” used in the present disclosure refers to a set of basic elements that can be used to characterize original data or signals in dictionary learning. The basic elements that constitute the dictionary can also be referred to as a dictionary subset. In some embodiments, when data compression is performed based on a given dictionary, the original data can be weighted using the dictionary subset in the dictionary to obtain a weighting coefficient corresponding to the dictionary subset. The sequence number of the dictionary subset with the larger absolute value of the weighting coefficient and the corresponding weighting coefficient can be recorded to achieve compression of the original data.

[0047] In some implementations, the dictionary may be a multidimensional matrix, such as a two-dimensional matrix or a three-dimensional matrix, wherein a row vector, a column vector, or a submatrix in the multidimensional matrix serves as a dictionary subset. The dictionary may also be a set consisting of multiple matrices or vectors, wherein the matrices or vectors constituting the dictionary serve as a dictionary subset. The training data may be a row vector, a column vector, or a submatrix, and may be weightedly represented by a subset in the dictionary (e.g., a row vector, a column vector, or a submatrix in the dictionary). In some implementations, the training data may be training data received by the first communication device 210 from another communication device. The other communication device may be a second communication device 220 or other communication device that shares a dictionary with the first communication device. In another implementation, the training data may be training data determined by the first communication device 210 itself.

[0048] The first communication device 210 determines (204) a set of subsets to be updated in the dictionary based on the training data. In some implementations, the first communication device 210 may determine a set of subsets to be trained in the dictionary based on the training data. The first communication device 210 may perform dictionary update training on the subsets to be trained in the dictionary based on the training data, and determine the subsets to be trained that meet the preset conditions as the subsets to be updated. For example, the first communication device 210 may determine the subset to be trained that is updated as the subset to be updated. In another implementation, when the difference between the updated value and the initial value in the subset to be trained reaches a difference threshold, the first communication device 210 may determine the subset to be trained as the subset to be updated. Exemplarily, the dictionary update training may be calculated using a K-singular value decomposition (K-SVD) algorithm.

[0049] The first communication device 210 determines (206) dictionary update information 212 corresponding to a set of subsets to be updated. In some embodiments, the dictionary update information 212 may include information indicating a subset in the set of subsets to be updated and an update value for the subset in the set of subsets to be updated. For example, the information indicating the subset to be updated may be implemented in the form of an index or a bitmap. In some embodiments, the update value of the subset may be an updated value of the subset. In some embodiments, the update value of the subset may be a difference between the updated value of the subset and the initial value.

[0050] The first communication device 210 sends (208) dictionary update information 212 to the second communication device 220. In some embodiments, the first communication device 210 may be the terminal device 110 in Figure 1, and the second communication device 220 may be the network device 210 in Figure 2. The terminal device 110 may perform a dictionary update operation and send the dictionary update information 212 to the network device 210 via an uplink channel. In another embodiment, the first communication device 210 may be the network device 210 in Figure 1, and the second communication device 220 may be the terminal device 110 in Figure 2. The network device 210 may perform a dictionary update operation and send the dictionary update information 212 to the terminal device 110 via a downlink channel.

[0051] The second communication device 220 receives (214) dictionary update information 212 from the first communication device 210. After receiving the dictionary update information 212, the second communication device 220 updates (216) the dictionary based on the dictionary update information 212. In this way, the first communication device 210 and the second communication device 220 share the updated dictionary. In the dictionary update method of the embodiment of the present application, only a subset of the dictionary is selected for update training each time, and only the information of the dictionary subset that needs to be updated is sent. Therefore, the computing resources consumed by the dictionary update and the wireless transmission resources required for sending the dictionary update information can be reduced. Since the dictionary update can be performed efficiently, the data compression performance can be improved.

[0052] FIG3A shows a schematic diagram of an exemplary implementation of a method 300A for determining dictionary update information according to some embodiments of the present application. The method 300A will be described in conjunction with FIG2 . The method 300A may be performed by the first communication device 210 in FIG2 .

[0053] In the method 300A, the first communication device 210 stores a dictionary for data compression (eg, Dict@t0 in FIG. 3A ). The dictionary Dict@t0 may be a column vector x ind The first communication device 210 can be based on the training data v m (m=1, 2, ..., M) to update the dictionary. It should be understood that the dictionary in Figure 3A is only exemplary and not intended to be limiting. The dictionary can be implemented as a multidimensional matrix (e.g., a two-dimensional matrix, a three-dimensional matrix, etc.) consisting of column vectors, row vectors, or submatrices, wherein the row vectors, column vectors, or submatrices in the multidimensional matrix serve as dictionary subsets. The dictionary can also be implemented as a set consisting of multiple matrices or vectors, wherein the matrices or vectors constituting the dictionary serve as dictionary subsets.

[0054] First, for each training data v m(m=1, 2, ..., M), the first communication device 210 can select K column vectors from the dictionary Dict@t0 as the subset to be trained. The selected K column vectors can be the ones that can most closely represent the training data v in a weighted manner. m vector, for example, the training data v m It can be approximately expressed as in Represents the weighted representation of training data v m The kth weight vector of Represents the weighted representation of training data v m The kth weight vector The corresponding weighting coefficient, ind mk Represents a vector The column number in the dictionary Dict@t0. K can be a pre-configured dictionary compression parameter. Training data v m The corresponding set of column numbers of the K weighted vectors in the dictionary Dict@t0 can be expressed as I m ={ind m1 ,ind m2 ,…,ind mK}.

[0055] The first communication device 210 may combine the sets of column numbers corresponding to all training data to obtain a set of column numbers I=I1∪I2∪…∪I M The first communication device 210 can perform an update operation only on the training subset corresponding to the column number set I, thereby obtaining an updated dictionary Dict@t1. When performing dictionary update training, the classic K-SVD algorithm can be used. By performing an update operation only on the training subset in the training subset set I, the amount of update operations can be reduced.

[0056] In some implementations, the first communication device 210 may store the changed dictionary subset in the dictionary subset corresponding to the set I. In another implementation, the first communication device 210 may determine the updated value of the dictionary subset that has changed in the dictionary subset corresponding to the set I. With initial value The difference value between the two sets is calculated, and the subset with the difference value greater than the threshold is determined as the subset to be updated. In some implementations, the difference value can be the updated value of the dictionary subset. With initial value The first or second norm of the difference between the two values, or other value indicating the size of the difference.

[0057] Returning to Figure 2 , the first communication device 210 may send dictionary update information to the second communication device 220. The dictionary update information may include an indication of the subset to be updated and a corresponding update value. For example, the update value of the subset to be updated may be the updated value of the subset to be updated. Alternatively, the update value of the subset to be updated may be the difference between the updated value and the initial value of the subset to be updated.

[0058] In some implementations, the first communication device 210 may send the dictionary update information directly to the second communication device 220. In other implementations, to further reduce bandwidth consumption, the first communication device 210 may compress the dictionary update information and send the compressed dictionary update information to the second communication device 220. For example, the dictionary update information 212 may include updated values ​​compressed based on compression control information. In other words, the updated values ​​of the subset may be compressed based on the compression control information before being sent. The compression control information may include at least one of the following: quantization method, number of quantization bits, transform method, or entropy coding method.

[0059] In some embodiments, the first communication device 210 may determine compression control information and send the compression control information to the second communication device 220. In this way, the second communication device 220 is able to decompress the dictionary update information.

[0060] In some embodiments, the second communication device 220 may determine compression control information and send the compression control information to the first communication device 210. In this way, the second communication device 220 is able to decompress the dictionary update information.

[0061] FIG3B shows a schematic diagram of an exemplary implementation of a method 300B for compressing dictionary update values ​​according to some embodiments of the present application. The method 300B will be described in conjunction with FIG2 . The method 300B may be executed by the first communication device 210 in FIG2 .

[0062] In method 300B, the first communication device 210 may perform a transform such as DFT, DCT, DWT, or other predefined transform on the update values ​​of the dictionary subset to be updated to obtain corresponding transform coefficients. The first communication device 210 may quantize the transform coefficients to obtain corresponding discrete values. For a complex signal, the first communication device 210 may quantize the real and imaginary parts of the complex signal separately. Alternatively, the first communication device 210 may quantize the amplitude and phase of the complex signal separately. For a real signal, the first communication device 210 may directly quantize the real signal. In some implementations, the quantization process may be uniform quantization. In some implementations, the quantization process may be non-uniform quantization. For different dictionaries, the first communication device 210 may use different quantization orders to process the transform coefficients corresponding to the dictionary update values. The first communication device 210 may perform entropy coding (such as, but not limited to, arithmetic coding, Huffman coding, etc.) on the discrete values ​​to obtain a compressed dictionary update value encoding stream.

[0063] In this way, first communication device 210 can efficiently compress dictionary update values, allowing second communication device 220, which shares a dictionary with first communication device 210, to obtain accurate dictionary update information, thereby improving the accuracy of dictionary compression. Because only the update values ​​of a subset of the dictionary need to be compressed and transmitted, dictionary update values ​​compressed using entropy coding consume fewer transmission resources.

[0064] The first communication device 210 may perform method 300B of compressing a dictionary update value based on a compression control parameter. In some embodiments, the compression control parameter may include a quantization method, a number of quantization bits, a transform method, an entropy coding method, and the like. The second communication device 220 may decompress the dictionary update information based on the same compression control parameter to update the dictionary based on the dictionary update value. In some implementations, the first communication device 210 may determine the compression control parameter and send compression control information including the compression control parameter to the second communication device 220. In another implementation, the second communication device 220 may determine the compression control parameter and send compression control information including the compression control parameter to the first communication device 210. In this way, multiple communication devices sharing a dictionary can complete the configuration of the compression control parameter. When the compression control parameter is determined by the terminal device 110 in Figure 1, the terminal device 110 may send the compression control information to the network device 120 via an uplink channel. When the compression control parameters are determined by the network device 120 in Figure 1, the network device 120 can send compression control information to the terminal device 110 through Radio Resource Control (RRC) signaling, Medium Access Control (MAC) signaling, or Physical Downlink Control Channel (PDCCH) signaling.

[0065] FIG3C shows a schematic diagram of an example implementation of compression control information 300C according to some embodiments of the present application. The compression control information 300C may include an indication of a quantization method, an indication of the number of quantization bits, an indication of a transform method, an indication of an entropy coding method, etc. For example, in some implementations, the first communication device 210 and the second communication device 220 may each store a set of quantization method configurations. The compression control information 300C may include an index of the quantization method. quant Similarly, in some implementations, the compression control information 300C may include an index of the transformation mode. trans In some implementations, the compression control information 300C may also include an index of the entropy coding method. entropy . In this way, the transmission resources consumed by sending the compression control information can be reduced. It should be understood that the compression control information 300C shown in Figure 3C is merely exemplary and not intended to be limiting. In some embodiments, at least one of the quantization method, the number of quantization bits, the transformation method, and the entropy coding method can be predetermined and therefore does not need to be included in the compression control information 300C. In some embodiments, other compression methods can also be used to compress the dictionary update value.

[0066] Returning to Figure 2, in some implementations, to obtain training data, the first communication device 210 may select training data from the data to be transmitted to the second communication device 220, and determine a subset to be updated in the dictionary based on the selected training data, and further determine dictionary update information. For example, the first communication device 210 may perform dictionary compression on the data to be transmitted based on the dictionary to obtain dictionary-compressed data. The first communication device 210 may decompress the obtained dictionary-compressed data based on the dictionary to obtain reconstructed data. The first communication device 210 may determine a loss function for the reconstructed data based on the data to be transmitted. Further, the first communication device 210 may determine the training data based on the loss function for the reconstructed data. For example, the first communication device 210 may determine that the data to be transmitted is included in the training data based on determining that the loss function for the reconstructed data meets a loss function threshold. In another example, the first communication device 210 may sort the multiple data to be transmitted based on the loss function for the reconstructed data corresponding to data in the multiple data to be transmitted. The first communication device 210 may select the training data from the multiple data to be transmitted based on the sorting. The first communication device 210 may determine a subset to be updated in the dictionary based on the determined training data, and further determine dictionary update information.

[0067] In some implementations, in order to obtain training data, the second communication device 220 may select training data from the data to be sent to the first communication device 210 and send the training data to the first communication device 210. The first communication device 210 may determine a subset to be updated in the dictionary based on the received training data, and further determine dictionary update information. For example, the second communication device 220 may perform data compression on the selected training data based on a compression control parameter (for example, using entropy coding for efficient compression). The first communication device 210 may receive compressed data corresponding to the training data from the second communication device 220, and decompress the received compressed data based on the same compression control parameter to obtain the training data. The compression control parameter may be at least one of the following: quantization method, number of quantization bits, transformation method, or entropy coding method, etc.

[0068] Figure 4 shows a schematic diagram of an example implementation of a method 400 for selecting training data according to some embodiments of the present application. Method 400 will be described in conjunction with Figures 1 and 2. Method 400 can be performed by the terminal device 110 in Figure 1 or the network device 120 in Figure 1 or the first communication device 210 in Figure 2 or the second communication device 220 in Figure 2. For example, in an example application scenario, the terminal device 110 can obtain channel state information (CSI) associated with the network device 120. The terminal device 110 and the network device 120 can share a dictionary corresponding to the CSI. For example, the terminal device 110 can compress the CSI based on the dictionary and send the dictionary-compressed CSI to the network device 120. Accordingly, the network device 120 can decompress the received CSI based on the same dictionary. Considering that the network device 120 has stronger computing power and more relaxed power consumption restrictions, the network device 120 can perform dictionary training or updating. At this time, the terminal device 110 needs to select training data from the obtained CSI and send the training data to the network device 120 through the uplink channel.

[0069] The following describes method 400 using terminal device 110 selecting training data as an example. Terminal device 110 can obtain the latest data to be sent to network device 120. Terminal device 110 only needs to send a portion of the data to be sent to network device 120 to serve as data for online dictionary training. It should be understood that method 400 can also be performed by other devices or communication devices.

[0070] As shown in Figure 4, terminal device 110 uses a conventional dictionary to perform dictionary compression on the latest data v to be transmitted to obtain dictionary-compressed data. The dictionary-compressed data includes K sets of dictionary subsets and corresponding weighting coefficients, where K can be a preconfigured dictionary compression parameter. The K sets of dictionary subsets and corresponding weighting coefficients can most closely represent the data to be transmitted in a weighted manner.

[0071] The terminal device 110 may perform dictionary decompression on the dictionary compressed data based on the dictionary to obtain the reconstructed data v′. The terminal device 110 may evaluate the loss function of the reconstructed data v′ relative to the original data to be transmitted v. For example, a loss function such as Normalized Mean Squared Error (NMSE=‖vv′‖ may be used. 2 / ‖v‖ 2 ), Generalized Cosine Similarity (GCS = v*v′ H / (‖v‖*‖v′‖)) and other loss functions.

[0072] The terminal device 110 can determine whether the corresponding data can be used for dictionary update training based on predefined rules. For example, in some implementations, the terminal device 110 can select data with an NMSE greater than a preset threshold or a GCS less than a preset threshold as training data. In another implementation, the terminal device 110 can sort the NMSE or GCS of all data to be sent, and select several data with the highest NMSE or the lowest GCS from all data to be sent as training data based on a preset number or preset percentage. In some implementations, the network device 120 can determine the training data selection rules and corresponding parameters (for example, a preset threshold of the loss function, the number or percentage of training data selected based on the sorting), and send the corresponding configuration information to the terminal device 110. In another implementation, the terminal device 110 can determine the training data selection rules and corresponding parameters by itself.

[0073] The terminal device 110 may send training data to the network device 120 via an uplink channel. To save bandwidth resources, the terminal device 110 may compress the training data using a method similar to the method 300B described with reference to FIG3B . The compression control parameters used to compress the training data may include a quantization method, a number of quantization bits, a transform method, an entropy coding method, and the like. In some embodiments, the compression control parameters may be the same as the compression control parameters in the compression control information 300C shown in FIG3C . In other words, the compression control parameters in the compression control information 300C shown in FIG3C may be used to compress the training data and the dictionary update information. In another embodiment, the compression control parameters used to compress the training data may be different from the compression control parameters in the compression control information 300C shown in FIG3C . For example, the compression control parameters used to compress the training data may be configured by the terminal device 110 or by the network device 120. For example, the compression control parameters used to compress the training data may be configured via indication information received from the network device 120.

[0074] In some implementations, the training data used for dictionary updates may be referred to as incremental training data. In other words, incremental training data is newly acquired by the communication device in the current environment and is used to perform online dictionary updates for an already generated dictionary. The communication device may periodically acquire sets of incremental training data. If the second communication device 220 acquires training data and the first communication device 210 performs dictionary update training to determine dictionary update information, the second communication device 220 may send the incremental training data set to the first communication device 210.

[0075] In some implementations, the dictionary can be generated based on a basic training data set, which can be selected from at least one predetermined basic training data set. Considering that the network device 120 has stronger computing power and more relaxed power consumption constraints, the network device 120 can generate the dictionary based on the basic training data set. For example, the network device 120 can store a set of basic training data sets. A set of basic training data sets can include at least one basic training data set pre-defined according to a protocol, and each basic training data set is used to perform initialization training of the dictionary for a corresponding communication scenario or environment. In some implementations, the network device 120 can select an applicable basic training data set from the stored set of basic training data sets based on, for example, a specific application or scenario, to perform initialization training of the dictionary, thereby generating a dictionary. The network device 120 can send the generated dictionary to the terminal device 110.

[0076] In some implementations, the network device 120 may perform initial dictionary training based on a set of basic training data sets to generate a set of dictionaries. The network device 120 may select an applicable dictionary from the stored set of dictionaries based on, for example, a specific application or scenario, and transmit the selected dictionary to the terminal device 110.

[0077] In some implementations, the network device 120 may perform initialization training of the dictionary based on a set of basic training data sets. The network device 120 may send the generated set of dictionaries to the terminal device 110. One of the network device 120 and the terminal device 110 may select an applicable dictionary based on a specific application or scenario, and send an indication of the selected dictionary to the other. It is understood that the indication may be implemented in the form of an explicit indication or an implicit indication. In some embodiments, one of the network device 120 and the terminal device 110 may select an applicable dictionary based on a specific application or scenario, and send the selected dictionary to the other. In some embodiments, the network device 120 and the terminal device 110 may select an applicable dictionary based on a specific application or scenario based on predefined rules, thereby allowing the network device 120 and the terminal 110 to share a dictionary.

[0078] In some implementations, to obtain training data, the first communication device 210 may periodically receive training data from the second communication device 220 to periodically perform dictionary update training. In another embodiment, the first communication device 210 may periodically perform dictionary update training based on training data determined by itself.

[0079] In some embodiments, the first communication device 210 may be implemented by the network device 120 or a chip thereof in Figure 1, and the second communication device 220 may be implemented by the terminal device 110 or a chip thereof in Figure 1. The terminal device 110 may periodically send training data to the network device 120 so that the network device 120 performs dictionary update training on the corresponding dictionary.

[0080] FIG5A shows a schematic diagram of an example implementation of a method 500A for periodically updating a dictionary according to some embodiments of the present application. For clarity, the method 500A will be described in conjunction with FIG1 . The method 500A may involve the first terminal device 110 - 1 and the network device 120 .

[0081] 5A , first terminal device 110-1 and network device 120 may share (502) a dictionary corresponding to data (e.g., CSI). First terminal device 110-1 may determine (504) that a dictionary update time has arrived. First terminal device 110-1 may send (506) training data 508 to network device 120 using reserved transmission resources. Network device 120 may receive (510) training data 508 from first terminal device 110-1. Network device 120 may determine (512) dictionary update information 516 based on training data 508.

[0082] Network device 120 may send (514) dictionary update information 516 to first terminal device 110-1. For example, transmission resources for sending dictionary update information 516 may be reserved. First terminal device 110-1 may receive (518) dictionary update information 516 from network device 120. Network device 120 and first terminal device 110-1 may share (520) the new dictionary. As a result, network device 120 and first terminal device 110-1 may perform data compression and decompression based on the new dictionary. In this way, signaling overhead may be reduced.

[0083] In some implementations, the second communication device 220 may send a resource request to the first communication device 210 based on a predetermined trigger condition for sending training data being satisfied. The first communication device 210 may send an indication of resource allocation to the second communication device 220 based on the resource request received from the second communication device 220. The second communication device 220 may send training data to the first communication device 210 based on the indication of resource allocation. In another embodiment, the first communication device 210 may perform dictionary update training based on a predetermined trigger condition for dictionary update training being satisfied.

[0084] In some embodiments, the first communication device 210 may be implemented by the network device 120 or a chip thereof in Figure 1, and the second communication device 220 may be implemented by the terminal device 110 or a chip thereof in Figure 1. The terminal device 110 may send a resource request to the network device 120 based on the trigger condition being satisfied to request resources for sending training data.

[0085] FIG5B shows a schematic diagram of an example implementation of a method 500B for aperiodic dictionary update according to some embodiments of the present application. For clarity, the method 500B will be described in conjunction with FIG1 . The method 500B may involve the first terminal device 110 - 1 and the network device 120 .

[0086] The first terminal device 110-1 and the network device 120 may share (502) a dictionary corresponding to data (e.g., CSI). The first terminal device 110-1 triggers (522) a dictionary update operation based on determining that a condition for triggering a dictionary update is satisfied. For example, the first terminal device 110-1 may estimate the NMSE or GCS of each data to be sent to the network device 120 after dictionary compression and decompression. When the NMSE of one or more data to be sent (e.g., a predetermined number of data to be sent or a predetermined percentage of data to be sent) is higher than a set threshold or the GCS is lower than a set threshold, the dictionary update operation may be triggered.

[0087] First terminal device 110-1 may send 524 a resource request 526 to network device 120. Network device 120 may receive 528 resource request 526 from first terminal device 110-1. Network device 120 may send 530 a resource allocation indication 532 to first terminal device 110-1. First terminal device 110-1 may receive 534 resource allocation indication 532 from network device 120.

[0088] First terminal device 110-1 may send 536 training data 538 to network device 120 based on resource allocation indication 532. Network device 120 may receive 540 training data 538 from first terminal device 110-1. Network device 120 may determine 542 dictionary update information 546 based on training data 538.

[0089] Network device 120 may send (544) dictionary update information 546 to first terminal device 110-1. First terminal device 110-1 may receive (548) dictionary update information 546 from network device 120. Network device 120 and first terminal device 110-1 may share (550) the new dictionary. Thus, network device 120 and first terminal device 110-1 may perform data compression and decompression based on the new dictionary. In this way, the accuracy of dictionary compression may be improved while reducing the amount of computation.

[0090] In some implementations, the first, second, and third communications devices may share the same dictionary. When sending dictionary update information, the first communications device may send the dictionary update information to the second and third communications devices. In other words, communications devices sharing the same dictionary may synchronously send training data to communications devices configured to perform dictionary update training. For example, the first communications device may be implemented by the network device 120 or its chip in Figure 1, and the second and third communications devices may be implemented by the terminal device 110 or its chip in Figure 1. For example, terminal devices 110 served by the same network device 120 may share the same dictionary. In some implementations, a group of communications devices may share the same dictionary. When sending dictionary update information, the first communications device may send the dictionary update information to communications devices within the same group. For example, terminal devices 110 served by the same network device 120 may be divided into multiple groups. Terminal devices within each group may share the same dictionary. In Figure 1, the first terminal device 110-1 and the second terminal device 110-2 may be included in the same group, and the third terminal device 110-3 may be included in another group. The second communication device may be implemented by the first terminal device 110 - 1 or a chip thereof in FIG. 1 , and the third communication device may be implemented by the second terminal device 110 - 2 or a chip thereof in FIG. 1 .

[0091] In some implementations, the first terminal device 110-1 and the second terminal device 110-2 may periodically send training data to the network device 120 so that the network device 120 can perform dictionary update training on the corresponding dictionaries. The training data received by the network device 120 may include training data received from the first terminal device 110-1 and training data received from the second terminal device 110-2.

[0092] Figure 6A shows a schematic diagram of an example implementation of a method for periodically updating a dictionary 600A according to some embodiments of the present application. For clarity, method 600A will be described in conjunction with Figure 1. Method 600A may involve first terminal device 110-1, second terminal device 110-2, and network device 120. Method 600A can be considered a specific implementation of method 500A shown in Figure 5A. Identical reference numerals represent identical steps or elements, and their detailed descriptions are omitted.

[0093] First terminal device 110-1, second terminal device 110-2, and network device 120 may share (602) a dictionary corresponding to data. For example, the data may be CSI. First terminal device 110-1 and second terminal device 110-2 are in a similar geographical location and in a similar channel state environment, and are therefore determined by network device 120 to be in the same group. In another implementation, all terminal devices served by network device 120 may share the same dictionary.

[0094] The second terminal device 110-2 may determine (604) that a dictionary update time has arrived. In some implementations, step 604 is performed simultaneously with step 504. The second terminal device 110-2 may send (606) training data 608 to the network device 120 using reserved transmission resources. In some implementations, step 606 is performed simultaneously with step 506. The network device 120 may receive (510) training data 508 and training data 608 from the first terminal device 110-1 and the second terminal device 110-2, respectively. The network device 120 may determine (512) dictionary update information 516 based on the training data 508 and training data 608. In some implementations, the network device 120 may determine the dictionary update information 516 based on the training data received from all terminal devices served by it. In another implementation, the network device 120 may determine the dictionary update information 516 based on the training data received from the terminal devices included in the same group.

[0095] The network device 120 may send (514) dictionary update information 516 to the first terminal device 110-1 and the second terminal device 110-2. In some implementations, the network device 120 may send the dictionary update information 516 to all terminal devices served by the network device 120 via broadcast. In another implementation, the network device 120 may send the dictionary update information 516 to the terminal devices in the same group via multicast.

[0096] Second terminal device 110-2 may receive (618) dictionary update information 516 from network device 120. In some implementations, step 618 is performed simultaneously with step 518. Network device 120, first terminal device 110-1, and second terminal device 110-2 may share (520) the new dictionary. As a result, network device 120, first terminal device 110-1, and second terminal device 110-2 may perform data compression and decompression based on the new dictionary. In this manner, signaling overhead may be reduced.

[0097] In some embodiments, a second communication device and a third communication device may communicate via a sidelink. For example, the second communication device may select a first training data set to be sent to the first communication device from a first data set acquired by the second communication device. The third communication device may select a second training data set to be sent to the first communication device from a second data set acquired by the third communication device. The second communication device may send the first training data set to the third communication device. The third communication device may update the second training data set based on the first training data set. For example, the third communication device may filter out data in the second training data set that is identical to the first training data set. For example, the third communication device may filter out data in the second training data set that is related to the first training data set. In another example, after receiving the first training data set, the third communication device may select a second training data set to be sent to the first communication device from a second data set acquired by the third communication device to reduce the correlation between the second training data set and the first training data set. In this way, the transmission resources required for the third communication device to send training data to the first communication device can be reduced, while also reducing the computational complexity of the dictionary update operation performed by the first communication device.

[0098] For example, the first terminal device 110-1 and the second terminal device 110-2 can communicate via a side link. Before sending the training data 608 to the network device 510, the second terminal device 110-2 can first receive the training data from the first terminal device 110-1. The training data can be part of the feature information of the training data 508. Alternatively, the training data can be a subset of the training data 508 or the training data 508 itself. The second terminal device 110-2 can filter the training data 608 based on the training data received from the first terminal device 110-1, select training data that is completely unrelated to the training data received from the first terminal device 110-1, and send the selected training data to the network device 120 for subsequent dictionary update operations by the network device 120. In this way, the transmission resources required for the second terminal device 110-2 to send the training data can be reduced.

[0099] In some implementations, at least one of the second and third communication devices may send a resource request to the first communication device based on a predetermined trigger condition for sending training data being met. The first communication device may send a resource allocation indication to at least one of the second and third communication devices based on the resource request received from at least one of the second and third communication devices. At least one of the second and third communication devices may send training data to the first communication device based on the resource allocation indication. In other words, the second and third communication devices sharing the same dictionary may asynchronously send training data to the first communication device. In another embodiment, the first communication device 210 may perform dictionary update training based on a predetermined trigger condition for dictionary update training being met.

[0100] In some embodiments, first terminal device 110-1 and second terminal device 110-2 may periodically send training data to network device 120 so that network device 120 can perform dictionary update training on the corresponding dictionaries. The training data received by network device 120 may include training data received from first terminal device 110-1 and training data received from second terminal device 110-2.

[0101] Figure 6B shows a schematic diagram of an example implementation of a method for aperiodic dictionary update 600B according to some embodiments of the present application. For clarity, method 600B will be described in conjunction with Figure 1. Method 600B may involve first terminal device 110-1, second terminal device 110-2, and network device 120. Method 600B can be considered as a specific implementation of method 500B shown in Figure 5B. The same reference numerals represent the same steps or elements, and their detailed descriptions are omitted.

[0102] First terminal device 110-1, second terminal device 110-2, and network device 120 may share (502) a dictionary corresponding to data (e.g., CSI). First terminal device 110-1 and second terminal device 110-2 are in a similar geographical location and in a similar channel state environment, and are therefore determined by network device 120 to be in the same group. In another implementation, all terminal devices served by network device 120 may share the same dictionary.

[0103] The network device 120 may send (544) dictionary update information 546 to the first terminal device 110-1 and the second terminal device 110-2. In some implementations, the network device 120 may send the dictionary update information 546 to all terminal devices served by the network device 120 via broadcast. In another implementation, the network device 120 may send the dictionary update information 546 to the terminal devices in the same group via multicast.

[0104] Second terminal device 110-2 may receive (648) dictionary update information 546 from network device 120. In some implementations, step 648 is performed simultaneously with step 548. Network device 120, second terminal device 110-2, and first terminal device 110-1 may share (550) the new dictionary. As a result, network device 120, first terminal device 110-1, and second terminal device 110-2 may perform data compression and decompression based on the new dictionary.

[0105] Based on determining that the conditions for triggering a dictionary update are met, the second terminal device 110-2 can trigger (622) a dictionary update operation. For example, the second terminal device 110-2 can estimate the NMSE or GCS of each data to be sent to the network device 120 after dictionary compression and decompression. When the NMSE of one or more data to be sent (for example, a predetermined number of data to be sent or a predetermined percentage of data to be sent) is higher than a set threshold or the GCS is lower than a set threshold, the dictionary update operation can be triggered. In some embodiments, the conditions for triggering a dictionary update of the first terminal device 110-1 and the second terminal device 110-2 can be the same. In some embodiments, the conditions for triggering a dictionary update of the first terminal device 110-1 and the second terminal device 110-2 can be different.

[0106] Second terminal device 110-2 may send 624 a resource request 626 to network device 120. Network device 120 may receive 628 the resource request 626 from second terminal device 110-2. Network device 120 may send 630 a resource allocation indication 632 to second terminal device 110-2. Second terminal device 110-2 may receive 634 the resource allocation indication 632 from network device 120. Second terminal device 110-2 may send 636 training data 638 to network device 120 based on the resource allocation indication 632.

[0107] Network device 120 may receive (640) training data 638 from second terminal device 110-2. Network device 120 may determine (642) dictionary update information 646 based on training data 638. Network device 120 may send (644) dictionary update information 646 to first terminal device 110-1 and second terminal device 110-2. In some implementations, network device 120 may send dictionary update information 646 to all terminal devices served by it via broadcast. In another implementation, network device 120 may send dictionary update information 646 to terminal devices in the same group via multicast.

[0108] First terminal device 110-1 may receive (548') dictionary update information 646 from network device 120. First terminal device 110-1 may receive (648') dictionary update information 646 from network device 120. In some implementations, step 648' is performed simultaneously with step 548'. Network device 120, first terminal device 110-1, and second terminal device 110-2 may share (650) the new dictionary. As a result, network device 120, first terminal device 110-1, and second terminal device 110-2 may perform data compression and decompression based on the new dictionary. In this way, the accuracy of dictionary compression may be improved while reducing the amount of computation.

[0109] In some implementations, when a first communication device sends multidimensional data to a second communication device, the first communication device may compress the multidimensional data based on multiple dictionaries and send the compressed data to the second communication device. The second communication device may decompress the received compressed data based on the same multiple dictionaries to obtain the multidimensional data. Alternatively or additionally, when the second communication device sends multidimensional data to the first communication device, the second communication device may compress the multidimensional data based on multiple dictionaries and send the compressed data to the first communication device. The first communication device may decompress the received compressed data based on the same multiple dictionaries to obtain the multidimensional data.

[0110] In some implementations, a communication device storing multiple dictionaries performs a dimensionality reduction operation on a multidimensional data matrix to be transmitted to obtain multiple sets of low-dimensional data. In some embodiments, the multiple dictionaries correspond to the multiple sets of low-dimensional data. For example, the low-dimensional data can be represented by a one-dimensional vector. The communication device can perform dictionary compression on the multiple sets of low-dimensional data based on the multiple dictionaries and transmit the multiple sets of dictionary-compressed low-dimensional data.

[0111] In some implementations, a communication device may receive multiple sets of dictionary-compressed low-dimensional data. The communication device may perform dictionary decompression on the multiple sets of dictionary-compressed low-dimensional data based on multiple dictionaries to obtain multiple sets of decompressed low-dimensional data. The communication device may perform a dimensionality increase operation on the multiple sets of low-dimensional data to obtain a multidimensional data matrix.

[0112] In some implementations, the multidimensional data matrix may include CSI, and the multiple groups of low-dimensional data may include at least one frequency domain description vector group and at least one spatial domain description vector group, wherein the number of frequency domain description vector groups in at least one frequency domain description vector group is the same as the number of spatial domain description vector groups in at least one spatial domain description vector group, and the number of frequency domain description vectors in the frequency domain description vector group is the same as the number of spatial domain description vectors in the spatial domain description vector group.

[0113] FIG7A shows a schematic diagram of an example implementation of a method 700A for compressing CSI according to some embodiments of the present application. The method 700A will be described in conjunction with FIG1 . The method 700A may be performed by the terminal device 110 in FIG1 .

[0114] As shown in FIG7A , the terminal device 110 obtains the CSI associated with the network device 120. The CSI may be a matrix of dimensions N. RX ×N TX ×N RB The three-dimensional matrix 702 represents, where N RX Indicates the number of receiving antennas of the terminal device 110, N TX N represents the number of transmitting antennas of the network device 120. RB The terminal device 110 performs SVD preprocessing on the data matrix corresponding to each resource block (RB) of the three-dimensional matrix 702 to perform a dimensionality reduction operation on the three-dimensional matrix 702, where each RB corresponds to a subcarrier or frequency domain signal, and the dimension of the data matrix corresponding to one RB is N. RX ×N TX For example, based on the SVD preprocessing for each RB, the terminal device 110 may convert the corresponding dimension N RX ×N TX The data matrix is ​​converted into r principal component right eigenvectors, i.e. r×N TX Based on the SVD preprocessing for all RBs, the terminal device 110 can RX ×N TX ×N RB The three-dimensional matrix 702 is converted to a dimension of r×N TX ×N RB It is understood that the terminal device 110 can also perform dimensionality reduction operations on the three-dimensional matrix 702 through other operations.

[0115] The terminal device 110 can generate a data matrix (dimension N) corresponding to each rank of the data matrix 704. TX ×N RB ) to perform a singular value decomposition (SVD) process on the three-dimensional matrix 704 to perform a dimensionality reduction operation. For example, based on the SVD process of each rank, the terminal device 110 can reduce the corresponding dimension N TX ×N RB The data matrix is ​​converted to obtain the eigenvectors of the most important components of d groups, and the eigenvectors of the most important components of each group include N dimensions. RB ×1 frequency description vector (FDV) 708 and dimension NTX Based on the SVD process for all ranks, the terminal device 110 can convert the spatial description vector (SDV) of dimension r×N into a spatial description vector (SDV) 710 of dimension r×N. TX ×N RB The three-dimensional matrix 704 is converted into r×d groups of feature vectors, each group of feature vectors may include FDV 708 and SDV 710. It is understood that the terminal device 110 may also perform dimensionality reduction operations on the three-dimensional matrix 706 through other operations.

[0116] In some embodiments, the terminal device 110 may perform a DFT codebook dimensionality reduction operation on the data matrix 704 to obtain a dimension of r×N′ TX ×N' RB A three-dimensional matrix 706, where N' TX ≤N TX , and N' RB ≤N RB The terminal device 110 can perform SVD operation on the three-dimensional matrix 706 to obtain multiple sets of dimensions N' RB ×1 feature vector FDV 708 and dimension N' TX ×1 feature vector SDV 710.

[0117] The terminal device 110 and the network device 120 can share the dictionary Dict for FDV FD And the dictionary Dict for SDV SD The terminal device 110 can be based on the dictionary Dict FD Compress r×d FDVs separately to obtain r×d×K1 group sequence numbers and corresponding weight coefficients, where each FDV can be represented by the dictionary Dict FD The dictionary columns and corresponding weight coefficients corresponding to the K1 group column numbers in the dictionary are weighted. K1 can be a pre-configured dictionary Dict FD Similarly, the terminal device 110 can compress the dictionary parameters based on the dictionary Dict SD Compress r×d SDVs to obtain r×d×K2 group sequence numbers and corresponding weight coefficients, where each SDV can be represented by the dictionary Dict SD The dictionary columns and corresponding weight coefficients corresponding to the K2 group column numbers in the dictionary are weighted. K2 can be a pre-configured dictionary Dict SD In some embodiments, K1 and K2 may be the same. Alternatively, K1 and K2 may be different. The specific implementation of dictionary compression will be described in detail below in conjunction with FIG. 7B .

[0118] Thus, the terminal device 110 can use the dictionary subset sequence number and coefficients to represent the three-dimensional matrix 702. In the example of method 700A, the dictionary subset sequence number and coefficients may include a dictionary based on the dictionary Dict FD The obtained r×d×K1 group sequence number and corresponding weight coefficient as well as the dictionary Dict SD The terminal device 110 may represent the dictionary subset sequence number and the coefficient as a bit sequence set and send the bit sequence set to the network device 120.

[0119] In some embodiments, the set of bit sequences totals bits, where L idx,i and L coeff,i are the number of bits required for a dictionary subset number (i.e., column number) and the number of bits required for a weighting coefficient corresponding to a dictionary subset, respectively, where L idx,1 and L coeff,1 Corresponding to the parameters of FDV, L idx,2 and L coeff,2 Corresponding to the parameters of SDV. In some embodiments, different numbers of quantization bits may be used for different coefficients.

[0120] The network device 120 can be based on the dictionary Dict FD and dictionary Dict SD The bit sequence set received from the terminal device 110 is converted into multiple sets of FDVs and SDVs, and the multiple sets of FDVs and SDVs are converted into N dimensions through corresponding dimensionality-raising operations. RX ×N TX ×N RB The three-dimensional matrix of is used to obtain CSI.

[0121] In this way, multiple sets of low-dimensional data can be obtained by transforming and reducing the dimensionality of high-dimensional data. Each set of low-dimensional data is compressed using a corresponding dictionary, thereby ensuring high accuracy of data compression while reducing the transmission resources required for data transmission.

[0122] FIG7B shows a schematic diagram of an example implementation of a dictionary compression method 700B according to some embodiments of the present application. The method 700B will be described in conjunction with FIG1 . The method 700B may be executed by the terminal device 110 in FIG1 .

[0123] As shown in FIG7B , for the data v to be compressed, the terminal device 110 may determine K dictionary subsets in the corresponding dictionary Dict, where the approximate value v' of the data v may be expressed as in represents the kth dictionary subset used to weight the data v, represents the kth dictionary subset used to weight the data v The corresponding weighting coefficient, ind k Can represent dictionary subsets The index in the dictionary Dict. ′ is determined to satisfy v and v ′ The difference between them is minimal. For example, v ′ It can be determined by minimizing NMSE or maximizing GCS.

[0124] FIG8A is a schematic diagram showing an example implementation of a bit sequence set according to some embodiments of the present application. The bit sequence set of FIG8A will be described in conjunction with the method 700A shown in FIG7A. In the method 700A, the terminal device 110 converts the three-dimensional matrix 702 into r×d groups of FDV bit sequences and r×d groups of SDV bit sequences, where each group of FDV bit sequences can represent K1 groups of column numbers and corresponding weighting coefficients for representing the corresponding FDV. Accordingly, each group of FDV bit sequences includes K1×(L coeff,1 +L idx,1 ) bits; Each group of SDV bit sequences can represent the K2 group sequence numbers and corresponding weighting coefficients used to represent the corresponding SDV. Accordingly, each group of SDV bit sequences includes K2×(L coeff,2 +L idx,2 The number of rank after conversion is r, where for each rank, there are d groups of FDV bit sequences and d groups of SDV bit sequences.

[0125] As shown in FIG8A , the bit sequence can be spliced ​​in the order of frequency domain components (FDVs) first and spatial components (SDVs). Within each component, the splicing can be performed in the order of rank. Under each rank, d groups of bit sequences are obtained. Each group of bit sequences corresponds to the information of one FDV or one SDV. Accordingly, each group of bit sequences contains K1 or K2 groups of coefficients and sequence number bit representations (respectively K1×(L coeff,1 +L idx,1 ) bits or K2×(L coeff,2 +L idx,2 ) bits).

[0126] For example, d groups of FDV bit sequences for each rank may be concatenated together, and d groups of SDV bit sequences for each rank may be concatenated together. In the order of ranks (e.g., Rank 1, Rank 2, ..., Rank r), d groups of FDV bit sequences for each rank may be concatenated together, thereby obtaining r×d groups of FDV bit sequences, which include r×d×K1×(L coeff,1 +L idx,1) bits. Similarly, the d groups of SDV bit sequences for each rank can be concatenated together in the order of rank (e.g., Rank 1, Rank 2, ..., Rank r), thereby obtaining r×d groups of SDV bit sequences, which include r×d×K2×(L coeff,2 +L idx,2 ) bits. r×d groups of FDV bit sequences and r×d groups of SDV bit sequences may be concatenated together to obtain a set of bit sequences to be transmitted.

[0127] FIG8B shows a schematic diagram of an example implementation of another bit sequence set according to some embodiments of the present application. The bit sequence set of FIG8B will be described in conjunction with the method 700A shown in FIG7A. In the method 700A, the terminal device 110 converts the three-dimensional matrix 702 into r×d groups of FDV bit sequences and r×d groups of SDV bit sequences, where each group of FDV bit sequences can represent K1 groups of column numbers and corresponding weighting coefficients for representing the corresponding FDV. Accordingly, each group of FDV bit sequences includes K1×(L coeff,1 +L idx,1 ) bits; Each group of SDV bit sequences can represent the K2 group sequence numbers and corresponding weighting coefficients used to represent the corresponding SDV. Accordingly, each group of SDV bit sequences includes K2×(L coeff,2 +L idx,2 ) bits. The rank number after conversion is r, where for each rank, there are d groups of FDV bit sequences (a total of d×K1×(L coeff,1 +L idx,1 ) bits) and d groups of SDV bit sequences (a total of d×K2×(L coeff,2 +L idx,2 ) bits).

[0128] As shown in FIG8B , the frequency domain components (FDVs) and spatial components (SDVs) corresponding to each rank may be concatenated, and then the concatenation may be performed in the order of the ranks. For example, d groups of FDV bit sequences for each rank may be concatenated together, and d groups of SDV bit sequences for each rank may be concatenated together. For each rank, d groups of FDV bit sequences and d groups of SDV bit sequences may be concatenated together, thereby obtaining a bit sequence for each rank, which includes d×K1×(L coeff,1 +L idx,1 )+d×K2×(L coeff,2 +L idx,2 The bit sequences for each rank may be concatenated together in the order of rank (eg, Rank 1, Rank 2, ..., Rank r), thereby obtaining a set of bit sequences to be transmitted.

[0129] It is understood that the packing method of the bit sequence set shown in Figures 8A and 8B is only illustrative and not intended to be limiting. Other suitable packing methods can be used. For example, in the bit sequence set shown in Figures 8A or 8B, the bits of the consecutive d groups of coefficients of the FDV or SDV for each rank (d × K1 × L coeff,1 bits or d×K2×L coeff,2 bits) are concatenated together, and the bits of the consecutive d groups of sequence numbers (d×K1×L idx,1 bits or d×K2×L idx,2 bits) are spliced ​​together and further spliced ​​to form the FDV bit sequence for each rank (a total of d×K1×(L coeff,1 +L idx,1 ) bits) and the SDV bit sequence for each rank (a total of d×K2×(L coeff,2 +L idx,2 ) bits). Alternatively, the bits of all coefficients can be concatenated together, and the bits of all sequence numbers can be concatenated together.

[0130] In some embodiments, a dimensionality reduction operation may be considered, and the parameters of the dimensionality reduction operation may be concatenated with the FDV / SDV bit sequence and sent together. For example, the sequence numbers of the spatial domain DFT orthogonal basis and the frequency domain DFT orthogonal basis may be represented in the form of a bitmap or a combination number.

[0131] In some embodiments, given the transmission resource constraints, the dictionary compression parameters can be dynamically adjusted. Returning to FIG. 7A , in method 700A, the terminal device 110 converts the three-dimensional matrix 702 into r×d groups of FDV bit sequences and r×d groups of SDV bit sequences, with a total of r×d×K1×(L coeff,1 +L idx,1 )+r×d×K2×(L coeff,2 +L idx,2 For example, assuming that CSI feedback resources are allocated in advance, the upper limit of the length of the compressed CSI bit sequence can be calculated as L max . According to L max To configure dictionary compression parameters.

[0132] For example, the rank number r can be pre-configured (e.g., by upper layer RRC signaling) or predetermined (e.g., a parameter calculated according to the channel state). coeff,1 +L idx,1 )+r×d×K2×(L coeff,2 +L idx,2 )≤L max Under the constraints of d, K1, K2, L coeff,1 、L coeff,2 、Lidx,1 and L idx,2 .

[0133] In some embodiments, in order to ensure similar compression and quantization effects on FDV and SDV, it is possible to first set K1 = K2 = K, L coeff,1 =L coeff,2 =L coeff , so that the above constraints are simplified to d×K×(2L coeff +L idx,1 +L idx,2 )≤L max / r.

[0134] In some embodiments, L idx,1 and L idx,2 Can be represented by Dict FD and Dict SD The number of columns determines and

[0135] In some embodiments, the effect of d on the GCS index after CSI data compression and reconstruction can be evaluated based on past training data, and the minimum d can be selected so that the mean value of GCS exceeds a preset threshold.

[0136] In some embodiments, under the constraint K×(2L coeff +L idx,1 +L idx,2 )≤L max Under / r / d, you can first give the number of coefficients K, and then set the quantization bit width L of the dictionary compression coefficients coeff .

[0137] Alternatively, different d can be set for different ranks, or different quantization bit widths L can be set for different coefficients. coeff .

[0138] In some embodiments, the terminal device 110 can configure the dictionary compression parameters d, K1, K2, L coeff,1 、L coeff,2 、L idx,1 and L idx,2 In some embodiments, the terminal device 110 may send the configured parameters to the network device 120 via an uplink channel.

[0139] In some embodiments, the dictionary compression parameters d, K1, K2, L can be determined by the network device 120. coeff,1 、L coeff,2 、L idx,1 and L idx,2and sends corresponding configuration information to the terminal device 110. For example, the network device 120 may send corresponding configuration information to the terminal device 110 via upper layer RRC signaling or MAC / PDCCH signaling to complete the configuration of the compression parameters.

[0140] Figures 9A to 9D illustrate exemplary implementations of information including dictionary compression parameters according to some embodiments of the present application. As shown in Figure 9A , the same dictionary compression parameters may be used for each rank. The number of dictionary compression coefficients for FDV and SDV is the same, and the quantization bit widths of the dictionary compression coefficients are the same.

[0141] As shown in Figure 9B, the same dictionary compression parameters may be used for each rank. The number of dictionary compression coefficients and the quantization bit width of the dictionary compression coefficients for FDV and SDV may be configured separately.

[0142] As shown in Figure 9C, the dictionary compression parameters for each rank can be configured separately. For each rank, the number of dictionary compression coefficients for FDV and SDV is the same, and the quantization bit width of the dictionary compression coefficients is the same. idx,1 and L idx,2 It can be determined by the number of dictionary columns and has nothing to do with the rank number.

[0143] As shown in Figure 9D, the dictionary compression parameters for each rank can be configured separately. For each rank, the number of dictionary compression coefficients for FDV and SDV can be configured separately, and the quantization bit width of the dictionary compression coefficients can be configured separately. idx,1 and L idx,2 It can be determined by the number of dictionary columns and has nothing to do with the rank number.

[0144] In some embodiments, at least a portion of the dictionary compression parameters may be preset or determined in advance. The information containing the dictionary compression parameters may only include the remaining parameters. For example, the number of quantization bits of the weighted coefficients obtained after dictionary compression (e.g., L coeff,1 、L coeff,2 ), for example, fixed to 8 bits. Once the size of the dictionary is determined, the number of index bits obtained after dictionary compression can be directly determined (for example, L idx,1 、L idx,2 In some embodiments, the rank number (e.g., r) can be obtained by evaluating the current channel quality. Alternatively, the rank number (e.g., r) can be determined in advance. If the above parameters are determined, only the number of subsets K used for dictionary compression and the number of feature vectors d retained by the dimensionality reduction operation need to be adjusted.

[0145] Figure 10 shows a schematic flow chart of a method 1000 implemented at a first communication device according to some embodiments of the present application. In one possible implementation, method 1000 can be implemented by the first communication device 210 in Figure 2 , which can be the terminal device 110 (or a chip of the terminal device 110) or the network device 120 (or a chip of the network device 120) in the example communication system 100 of Figure 1 . In other possible implementations, method 1000 can also be implemented by other electronic devices independent of the example communication system 100. As an example, the method 1000 will be described below using the first communication device 210 in Figure 2 as an example.

[0146] At 1020 , the first communication device 210 obtains training data, where the training data is used to train a dictionary for data compression.

[0147] At 1040 , the first communication device 210 determines a set of subsets to be updated in the dictionary based on the training data.

[0148] At 1060 , the first communication device 210 determines dictionary update information corresponding to a set of subsets to be updated.

[0149] At 1070 , the first communication device 210 sends dictionary update information.

[0150] In some embodiments, the dictionary update information may include: information indicating a subset of a set of subsets to be updated; and an update value for the subset of the set of subsets to be updated. The update value may be one of the following: an updated value of the subset; or a difference between the updated value of the subset and an initial value.

[0151] In some embodiments, the dictionary update information may include an update value compressed based on the compression control information. The compression control information may include at least one of the following: a quantization method, a quantization bit number, a transformation method, or an entropy coding method.

[0152] In some embodiments, the method may further include: determining compression control information; and sending the compression control information.

[0153] In some embodiments, the method may further include receiving compression control information.

[0154] In some embodiments, obtaining the training data may include receiving the training data from the second communication device.

[0155] In some embodiments, acquiring the training data may include periodically receiving the training data from the second communication device.

[0156] In some embodiments, the method may further comprise: receiving a resource request from the second communication apparatus; and sending an indication of the resource allocation to the second communication apparatus.

[0157] In some embodiments, the dictionary may be generated based on a basic training data set, and the basic training data set may be selected from at least one predetermined basic training data set.

[0158] In some embodiments, sending the dictionary update information may include sending the dictionary update information to the second communication apparatus and the third communication apparatus.

[0159] In some embodiments, the method may further include: performing a dimensionality reduction operation on the multidimensional data matrix to be sent to obtain multiple sets of low-dimensional data, wherein the first communication device stores multiple dictionaries including dictionaries, and the multiple dictionaries correspond to the multiple sets of low-dimensional data respectively; based on the multiple dictionaries, dictionary compression is performed on the multiple sets of low-dimensional data; and the multiple sets of low-dimensional data compressed by the dictionary are sent.

[0160] In some embodiments, the method may further include: receiving multiple sets of dictionary-compressed low-dimensional data; performing dictionary decompression on the multiple sets of dictionary-compressed low-dimensional data based on multiple dictionaries including a dictionary; and performing a dimensionality increase operation on the multiple sets of dictionary-decompressed low-dimensional data to obtain a multidimensional data matrix.

[0161] Figure 11 shows a schematic flow chart of a method 1100 implemented at a second communication device according to some embodiments of the present application. In one possible implementation, method 1100 can be implemented by the second communication device 220 in Figure 2, which can be the network device 120 (or a chip of the network device 120) or the terminal device 110 (or a chip of the terminal device 110) in the example communication system 100 of Figure 1. In other possible implementations, method 1100 can also be implemented by other electronic devices independent of the example communication system 100. As an example, the method 1100 will be described below using the second communication device 220 in Figure 2 as an example.

[0162] At 1120, the second communication device 220 receives dictionary update information for a dictionary used for data compression from the first communication device 210. The dictionary update information includes indication information of a subset in a set of subsets to be updated in the dictionary and an update value for the subset in the set of subsets to be updated.

[0163] At 1140 , the second communication device 220 updates the dictionary based on the dictionary update information.

[0164] In some embodiments, the dictionary update information may include: information indicating a subset of a set of subsets to be updated; and an update value for the subset of the set of subsets to be updated. The update value may be one of the following: an updated value of the subset; or a difference between the updated value of the subset and an initial value.

[0165] In some embodiments, the dictionary update information may include an update value compressed based on the compression control information. The compression control information may include at least one of the following: a quantization method, a quantization bit number, a transformation method, or an entropy coding method.

[0166] In some embodiments, the method may further include: determining compression control information; and sending the compression control information.

[0167] In some embodiments, the method may further include receiving compression control information.

[0168] In some embodiments, the method may further include: sending training data compressed based on compression control information to the first communication device, wherein the compression control information includes at least one of the following: quantization method, number of quantization bits, transformation method or entropy coding method.

[0169] In some embodiments, sending the compressed training data to the first communication device may include periodically sending the compressed training data to the first communication device.

[0170] In some embodiments, sending the compressed training data to the first communication device may include: sending a resource request to the first communication device; receiving an indication of resource allocation from the first communication device; and sending the compressed training data to the first communication device based on the indication of resource allocation.

[0171] In some embodiments, the resource request may be sent to the first communication device based on at least one of the following: the amount of training data reaches a predetermined threshold; or the percentage of training data in the data to be sent to the first communication device reaches a predetermined threshold.

[0172] In some embodiments, the dictionary may be generated based on a basic training data set, and the basic training data set may be selected from at least one predetermined basic training data set.

[0173] In some embodiments, the method may further include sending the training data to a third communication device.

[0174] In some embodiments, the training data is first training data, and the method further includes: receiving second training data from a third communication device, wherein the first training data is determined based on the second training data.

[0175] In some embodiments, the method may further include: performing a dimensionality reduction operation on the multidimensional data matrix to be sent to obtain multiple sets of low-dimensional data, wherein the first communication device stores multiple dictionaries including dictionaries, and the multiple dictionaries correspond to the multiple sets of low-dimensional data respectively; based on the multiple dictionaries, dictionary compression is performed on the multiple sets of low-dimensional data; and the multiple sets of low-dimensional data compressed by the dictionary are sent.

[0176] In some embodiments, the method may further include: receiving multiple sets of dictionary-compressed low-dimensional data; performing dictionary decompression on the multiple sets of dictionary-compressed low-dimensional data based on multiple dictionaries including a dictionary; and performing a dimensionality increase operation on the multiple sets of dictionary-decompressed low-dimensional data to obtain a multidimensional data matrix.

[0177] According to embodiments of the present application, efficient dynamic dictionary updates can be supported. For example, CSI-related dictionaries can adapt to time-varying channels. Some embodiments of the present application can provide an efficient data compression method based on a dynamic dictionary, thereby improving overall compression efficiency and saving bandwidth resources. In some embodiments, flexible and controllable fixed-length compression can be achieved, facilitating flexible matching of transmission resources.

[0178] Figure 12 is a schematic diagram of the structure of possible communication devices (also referred to as communication equipment) provided in an embodiment of the present disclosure. These communication devices can implement the functions of the terminal device or network device in the above method embodiments, and thus can also achieve the beneficial effects of the above method embodiments. In the embodiment of the present disclosure, the communication device can be the first communication device 210 or the second communication device 220 in Figure 2 or the terminal device 110 or network device 120 in Figure 1, or a module (such as a chip) thereof.

[0179] As shown in Figure 12, a communication device 1200 includes a processing unit 1210, a receiving unit 1220, and a sending unit 1230. The communication device can be used to implement the functions of dynamically updating a dictionary or compressing data based on a dictionary in the method embodiments shown in any of Figures 1 to 11 above. In some embodiments, the processing unit can be a processor, the sending unit can be a transmitter, and the receiving unit can be a receiver.

[0180] FIG13 is a simplified block diagram of a possible communication device (also referred to as a communication apparatus) provided in accordance with an embodiment of the present disclosure. As shown in FIG13 , the communication device 1300 includes a processor 1310 and an interface circuit 1320. The processor 1310 and the interface circuit 1320 are coupled to each other. It will be appreciated that the interface circuit 1320 may be a transceiver or an input / output interface. Optionally, the communication device 1300 may further include a memory 1330 for storing instructions executed by the processor 1310 or for storing input data required by the processor 1310 to execute instructions or for storing data generated after the processor 1310 executes instructions.

[0181] When the communication device 1300 is used to implement the method in the above method embodiment, the processor 1310 is used to execute the functions of the above processing unit 1210, and the interface circuit 1320 is used to execute the functions of the above receiving unit 1220 and the sending unit 1230.

[0182] It is understood that the processor in the embodiments of the present disclosure may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.

[0183] The embodiments of the present disclosure provide a communication system. The communication system may include the communication device involved in the embodiment shown in Figure 12 above, such as the first communication device 210 or the second communication device 220 in Figure 2 or the terminal device 110 or the network device 120 in Figure 1 or a module thereof (such as a chip). Optionally, the first communication device 210 or the second communication device 220 or the terminal device 110 or the network device 120 or a module thereof (such as a chip) in the communication system may execute any of the communication methods shown in Figures 1 to 11.

[0184] The present disclosure also provides a circuit that can be coupled to a memory and can be used to execute the process related to the first communication device 210 or the second communication device 220 or the terminal device 110 or the network device 120 or a module (such as a chip) thereof in any of the above method embodiments. The chip system may include the chip and may also include other components such as a memory or a transceiver.

[0185] It should be understood that the processor mentioned in the embodiments of the present disclosure may be a CPU, or may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0186] It should also be understood that the memory mentioned in the embodiments of the present disclosure may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM).

[0187] It should be noted that when the processor is a general-purpose processor, DSP, ASIC, FPGA or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, the memory (storage module) is integrated into the processor.

[0188] It should be noted that the memory described herein is intended to include, but not be limited to, these and any other suitable types of memory.

[0189] It should be understood that in the various embodiments of the present disclosure, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present disclosure.

[0190] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0191] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0192] In the several embodiments provided in the present disclosure, it should be understood that the disclosed communication methods and devices can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is merely a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0193] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed across multiple network elements. Some or all of these elements may be selected to achieve the purpose of this embodiment according to actual needs.

[0194] In addition, each functional module in each embodiment of the present disclosure may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0195] If this function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure, or the part that makes the contribution, or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method of each embodiment of the present disclosure. The aforementioned computer-readable storage medium can be any available medium that can be accessed by a computer. By way of example and not limitation, computer-readable media may include random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM), universal serial bus flash disk, mobile hard disk, or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer.

[0196] As used herein, the term "including" and similar terms should be understood as open inclusion, i.e., "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. can refer to different or the same objects and are only used to distinguish the objects referred to, and do not imply a specific spatial order, temporal order, order of importance, etc. of the objects referred to. In some embodiments, values, processes, selected items, determined items, devices, means, components, assemblies, etc. are referred to as "best", "lowest", "highest", "minimum", "maximum", etc. It should be understood that such descriptions are intended to indicate that a selection can be made from a number of available functional options, and that such a selection need not be better, lower, higher, smaller, larger, or otherwise preferred than other options in other aspects or all aspects. As used herein, the term "determine" can encompass a variety of actions. For example, "determine" can include calculating, computing, processing, deriving, investigating, searching (e.g., searching in a table, database, or another data structure), ascertaining, etc. Furthermore, "determining" may include receiving (eg, receiving information), accessing (eg, accessing data in a memory), etc. Furthermore, "determining" may include resolving, selecting, choosing, establishing, etc.

[0197] The embodiments may be further described using the following examples:

[0198] 1. A method comprising: obtaining, by a first communication device, training data for training a dictionary for data compression; determining, based on the training data, a set of subsets in the dictionary to be updated; determining dictionary update information corresponding to the set of subsets to be updated; and transmitting the dictionary update information. This method supports efficient dictionary updates and reduces computing and wireless transmission resources consumed by dictionary updates. This improves the accuracy of dictionary compression.

[0199] 2. The method of Example 1, wherein the set of subsets to be updated is determined based on update values ​​of subsets in a set of subsets to be trained in the dictionary, the set of subsets to be trained is determined based on the training data and the dictionary, and the update values ​​of the subsets are determined based on the training data and the set of subsets to be trained. This reduces computing resources consumed by dictionary update training.

[0200] 3. The method according to example 1 or 2, wherein the dictionary update information includes: information indicating a subset of the set of subsets to be updated; and an update value corresponding to the subset of the set of subsets to be updated, the update value being one of: an updated value of the subset; or a difference between the updated value and an initial value of the subset. This reduces wireless transmission resources required for sending the dictionary update information.

[0201] 4. The method of Example 3, wherein the dictionary update information includes the updated value compressed based on compression control information, wherein the compression control information includes at least one of: a quantization method, a number of quantization bits, a transform method, or an entropy coding method. This reduces resources consumed by transmitting the dictionary update information.

[0202] 5. The method according to example 4 further comprises: determining the compression control information; and sending the compression control information. In this way, the compression efficiency and accuracy of the dictionary compression information can be improved, thereby facilitating dictionary sharing.

[0203] 6. The method according to example 4 further comprises: receiving the compression control information. This can improve the compression efficiency and accuracy of the dictionary compression information, thereby facilitating dictionary sharing.

[0204] 7. The method according to any one of Examples 3 to 6, further comprising: updating the dictionary based on the set of subsets to be updated and the differential value corresponding to the subset in the set of subsets to be updated. This enables efficient dynamic dictionary updating, thereby improving the accuracy of dictionary compression.

[0205] 8. The method of any one of Examples 1 to 7, wherein obtaining the training data comprises: performing dictionary compression on the data to be transmitted based on the dictionary to obtain dictionary-compressed data; decompressing the dictionary-compressed data based on the dictionary to obtain reconstructed data; determining a loss function for the reconstructed data based on the data to be transmitted; and determining the training data based on the loss function for the reconstructed data. This reduces the computational resources consumed by dictionary update training.

[0206] 9. The method of example 8, wherein determining the training data includes determining that the to-be-sent data is included in the training data based on determining that the loss function of the reconstructed data meets a loss function threshold. This improves the accuracy of dictionary compression while reducing the computational resources consumed by dictionary update training.

[0207] 10. The method of example 8, wherein determining the training data comprises:

[0208] The plurality of data to be transmitted is sorted based on the loss function of the reconstructed data corresponding to data in the plurality of data to be transmitted; and the training data is selected from the plurality of data to be transmitted based on the sorting. This improves the accuracy of dictionary compression while reducing the computing resources consumed by dictionary update training.

[0209] 11. The method of any one of Examples 1 to 7, wherein obtaining the training data comprises: receiving compressed data corresponding to the training data from a second communication device; and decompressing the compressed data based on compression control information to obtain the training data, wherein the compression control information includes at least one of: a quantization method, a number of quantization bits, a transform method, or an entropy coding method. Thus, by compressing the training data, transmission bandwidth consumption of the training data can be reduced.

[0210] 12. The method of any one of Examples 1 to 7 and 11, wherein obtaining the training data comprises periodically receiving the training data from a second communication device. This improves the accuracy of dictionary compression.

[0211] 13. The method of any one of Examples 1 to 7 and 11, further comprising: receiving a resource request from a second communication device; and sending a resource allocation indication to the second communication device. This improves the accuracy of dictionary compression while reducing the computational resources consumed by dictionary update training.

[0212] 14. The method of any one of Examples 1 to 7 and 11 to 13, wherein sending the dictionary update information comprises sending the dictionary update information to the second communication device and the third communication device. This allows the dictionary to be shared among a group of communication devices, thereby reducing computing resources consumed by the dynamic dictionary.

[0213] 15. The method of example 14, wherein the training data comprises at least one of: first training data received from the second communication device; or second training data received from the third communication device. Thus, the dynamic dictionary can be adapted to a group of communication devices.

[0214] 16. The method according to any one of Examples 1 to 15, further comprising: receiving a set of bit sequences; determining, based on the set of bit sequences, a plurality of dictionary subsets and a plurality of weighting coefficients corresponding to a plurality of dictionaries including the dictionary; determining, based on the plurality of dictionaries, a plurality of sets of low-dimensional data, wherein the low-dimensional data in the plurality of sets of low-dimensional data can be weightedly represented based on a set of dictionary subsets in the plurality of sets of dictionary subsets and a set of weighting coefficients in the plurality of sets of weighting coefficients; and performing a dimensionality increase operation on the plurality of sets of low-dimensional data to obtain a multidimensional data matrix. Thus, a dynamic dictionary can be adapted to a group of communication devices.

[0215] 17. The method according to any one of Examples 1 to 16 further includes: performing a dimensionality reduction operation on the multidimensional data matrix to be sent to obtain multiple sets of low-dimensional data, wherein the first communication device stores multiple dictionaries including the dictionary, and the multiple dictionaries respectively correspond to the multiple sets of low-dimensional data; based on the multiple dictionaries, dictionary compression is performed on the multiple sets of low-dimensional data to determine at least one dictionary subset and corresponding weighting coefficient in the dictionary for low-dimensional data in a set of low-dimensional data in the multiple sets of low-dimensional data, and the low-dimensional data can be weighted based on the at least one dictionary subset and the weighting coefficient; based on the at least one dictionary subset corresponding to the low-dimensional data and the weighting coefficient, a set of bit sequences corresponding to the multiple sets of low-dimensional data is determined; and sending the set of bit sequences. In this way, dictionary decompression of high-dimensional data can be achieved, thereby improving data compression performance while reducing the transmission resources required for data transmission.

[0216] 18. The method of example 17, wherein the dimensionality reduction operation comprises at least one of: a singular value decomposition operation; a tensor decomposition operation; a low-rank decomposition operation; and a discrete Fourier transform codebook dimensionality reduction operation. Thus, the transmission resources required for data transmission can be reduced.

[0217] 19. The method of example 17 or 18, wherein the multidimensional data matrix includes channel state information; wherein the multiple groups of low-dimensional data include at least one frequency domain description vector group and at least one spatial domain description vector group, the number of frequency domain description vector groups in the at least one frequency domain description vector group is the same as the number of spatial domain description vector groups in the at least one spatial domain description vector group, and the number of frequency domain description vectors in the frequency domain description vector group is the same as the number of spatial domain description vectors in the spatial domain description vector group. Thus, dictionary compression of the channel state information can be achieved.

[0218] 20. The method of example 19, wherein determining the set of bit sequences comprises: for a frequency domain description vector group in the at least one frequency domain description vector group, concatenating a frequency domain indication information bit sequence of indication information of the at least one subset corresponding to the frequency domain description vectors in the frequency domain description vector group and a corresponding frequency domain coefficient bit sequence of the weighting coefficients, thereby determining at least one frequency domain bit sequence corresponding to the at least one frequency domain description vector group; and for a spatial domain description vector group in the at least one spatial domain description vector group, concatenating a spatial domain indication information bit sequence of indication information of the at least one subset corresponding to the spatial domain description vectors in the spatial domain description vector group and a corresponding spatial domain coefficient bit sequence of the weighting coefficients, thereby determining at least one spatial domain bit sequence corresponding to the at least one spatial domain description vector group. Thus, dictionary compression of channel state information can be achieved.

[0219] 21. The method of Example 20, wherein determining the set of bit sequences further comprises: determining a total frequency-domain bit sequence by concatenating the at least one frequency-domain bit sequence corresponding to the at least one frequency-domain description vector group; determining a total spatial-domain bit sequence by concatenating the at least one spatial-domain bit sequence corresponding to the at least one spatial-domain description vector group; and determining the set of bit sequences by concatenating the total frequency-domain bit sequence and the total spatial-domain bit sequence. Thus, transmission resources required for data transmission can be reduced, and the accuracy of dictionary compression and decompression can be improved.

[0220] 22. The method of example 20, wherein determining the set of bit sequences further comprises: determining at least one frequency-domain and spatial-domain bit sequence by concatenating a frequency-domain bit sequence in the at least one frequency-domain bit sequence with a corresponding spatial-domain bit sequence in the at least one spatial-domain bit sequence; and determining the set of bit sequences by concatenating the at least one frequency-domain and spatial-domain bit sequence. This reduces transmission resources required for data transmission and improves the accuracy of dictionary compression and decompression.

[0221] 23. The method of any one of Examples 17 to 22, further comprising: determining a compression parameter for the dictionary compression based on a predetermined bit sequence set transmission resource. This facilitates reserving transmission resources for the dictionary-compressed data, thereby saving signaling overhead and improving data transmission efficiency.

[0222] 24. The method according to any one of Examples 20 to 22, further comprising: determining an upper limit on the length of the bit sequence set based on a predetermined bit sequence set transmission resource; determining the number of frequency domain description vector groups in the at least one frequency domain description vector group; and determining a compression parameter for the dictionary compression based on the upper limit on the length of the bit sequence set and the number of frequency domain description vector groups in the at least one frequency domain description vector group. This improves the flexibility of the dictionary compression operation, thereby flexibly matching transmission resources.

[0223] 25. The method according to any one of Examples 20 to 22, further comprising: receiving compression parameters for the dictionary compression. This improves the flexibility of the dictionary compression operation, thereby flexibly matching transmission resources.

[0224] 26. The method according to example 24 or 25, wherein the compression parameter comprises at least one of the following: the number of frequency domain description vectors in the frequency domain description vector group; the number of subsets of the at least one subset corresponding to the frequency domain description vector; the number of subsets of the at least one subset corresponding to the spatial domain description vector; the length of the frequency domain indication information bit sequence; the length of the frequency domain coefficient bit sequence; the length of the spatial domain indication information bit sequence; or the length of the spatial domain coefficient bit sequence. Thus, the flexibility of the dictionary compression operation can be improved, thereby flexibly matching transmission resources.

[0225] 27. The method of any one of Examples 1 to 26, wherein the dictionary is generated based on a base training data set selected from at least one predetermined base training data set. In this manner, communication devices sharing a dictionary can directly generate a dictionary for a predetermined scenario without having to obtain a large amount of training data before each dictionary generation.

[0226] 28. A method comprising: a second communication device receiving dictionary update information for a dictionary used for data compression from a first communication device, wherein the dictionary update information includes indication information of a subset in a set of subsets to be updated in the dictionary and an update value corresponding to the subset in the set of subsets to be updated; and updating the dictionary based on the dictionary update information.

[0227] 29. The method of example 28, wherein the updated value is one of: an updated value of the subset; or a difference between the updated value and an initial value of the subset.

[0228] 30. A method according to Example 28 or 29, wherein the dictionary update information includes the updated value compressed based on compression control information, and the compression control information includes at least one of the following: quantization method, number of quantization bits, transformation method or entropy coding method.

[0229] 31. The method of Example 30, further comprising: receiving the compression control information.

[0230] 32. The method of Example 30 further comprising: determining the compression control information; and sending the compression control information.

[0231] 33. The method according to any one of Examples 28 to 32 further includes: determining training data; compressing the compressed data based on compression control information, wherein the compression control information includes at least one of the following: quantization method, number of quantization bits, transformation method or entropy coding method; and sending the compressed training data to the first communication device.

[0232] 34. The method of example 33, wherein determining the training data comprises: dictionary-compressing data to be sent to the first communication device based on the dictionary to obtain dictionary-compressed data; decompressing the dictionary-compressed data based on the dictionary to obtain reconstructed data; and determining a loss function for the reconstructed data based on the data to be sent; and

[0233] The training data is determined based on the loss function of the reconstructed data.

[0234] 35. A method according to Example 34, wherein determining the training data based on the loss function includes: determining the data to be sent as being included in the training data based on determining that the loss function of the reconstructed data reaches a loss function threshold.

[0235] 36. A method according to Example 34, wherein determining the training data based on the loss function includes: based on the loss function of the reconstructed data corresponding to the data in the multiple data to be sent to the first communication device, the first communication device sorts the multiple data to be sent; and based on the sorting, selects the training data from the multiple data to be sent.

[0236] 37. A method according to any one of Examples 33 to 36, wherein sending the compressed training data to the first communication device includes: periodically sending the compressed training data to the first communication device.

[0237] 38. A method according to any one of Examples 33 to 36, wherein sending the compressed training data to the first communication device includes: sending a resource request to the first communication device; receiving an indication of resource allocation from the first communication device; and sending the compressed training data to the first communication device based on the indication of resource allocation.

[0238] 39. A method according to Example 38, wherein a resource request is sent to the first communication device based on at least one of the following: the number of training data reaches a predetermined number threshold; or the percentage of the training data in the data to be sent to the first communication device reaches a predetermined percentage threshold.

[0239] 40. The method of any one of Examples 33 to 39, further comprising: sending the training data to a third communication device, whereby the second communication device and the third communication device can adjust the training data to be sent to the first communication device, thereby saving transmission resources and computing resources.

[0240] 41. The method of any one of Examples 33 to 40, wherein the training data is first training data, the method further comprising: receiving second training data from a third communication device, wherein the first training data is determined based on the second training data. This enables the second and third communication devices to adjust the training data to be sent to the first communication device, thereby saving transmission resources and computing resources.

[0241] 42. The method according to any one of Examples 28 to 41 further includes: receiving a set of bit sequences; determining, based on the set of bit sequences, multiple groups of dictionary subsets and multiple groups of weighting coefficients corresponding to multiple dictionaries including the dictionary; determining, based on the multiple dictionaries, multiple groups of low-dimensional data, wherein the low-dimensional data in the multiple groups of low-dimensional data can be weightedly represented based on a group of dictionary subsets in the multiple groups of dictionary subsets and a group of weighting coefficients in the multiple groups of weighting coefficients; and performing a dimensionality increase operation on the multiple groups of low-dimensional data to obtain a multidimensional data matrix.

[0242] 43. The method according to any one of Examples 28 to 42 further includes: performing a dimensionality reduction operation on a multidimensional data matrix to be sent to the first communication device to obtain multiple sets of low-dimensional data, wherein the second communication device stores multiple dictionaries including the dictionary, and the multiple dictionaries respectively correspond to the multiple sets of low-dimensional data; based on the multiple dictionaries, dictionary compression is performed on the multiple sets of low-dimensional data to determine at least one dictionary subset and corresponding weighting coefficient in the dictionary for low-dimensional data in a group of low-dimensional data in the multiple sets of low-dimensional data, and the low-dimensional data can be weightedly represented based on the at least one dictionary subset and the weighting coefficient; based on the at least one dictionary subset corresponding to the low-dimensional data and the weighting coefficient, a set of bit sequences corresponding to the multiple sets of low-dimensional data is determined; and the set of bit sequences is sent to the first communication device.

[0243] 44. A method according to example 43, wherein the dimensionality reduction operation includes at least one of the following: a singular value decomposition operation; a tensor decomposition operation; a low rank decomposition operation; and a discrete Fourier transform codebook dimensionality reduction operation.

[0244] 45. A method according to Example 43 or 44, wherein the multidimensional data matrix includes channel state information associated with the first communication device; wherein the multiple groups of low-dimensional data include at least one frequency domain description vector group and at least one spatial domain description vector group, the number of frequency domain description vector groups in the at least one frequency domain description vector group is the same as the number of spatial domain description vector groups in the at least one spatial domain description vector group, and the number of frequency domain description vectors in the frequency domain description vector group is the same as the number of spatial domain description vectors in the spatial domain description vector group.

[0245] 46. ​​A method according to Example 45, wherein determining the bit sequence set includes: for a frequency domain description vector group in the at least one frequency domain description vector group, determining at least one frequency domain bit sequence corresponding to the at least one frequency domain description vector group by splicing the frequency domain indication information bit sequence of the indication information of the at least one subset corresponding to the frequency domain description vector in the frequency domain description vector group and the frequency domain coefficient bit sequence of the corresponding weighting coefficient; and for a spatial domain description vector group in the at least one spatial domain description vector group, determining at least one spatial domain bit sequence corresponding to the at least one spatial domain description vector group by splicing the spatial domain indication information bit sequence of the indication information of the at least one subset corresponding to the spatial domain description vector in the spatial domain description vector group and the spatial domain coefficient bit sequence of the corresponding weighting coefficient.

[0246] 47. A method according to Example 46, wherein determining the bit sequence set further includes: determining a total frequency domain bit sequence by splicing the at least one frequency domain bit sequence corresponding to the at least one frequency domain description vector group; determining a total spatial domain bit sequence by splicing the at least one spatial domain bit sequence corresponding to the at least one spatial domain description vector group; and determining the bit sequence set by splicing the total frequency domain bit sequence and the total spatial domain bit sequence.

[0247] 48. A method according to Example 46, wherein determining the bit sequence set further includes: determining at least one frequency domain spatial domain bit sequence by splicing the frequency domain bit sequence in the at least one frequency domain bit sequence with the corresponding spatial domain bit sequence in the at least one spatial domain bit sequence; and determining the bit sequence set by splicing the at least one frequency domain spatial domain bit sequence.

[0248] 49. The method according to any one of Examples 43 to 48 further includes: determining compression parameters for the dictionary compression based on a predetermined bit sequence set transmission resource.

[0249] 50. The method according to any one of Examples 46 to 48 further includes: determining an upper limit on the length of the bit sequence set based on a predetermined bit sequence set transmission resource; determining the number of frequency domain description vector groups in the at least one frequency domain description vector group; and determining a compression parameter for the dictionary compression based on the upper limit on the length of the bit sequence set and the number of frequency domain description vector groups in the at least one frequency domain description vector group.

[0250] 51. The method of any one of Examples 46 to 48, further comprising receiving compression parameters for the dictionary compression from the first communication device.

[0251] 52. A method according to Example 50 or 51, wherein the compression parameter includes at least one of the following: the number of frequency domain description vectors in the frequency domain description vector group; the number of subsets of the at least one subset corresponding to the frequency domain description vector; the number of subsets of the at least one subset corresponding to the spatial domain description vector; the length of the frequency domain indication information bit sequence; the length of the frequency domain coefficient bit sequence; the length of the spatial domain indication information bit sequence; or the length of the spatial domain coefficient bit sequence.

[0252] 53. A method according to any one of Examples 28 to 52, wherein the dictionary is generated based on a base training data set, and the base training data set is selected from at least one predetermined base training data set.

[0253] 54. A first communication device, characterized in that it includes a module or unit for executing the method described in any one of Examples 1 to 27.

[0254] 55. A second communication device, characterized in that it includes a module or unit for executing the method described in any one of Examples 28 to 53.

[0255] 56. A first communication device comprises a processor, the processor is coupled to a memory, the memory stores instructions, and when the instructions are executed by the processor, the first communication device performs the method according to any one of Examples 1 to 27.

[0256] 57. A second communication device comprises a processor, the processor is coupled to a memory, the memory stores instructions, and when the instructions are executed by the processor, the second communication device performs the method according to any one of Examples 28 to 53.

[0257] 58. A computer-readable storage medium storing instructions that, when executed, cause the method of any one of Examples 1 to 53 to be performed.

[0258] 59. A computer program product comprising instructions that, when executed, cause the method of any one of Examples 1 to 53 to be performed.

[0259] 60. A communication system, characterized in that it includes the first communication device described in Example 54 or 56 and the second communication device described in Example 55 or 57.

[0260] The above is only a specific embodiment of the present disclosure, but the scope of protection of the embodiments of the present disclosure is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the embodiments of the present disclosure should be included in the scope of protection of the embodiments of the present disclosure. Therefore, the scope of protection of the embodiments of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A method comprising: The first communication device acquires training data, wherein the training data is used to train a dictionary for data compression; Based on the training data, determining a set of subsets to be updated in the dictionary; Determining dictionary update information corresponding to the set of subsets to be updated; as well as The dictionary update information is sent.

2. A method comprising: The second communication device receives dictionary update information for a dictionary used for data compression from the first communication device, wherein the dictionary update information includes indication information of a subset in a set of subsets to be updated in the dictionary and an update value for the subset in the set of subsets to be updated; as well as Based on the dictionary update information, the dictionary is updated.

3. The method according to any one of claims 1 or 2, wherein the dictionary update information comprises: Indicative information of a subset in the set of subsets to be updated; as well as For the update value of a subset in the set of subsets to be updated, the update value is one of the following: an updated value of the subset; or a difference value between the updated value of the subset and an initial value.

4. The method according to claim 3, wherein the dictionary update information includes the updated value compressed based on compression control information, and the compression control information includes at least one of the following: quantization method, number of quantization bits, transformation method or entropy coding method.

5. The method according to claim 4, further comprising: determining the compression control information; as well as The compression control information is sent.

6. The method according to claim 4, further comprising: The compression control information is received.

7. The method according to any one of claims 1 or 3 to 6, wherein obtaining the training data comprises: The training data is received from a second communication device.

8. The method according to any one of claims 1 or 3 to 7, wherein obtaining the training data comprises: The training data is periodically received from the second communication device.

9. The method according to any one of claims 1 or 3 to 7, further comprising: receiving a resource request from a second communication device; as well as An indication of resource allocation is sent to the second communication device.

10. The method according to any one of claims 1 or 3 to 9, wherein sending the dictionary update information comprises: The dictionary update information is sent to the second communication device and the third communication device.

11. The method according to any one of claims 2 to 6, further comprising: Training data compressed based on compression control information is sent to the first communication device, wherein the compression control information includes at least one of the following: a quantization method, a number of quantization bits, a transformation method, or an entropy coding method.

12. The method of claim 11, wherein sending the compressed training data to the first communication device comprises: The compressed training data is periodically sent to the first communication device.

13. The method according to any one of claims 11 or 12, wherein sending the compressed training data to the first communication device comprises: sending a resource request to the first communication device; receiving an indication of resource allocation from the first communication device; as well as The compressed training data is sent to the first communications device based on the indication of the resource allocation.

14. The method of claim 13, wherein sending the resource request to the first communication device is based on at least one of: The number of the training data reaches a predetermined number threshold; or The percentage of the training data to the data to be sent to the first communication device reaches a predetermined percentage threshold.

15. The method according to any one of claims 11 to 14, further comprising: The training data is sent to a third communication device.

16. The method according to any one of claims 11 to 15, wherein the training data is first training data, the method further comprising: Second training data is received from a third communication device, wherein the first training data is determined based on the second training data. 17 . The method according to claim 1 , wherein the dictionary is generated based on a base training data set, the base training data set being selected from at least one predetermined base training data set.

18. The method according to any one of claims 1 to 17, further comprising: Performing a dimensionality reduction operation on the multidimensional data matrix to be sent to obtain multiple sets of low-dimensional data, wherein the first communication device stores multiple dictionaries including the dictionary, and the multiple dictionaries correspond to the multiple sets of low-dimensional data respectively; Based on the multiple dictionaries, dictionary compression is performed on the multiple groups of low-dimensional data; and The plurality of sets of low-dimensional data compressed by the dictionary are sent.

19. The method according to any one of claims 1 to 17, further comprising: receiving a plurality of sets of low-dimensional data compressed by a dictionary; Based on a plurality of dictionaries including the dictionary, dictionary decompressing the plurality of groups of low-dimensional data compressed by the dictionary; as well as A dimension-increasing operation is performed on the multiple groups of low-dimensional data decompressed by the dictionary to obtain a multi-dimensional data matrix.

20. A first communication device, characterized in that: Comprising modules or units for executing the method of any one of claims 1 or 3 to 11 or any one of 18 to 19.

21. A second communication device, characterized in that: The method comprises a module or a unit for executing the method of any one of claims 2 to 6 or any one of claims 12 to 19.

22. A first communication device, comprising a processor, the processor being coupled to a memory, the memory storing instructions, and when the instructions are executed by the processor, causing the first communication device to execute the method according to any one of claims 1 or 3 to 11 or any one of 18 to 19.

23. A second communication device, comprising a processor, wherein the processor is coupled to a memory, wherein the memory stores instructions, and when the instructions are executed by the processor, the second communication device executes the method according to any one of claims 2 to 6 or any one of claims 12 to 19.

24. A computer-readable storage medium storing instructions, wherein the instructions, when executed, cause the method of any one of claims 1 to 19 to be executed.

25. A computer program product comprising instructions which, when executed, cause the method of any one of claims 1 to 19 to be performed.

26. A communication system, characterized in that: Includes the first communication device described in claim 20 or 22 and the second communication device described in claim 21 or 23.

Images