Method and device for communication, equipment, storage medium and program product

Abstract

The embodiment of the invention provides a communication method and device, a storage medium and a program product. In the method, a data compression device transmits a first data matrix. Furthermore, the data compression device transmits a projection result and a projection residual error of the second data matrix. Wherein the projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual error is determined according to the second data matrix and the projection result. Thus, the correlation between the first data matrix and the second data matrix can be fully utilized, the compression efficiency is improved, and the transmission bandwidth or storage resources are saved.

Description

Technical Field

[0001] Embodiments of the present application generally relate to the field of communications, and more specifically to a method for communications, a terminal device, a network device, a computer-readable storage medium, and a computer program product. Background Art

[0002] Air interface native data may include synaesthesia data, artificial intelligence (AI) model data, and channel state information (CSI) data of multi-antenna systems. Native data has high dimensions and large data volume, and its interaction and transmission will occupy a large amount of air interface resources. Data compression is required to reduce the occupation of air interface resources or save storage resources. Other data, such as application interaction data in the Internet, also need to be compressed to save network bandwidth or storage resources. Summary of the invention

[0003] The embodiments of the present application provide a technical solution for data compression and decompression, which can provide higher compression efficiency and save transmission resources. In addition to being applicable to air interface native data, the embodiments of the present disclosure can also be applicable to, for example, Internet interactive data or other data, and can be widely used in terminal devices or network side devices in future wireless communication scenarios.

[0004] In the first aspect, a communication method is provided. The method can be executed by a data compression device. Unless otherwise specified, the data compression device in the embodiment of the present application can refer to the data compression device itself (for example, implemented as a terminal device, a network device), or a component in the data compression device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can realize all or part of the functions of the data compression device. The following is described by taking the execution subject as an example of a data compression device. In this method, the data compression device sends a first data matrix. Furthermore, the data compression device sends the projection result and projection residual of the second data matrix. The projection result is determined according to the projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result. In this way, the correlation between the first data matrix and the second data matrix can be fully utilized to improve the compression efficiency and save transmission bandwidth or storage resources.

[0005] In the second aspect, a communication method is provided. The method can be executed by a data decompression device. Unless otherwise specified, the data decompression device in the embodiment of the present application can refer to the data decompression device itself (for example, implemented as a terminal device, a network device), or a component in the data decompression device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can realize all or part of the functions of the data decompression device. The following is described as an example in which the execution subject is a data decompression device. In this method, the data decompression device receives a first data matrix. Furthermore, the data decompression device receives the projection result and the projection residual of the second data matrix. Furthermore, the data decompression device obtains the second data matrix or an approximate matrix of the second data matrix based on the first data matrix, the projection result and the projection residual. The projection result is determined according to the projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result. In this way, the data compressed by the data compression device can be decompressed and restored, and the correlation between the first data matrix and the second data matrix is ​​fully utilized to improve the compression efficiency and save transmission bandwidth or storage resources.

[0006] In some implementations, the projection matrix is ​​determined in the following manner: the data compression device selects a predetermined number of columns in a reference matrix to form the projection matrix, and the reference matrix is ​​the first data matrix or a submatrix of the first data matrix. In this way, the projection matrix can be constructed in a simple manner, saving computational effort.

[0007] In some implementations, the projection matrix is ​​determined in the following manner: the data compression device determines a projection base matrix based on a reference matrix, and selects a predetermined number of columns in the projection base matrix to form a projection matrix. The reference matrix is ​​a first data matrix or a submatrix of the first data matrix. In this way, a higher energy basis can be extracted through LRMA, the features of the reference matrix can be fully extracted, and the compression error can be reduced.

[0008] In some implementations, determining the projection base matrix according to the reference matrix includes: performing matrix decomposition on the reference matrix to determine the projection base matrix. In this way, the features of the reference matrix can be fully extracted and the compression error can be reduced.

[0009] In some implementations, the matrix decomposition includes any one of the following: low rank matrix approximation (LRMA) decomposition, SVD decomposition, or QR decomposition. In this way, the matrix decomposition can be flexibly implemented.

[0010] In some implementations, the data compression device further sends one or more of the following: first indication information for indicating the number of columns of the projection matrix, second indication information for indicating whether the projection matrix is ​​based on a reference matrix or a projection base matrix obtained by performing LRMA decomposition on the reference matrix, or third indication information for indicating the position of the columns of the projection matrix in the reference matrix or the projection base matrix. In this way, synchronization between the data compression device and the data decompression device can be achieved, which is conducive to accurate decompression.

[0011] In some implementations, the data decompression device further receives one or more of the following: first indication information for indicating the number of columns of the projection matrix, second indication information for indicating whether the projection matrix is ​​based on a reference matrix or a projection base matrix obtained by performing LRMA decomposition on the reference matrix, or third indication information for indicating the position of the columns of the projection matrix in the reference matrix or the projection base matrix. In this way, synchronization between the data compression device and the data decompression device can be achieved, which is conducive to accurate decompression.

[0012] In some implementations, the data compression device determines the projection matrix, including: the data compression device selects a subspace of a predetermined number of dimensions in a column space spanned by a reference matrix, wherein the projection matrix represents the subspace, and wherein the reference matrix is ​​a first data matrix or a submatrix of the first data matrix. In this way, the second data matrix can be accurately fitted to reduce compression errors.

[0013] In some implementations, the subspace is selected so that the norm of the projection residual after the second data matrix is ​​projected onto the projection matrix is ​​minimized. In this way, the second data matrix can be accurately fitted and the compression error can be reduced.

[0014] In some implementations, the data compression device determines the projection matrix, including: the data compression device performs QR decomposition on the reference matrix to determine the Q matrix. Further, the data compression device performs singular value decomposition on the transpose of the Q matrix and the product of the data matrix to obtain the eigenvector matrix. Further, the data compression device determines the subspace selection matrix based on the first predetermined number of columns of the eigenvector matrix. Further, the data compression device determines the projection matrix based on the subspace selection matrix and the reference matrix. In this way, the generated projection matrix can accurately fit the second data matrix and reduce the compression error.

[0015] In some implementations, the data compression device determines the projection matrix including: the data compression device performs QR decomposition on the reference data C to obtain a Q matrix, where the formula of the QR decomposition is C=Q×R, The columns are orthogonal, Furthermore, the data compression device performs the matrix Q T G performs SVD decomposition to obtain the U matrix, where G is the second data matrix, and the formula is Q T G=U∑VT Then, the data compression device obtains the first d columns of the matrix U, which are recorded as Then, the data compression device obtains the subspace selection matrix Then, the data compression device obtains the projection matrix P * =C×S * In this way, the generated projection matrix can accurately fit the second data matrix and reduce the compression error.

[0016] In some implementations, the data compression device determines the projection matrix including: the data compression device solves the following optimization problem, where G is the second data matrix, C is the reference matrix In this way, the generated projection matrix can accurately fit the second data matrix and reduce compression errors.

[0017] In some implementations, the method performed by the data compression device further includes: sending fourth indication information, the fourth indication information being used to indicate a subspace selection matrix, wherein the subspace selection matrix is ​​used to determine a projection matrix together with a reference matrix. In this way, the subspace selection matrix can be accurately indicated to the data decompression device, facilitating accurate decompression.

[0018] In some implementations, the method performed by the data decompression device further includes: receiving fourth indication information, the fourth indication information being used to indicate a subspace selection matrix, wherein the subspace selection matrix is ​​used to determine a projection matrix together with a reference matrix. In this way, the subspace selection matrix can be accurately indicated from the data compression device, which is conducive to accurate decompression.

[0019] In some implementations, the method performed by the data compression device further includes: receiving feedback information for the first data matrix; and determining the second data matrix based on the feedback information and the first data. In this way, at least a portion of the first data matrix is ​​accurately selected to compress the second data matrix, thereby improving compression accuracy.

[0020] In some implementations, the method performed by the data compression device further includes: sending fifth indication information for the first data matrix; and determining the second data matrix based on the fifth indication information and the first data matrix.

[0021] In some implementations, the method performed by the data decompression device further includes: sending feedback information for the first data matrix; and determining the second data matrix based on the feedback information and the first data. In this way, at least a portion of the first data matrix is ​​accurately selected to decompress the second data matrix, thereby improving the decompression accuracy.

[0022] In some implementations, the method performed by the data decompression device further includes: receiving fifth indication information for the first data matrix; and determining the second data matrix based on the fifth indication information and the first data matrix.

[0023] In some implementations, the feedback information or the fifth indication information includes one or more of the following: subset indication information for indicating the first data matrix or a submatrix of the first data matrix, or confidence information for indicating the confidence of the first data. In this way, a portion of the first data matrix can be accurately indicated, which is conducive to accurate compression of the second data matrix.

[0024] In some implementations, the first data matrix has a first granularity, the second data matrix has a second granularity, and the first granularity is greater than the second granularity. In this way, the data compression method can be applied to data of different granularities, expanding its scope of application.

[0025] In some implementations, the first data matrix is ​​sampled data of the geographic space using a first resolution, the second data is sampled data of the geographic space using a second resolution higher than the first resolution, and the subset indication information indicates the spatial position information of the subset of the first data. Additionally or alternatively, the method performed by the data compression device may also be: the first data matrix is ​​a plurality of cluster centers of a plurality of data classes determined by clustering the original data, the second data matrix is ​​data contained in one or more of the plurality of data classes, and the subset indication information indicates one or more of the plurality of cluster centers corresponding to the one or more data classes. In this way, the second data matrix may be collected and compressed in different resolutions or clustering ways, thereby improving compression accuracy and reducing errors.

[0026] In some implementations, the reference matrix is ​​determined based on one or more of the following: feedback information for the first data matrix, a submatrix of the first data matrix, or the first data matrix. In this way, a reference matrix can be generated in a variety of flexible ways to facilitate data compression.

[0027] In some implementations, the second data matrix is ​​not grouped but corresponds to the reference matrix as a whole, or the second data matrix is ​​divided into multiple data groups, and the multiple data groups correspond to multiple sub-matrices of the reference matrix respectively. In this way, the second data matrix can be processed flexibly in a grouped or non-grouped manner, which is conducive to appropriate compression processing.

[0028] In some implementations, the method performed by the data compression device further includes sending one or more of the following: first configuration information for indicating whether the feedback information for the first data matrix is ​​based on an index or a bitmap, second configuration information for indicating whether the confidence feedback for the first data matrix is ​​enabled, third configuration information for indicating whether the first data matrix is ​​based on a geographic location or a cluster, fourth configuration information for indicating whether the reference matrix is ​​determined based on the feedback information for the first data matrix or based on the first data matrix, or fifth configuration information for indicating whether the second data matrix as a whole corresponds to the reference matrix or whether the multiple data groups divided from the second data matrix correspond to multiple sub-matrices of the reference matrix. In this way, the data compression device can accurately send configuration information to the data decompression device, which is conducive to synchronization between the two parties and accurate decompression by the data decompression device.

[0029] In some implementations, the method performed by the data compression device further includes: before sending the first data matrix, sending at least one of the first configuration information, the second configuration information, the third configuration information, the fourth configuration information, or the fifth configuration information. This facilitates synchronization between the data compression device and the data decompression device, and facilitates accurate decompression by the data decompression device.

[0030] In some implementations, the method performed by the data decompression device further includes receiving one or more of the following: first configuration information for indicating whether the feedback information for the first data matrix is ​​based on an index or a bitmap, second configuration information for indicating whether the confidence feedback for the first data matrix is ​​enabled, third configuration information for indicating whether the first data matrix is ​​based on a geographic location or a cluster, fourth configuration information for indicating whether the reference matrix is ​​determined based on the feedback information for the first data matrix or based on the first data matrix, or fifth configuration information for indicating whether the second data matrix as a whole corresponds to the reference matrix or whether the multiple data groups divided from the second data matrix correspond to multiple sub-matrices of the reference matrix. In this way, the data decompression device can accurately receive the configuration information from the data compression device, which is beneficial to the synchronization of both parties and the accurate decompression of the data decompression device.

[0031] In some implementations, the method performed by the data decompression device further includes: before receiving the first data matrix, receiving at least one of the first configuration information, the second configuration information, the third configuration information, the fourth configuration information, or the fifth configuration information. This facilitates synchronization between the data compression device and the data decompression device, and facilitates accurate decompression by the data decompression device.

[0032] In some implementations, the method performed by the data compression device further includes sending one or more of the following: sixth configuration information for indicating whether the projection matrix is ​​based on the column space of the reference matrix or on the subspace of the column space, or seventh configuration information for identifying the reference matrix from the first data matrix. In this way, the generation method of the projection matrix can be accurately identified, which is beneficial to the synchronization of the data compression device and the data decompression device, and is beneficial to the data decompression device for accurate decompression.

[0033] In some implementations, the data compression device sends at least one of the sixth configuration information or the seventh configuration information before sending the projection result and the projection residual. In this way, the method for generating the projection matrix can be accurately identified, which is beneficial to the synchronization of the data compression device and the data decompression device, and is beneficial to the data decompression device for accurate decompression.

[0034] In some implementations, the method performed by the data decompression device further includes receiving one or more of the following: sixth configuration information for indicating whether the projection matrix is ​​based on the column space of the reference matrix or on the subspace of the column space, or seventh configuration information for identifying the reference matrix from the first data matrix. In this way, the generation method of the projection matrix can be accurately identified, which is beneficial to the synchronization of the data compression device and the data decompression device, and is beneficial to the data decompression device for accurate decompression.

[0035] In some implementations, the data decompression device receives at least one of the sixth configuration information or the seventh configuration information before receiving the projection result and the projection residual. In this way, the method for generating the projection matrix can be accurately identified, which is beneficial to the synchronization of the data compression device and the data decompression device, and is beneficial to the data decompression device for accurate decompression.

[0036] In a third aspect, a device is provided. The device may refer to the data compression device itself (for example, implemented as a terminal device, a network device), or a component in the data compression device (for example, a processor, a chip, or a chip system, etc.), or may be a logic module or software that can realize all or part of the functions of the data compression device. The following description is given by taking the device as a data compression device as an example. The device includes: a first data matrix sending module for sending a first data matrix, and a result sending module for sending a projection result and a projection residual of a second data matrix. The projection result is determined according to the projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result. In this way, the correlation between the first data matrix and the second data matrix can be fully utilized to improve the compression efficiency and save transmission bandwidth or storage resources.

[0037] In a fourth aspect, a device is provided. The device may refer to the data decompression device itself (for example, implemented as a network device, a terminal device), or a component in the data decompression device (for example, a processor, a chip, or a chip system, etc.), or may be a logic module or software that can realize all or part of the functions of the data decompression device. The following description is taken as an example that the device is a data decompression device. The device includes: a first data matrix receiving module for receiving a first data matrix, a result receiving module for receiving a projection result and a projection residual of a second data matrix, and a second data matrix acquisition module for acquiring a second data matrix or an approximate matrix of the second data matrix based on the first data matrix, the projection result and the projection residual. Among them, the projection result is determined according to the projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result. In this way, the data compressed by the data compression device can be decompressed and restored, and the correlation between the first data matrix and the second data matrix can be fully utilized to improve the compression efficiency and save transmission bandwidth or storage resources.

[0038] In a fifth aspect, a system is provided, comprising the apparatus in the third aspect and the fourth aspect. In this way, the correlation between the first data matrix and the second data matrix can be fully utilized to improve compression efficiency and save transmission bandwidth or storage resources.

[0039] In a sixth aspect, a device is provided, which may be a data compression device or a data decompression device in the above method embodiment, or a chip set in the data compression device or the data decompression device. The device includes a processor and a memory. The memory is used to store a computer program or instruction, and when the processor runs the computer program or instruction, the data compression device or the data decompression device executes the method executed by the data compression device or the data decompression device in the above method embodiment.

[0040] In a seventh aspect, an embodiment of the present application provides a computer-readable storage medium storing machine-executable instructions. When the machine-executable instructions are executed, the method performed by the data compression device or the data decompression device in the above aspects is implemented.

[0041] In a ninth aspect, a computer program product is provided, the computer program product comprising: a computer program code, when the computer program code is run, the method performed by the data compression device or the data decompression device in the above aspects is executed. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1A A communication system in which the embodiments of the present application can be implemented.

[0043] Figure 1BThe block diagram is a block diagram for compressing and transmitting raw data over an air interface.

[0044] Figure 1C A schematic diagram of a low-rank matrix approximation.

[0045] Figure 1D A schematic diagram of a first scenario of data transmission.

[0046] Figure 1E A schematic diagram of a second scenario of data transmission.

[0047] Figure 2 Flow chart of data compression and decompression in an embodiment of the present application.

[0048] Figure 3A This is a flowchart of transmitting the first data matrix and the second data matrix in an embodiment of the present application.

[0049] Figure 3B It is a schematic diagram of transmitting the first data matrix and the second data matrix in an embodiment of the present application.

[0050] Figure 4A A schematic diagram of performing projection compression on a given projection matrix in an embodiment of the present application.

[0051] Figure 4B A flowchart for performing projection compression on a given projection matrix in an embodiment of the present application.

[0052] Figure 5A A schematic diagram of generating a projection matrix by column selection or LRMA decomposition in an embodiment of the present application.

[0053] Figure 5B This is a flowchart of generating a projection matrix for data compression by column selection or LRMA decomposition in an embodiment of the present application.

[0054] Fig. 6A It is a schematic diagram of subspace projection in an embodiment of the present application.

[0055] Figure 6B A flowchart of generating a projection matrix for data compression for subspace projection in an embodiment of the present application.

[0056] Fig. 7A is a schematic diagram of a radio frequency map (RF map) in an embodiment of the present application.

[0057] Figure 7B It is a schematic diagram of the location-based coarse / fine-grained data division and retrieval method in an embodiment of the present application.

[0058] Figure 7C It is a schematic diagram of the clustering-based retrieval method for coarse / fine-grained data division in an embodiment of the present application.

[0059] Fig.7D This is a flowchart of data compression based on retrieval in an embodiment of the present application.

[0060] Fig. 8A It is a schematic diagram of unpacked data compression in an embodiment of the present application.

[0061] Figure 8B It is a schematic diagram of packet data compression in an embodiment of the present application.

[0062] Fig. 9 This is a flowchart of the comprehensive signaling interaction in an embodiment of the present application.

[0063] Fig. 10A It is a schematic diagram of the effect of compressing radio frequency map data in an embodiment of the present application.

[0064] Fig. 10B It is a schematic diagram of the effect of compressing the channel matrix in an embodiment of the present application.

[0065] Fig.11 Flow chart of the data compression method in an embodiment of the present application.

[0066] Fig.12 This is a flow chart of the data decompression method in an embodiment of the present application.

[0067] Fig.13 A simplified block diagram of an example device showing a possible implementation of an embodiment of the present application.

[0068] Fig.14 A simplified block diagram of a communication device of a possible implementation method in an embodiment of the present application is shown.

[0069] Fig.15 A simplified block diagram of a network device in a possible implementation manner of an embodiment of the present application is shown. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the embodiments of the present application will be further described in detail with reference to the accompanying drawings. The specific operation methods and functional descriptions in the method embodiments can also be applied to the device embodiments or system embodiments.

[0071] Air interface native data mainly includes synaesthesia data, AI model data, and CSI data of multi-antenna systems. Native data has high dimensions and large data volume, and its interaction and transmission will occupy a large amount of air interface resources. Data compression technology can greatly reduce the occupation of air interface resources while meeting certain distortion requirements or task accuracy requirements. Data compression is required to reduce the occupation of air interface resources. Other data, such as application interaction data in the Internet, also need to be compressed to save network bandwidth or storage resources.

[0072] Figure 1A A communication system in which data compression and decompression can be implemented in the embodiments of the present application. Figure 1A As shown, the communication method provided in the embodiment of the present application can be applied to a wireless communication system 100. In the wireless communication system 100, a terminal device 101 and a network device 103 are shown. Data compression can be implemented in the terminal device 101, and data decompression can be implemented in the network device 103. Or conversely, data compression can be implemented in the network device 103, and data decompression can be implemented in the terminal device 101. In the wireless communication system 100, a network device 103 such as a base station (BS) provides communication services to a terminal device 101 such as a mobile station (MS). The base station includes a baseband unit (BBU) and a remote radio unit (RRU). The BBU and the RRU can be placed in different places, for example: the RRU is remote and placed in an area with high traffic volume, and the BBU is placed in a central computer room. The BBU and the RRU can also be placed in the same computer room. The BBU and the RRU can also be different components under one rack. It is understood by those skilled in the art that data compression and decompression can also be implemented between two terminal devices using side link communication, or between two network devices, or between two devices using wired links. Data compression and decompression can also be used for data storage in media such as hard disks, Flash, read-only memory (ROM), random access memory (RAM), etc., and the embodiments of the present disclosure are not limited to this.

[0073] The wireless communication systems in the embodiments of the present application include but are not limited to: Narrow Band-Internet of Things (NB-IoT), Global System for Mobile Communications (GSM), Enhanced Data rate for GSM Evolution (EDGE), Wideband Code Division Multiple Access (WCDMA), Code Division Multiple Access 2000 (CDMA2000), Time Division-Synchronization Code Division Multiple Access (TD-SCDMA), Long Term Evolution (LTE), three major application scenarios of 5G mobile communication systems, eMBB, URLLC and eMTC, and 6G, etc.

[0074] It should be understood that the above wireless communication system can be applied to high-frequency scenarios such as millimeter waves (above 6G) as well as low-frequency scenarios (sub6G). The application scenarios of the wireless communication system include but are not limited to the fifth generation system (5G), new radio (NR) communication system and other communication systems or future evolved public land mobile network (PLMN) systems.

[0075] The terminal device 101 shown above can be a user equipment (UE), a terminal, an access terminal, a terminal unit, a terminal station, a mobile station (MS), a remote station, a remote terminal, a mobile terminal, a wireless communication device, a terminal agent or a terminal device, etc. The terminal device 110 can also be a communication chip with a communication module, or a vehicle with a communication function, or a vehicle-mounted device (such as a vehicle-mounted communication device, a vehicle-mounted communication chip), etc. The terminal device 101 can have a wireless transceiver function, which can communicate with one or more network devices of one or more communication systems (such as wireless communication) and receive network services provided by the network devices. The network devices here include but are not limited to the network device (103) shown in the figure. A person of ordinary skill in the art can understand that Figure 1AThe data compression and decompression scenarios shown may also be applicable between network devices, or between terminal devices, etc., and the present disclosure does not limit this.

[0076] Among them, the terminal device 101 can be a cellular phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA) device, a handheld device with wireless communication function, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a wearable device, a terminal device in a future 5G network, or a terminal device in a future evolved PLMN network, etc.

[0077] The terminal device 101 may specifically be a mobile phone, a tablet computer, a computer with wireless transceiver function, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a wireless terminal in industrial control, a wireless terminal in self driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in a smart city, a wireless terminal in a smart home, etc.

[0078] In addition, the terminal device 101 can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted. The terminal device 101 can also be deployed on the water surface (such as a ship, etc.). The terminal device 101 can also be deployed in the air (for example, on an airplane, a balloon, and a satellite, etc.). The network device (103) can be an access network device (or an access network point). Among them, the access network device refers to a device that provides a network access function, such as a radio access network (RAN) base station, etc. The network device (103) can specifically include a base station (BS), or a base station and a wireless resource management device for controlling the base station, etc. The network device (103) can also include a relay station (relay device), an access point, a base station in a 5G network or an NR base station, a base station in a future evolved PLMN network, etc. The network device (103) can be a wearable device or a vehicle-mounted device. The network device (103) can also be a communication chip with a communication module.

[0079] For example, the network device (103) includes, but is not limited to: a base station (g nodeB, gNB) in 5G, an evolved node B (evolved node B, eNB) in a long term evolution (LTE) system, a radio network controller (RNC), a wireless controller under a cloud radio access network (CRAN) system, a base station controller (BSC), a home base station (for example, home evolved nodeB, or home node B, HNB), a baseband unit (BBU), a transmission point (transmitting and receiving point, TRP), a transmitting point (transmitting point, TP), a mobile switching center, and may also be an evolved NB (eNB or eNodeB) in LTE, a base station device in a future 5G network, or an access network device in a future evolved PLMN network, or a wearable device or a vehicle-mounted device.

[0080] In some deployments, the network device may include a centralized unit (CU) and a distributed unit (DU). The network device may also include an active antenna unit (AAU). The CU implements some functions of the network device, and the DU implements some functions of the network device. For example, the CU is responsible for processing non-real-time protocols and services, and implementing the functions of the radio resource control (RRC) and packet data convergence protocol (PDCP) layers. The DU is responsible for processing physical layer protocols and real-time services, and implementing the functions of the radio link control (RLC) layer, the media access control (MAC) layer, and the physical (PHY) layer. The AAU implements some physical layer processing functions, radio frequency processing, and related functions of active antennas. Since the information of the RRC layer will eventually become the information of the PHY layer, or be converted from the information of the PHY layer, under this architecture, high-level signaling, such as RRC layer signaling, can also be considered to be sent by the DU, or by the DU+AAU. It is understandable that the network device may be a device including one or more of a CU node, a DU node, and an AAU node. In addition, the CU may be divided into a network device in an access network (radio access network, RAN), or the CU may be divided into a network device in a core network (core network, CN), and the embodiments of the present application are not limited to this. Examples of network devices include, but are not limited to, Node B (NodeB or NB), evolved NodeB (eNodeB or eNB), next generation NodeB (gNB), transmit receive point (TRP), remote radio unit (RRU), radio head (RH), remote radio head (RRH), IAB node, low power node, such as a micro-micro node, a micro-micro node, a reconfigurable intelligent surface (RIS), a network controlled repeater, and the like.

[0081] In addition, the network device (103) can be connected to a core network (CN) device, and the core network device can be used to provide core network services for the access network device (103) and the terminal device (101). The core network device can correspond to different devices in different systems. For example, in 3G, the core network device can correspond to the serving GPRS support node (SGSN) of the general packet radio service (GPRS) and / or the gateway GPRS support node (GGSN) of GPRS. In 4G, the core network device can correspond to the mobility management entity (MME) and / or the serving gateway (S-GW). In 5G, the core network device can correspond to the access and mobility management function (AMF), the session management function (SMF) or the user plane function (UPF).

[0082] Figure 1B It is a block diagram for compressing and transmitting air interface native data. In block diagram 110, at the transmitting end, the perception data, AI data and CSI data are subjected to data identification and filtering 113, data transformation and quantization 115, data selection and channel mapping 117 in physical source coding 111, and then channel coding 119, and enter the receiving end through the channel. At the receiving end, the data received from the channel is subjected to channel decoding 121 and source decoding 123 to obtain data for the task. The channel can be, for example, a wireless channel 121. It can be understood by those of ordinary skill in the art that the channel can also be a wired channel, and the embodiments of the present disclosure are not limited to this. It can be understood by those of ordinary skill in the art that this scenario of compressed data transmission through the channel can also be replaced by a compressed data storage scenario in the medium, and the embodiments of the present disclosure are not limited to this.

[0083] Based on specific scenarios, there are various forms of redundancy in the air interface native data. Mining data redundancy can be used for data compression. For example, a large amount of content in the perceived original signal has little impact on subsequent tasks, and discarding it can greatly reduce the amount of data. The perceived point cloud data is correlated in time and space, and the channel matrix has a strong correlation in the frequency domain and the spatial angle domain.

[0084] Low Rank Matrix Approximation (LRMA) is a method of mining data correlation for data compression. Figure 1C As shown. Based on the Eckart-Young-Mirsky theory or truncated singular value decomposition (SVD), the matrix A with m rows and n columns is approximately equal to the matrix B with m rows and k columns multiplied by the matrix E with k rows and n columns. The columns of the matrix B are orthogonal and can be used as a projection subspace. The projection error of each column of A on this projection subspace approaches the minimum value. The compression method based on LRMA can mine the correlation redundancy between columns in the data matrix A. In many scenarios of native data compression, the system can transmit data multiple times instead of just once. For example, the CSI scenario may use periodic interval feedback. As in Figure 1D In the first scenario (CSI scenario), the terminal device 141 performs two channel measurements 146 and 156 based on the reference signals sent by the network device 143, such as the first reference signal 145 and the second reference signal 155, respectively, to obtain the first channel data 150 and the second channel data 160. For example, the first channel data 150 and the second channel data 160 may be the channel frequency domain transfer function between the terminal device 141 and the network device 143. In the scenario where the terminal device 141 is not moving very fast, the first channel data 150 and the second channel data 160 may have time correlation. Figure 1E In the point cloud or RF map scene (second scene) shown, the effect of hierarchical transmission can be achieved through interaction. The network device 173 first transmits a large range of coarse-grained data 175 to the terminal device 171, and then transmits a small range of fine-grained data 185 based on the perception measurement 176 and feedback 180 of the terminal device, which can greatly reduce the amount of transmitted data. The coarse-grained data 175 and the fine-grained data 185 in this scene have spatial correlation. Figure 1D , Figure 1E In the example, the terminal devices 161 and 171 may be Figure 1A In the implementation of the terminal device 101, the network devices 163 and 173 may be Figure 1A In addition to the implementation of the network device 103 Figure 1E The spatial granularity of point cloud data and radio frequency map data in Figure 1D The channel data may also have frequency domain granularity. For example, the first channel data 150 is frequency domain coarse-grained data, and the second channel data 160 is frequency domain fine-grained data. The terminal device 141 interacts with the feedback ( Figure 1D(not shown) generates the second channel data 160 of fine granularity in the frequency domain. A person skilled in the art can understand that there may be other granularity resolution methods, such as time domain granularity, etc., which is not limited in the embodiments of the present disclosure.

[0085] In the embodiment of the present disclosure, since there is a correlation between the first channel data 150 and the second channel data 160, and there is a correlation between the coarse-grained data 175 and the fine-grained data 185, at least a portion of the first channel data 150 or the coarse-grained data 175 can be selected as a reference for compressing the second channel data 160 or the fine-grained data 185. The first granularity of the coarse-grained data 175 is greater than the second granularity of the fine-grained data 185. In the embodiment of the present disclosure, the above data can be in the form of a matrix, and the first channel data 150 and the coarse-grained data 175 can be collectively referred to as a first data matrix, and the second channel data 160 and the fine-grained data 185 can be collectively referred to as a second data matrix. It can be understood by those skilled in the art that the above data can also be in the form of a vector, and a vector is a specific form of a matrix.

[0086] Figure 2 This is a flowchart of data compression and decompression in an embodiment of the present application, which specifically describes how to perform efficient data compression and decompression in multiple transmission scenarios of data such as air interface native data.

[0087] In process 200, the data compression device 201 sends (204) a first data matrix 205. The data compression device 201 sends (208) a projection result and a projection residual 210 of a second data matrix. The projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result. The data decompression device 203 receives the first data matrix 205, and the projection result and the projection residual 210 of the second data matrix. At 215, the data decompression device 203 obtains a second data matrix or an approximate matrix of the second data matrix based on the first data matrix, the projection result and the projection residual. In this way, the correlation between the first data matrix and the second data matrix can be fully utilized to improve compression efficiency and save transmission bandwidth or storage resources. The data compression device 201 can correspond to Figure 1A In the terminal device 101, the data compression device 203 may correspond to Figure 1A Alternatively, the data compression device 201 may correspond to the network device 103 in Figure 1A The network device 103 in the data compression device 203 may correspond to Figure 1A The terminal device 101 in.

[0088] Figure 3A This is a flowchart of transmitting the first data matrix and the second data matrix in an embodiment of the present application. Figure 3AThe data compression device 301 and the data decompression device 303 in the embodiment can be respectively Figure 2 Implementation of the data compression device 201 and the data decompression device 203.

[0089] In process 300, the data compression device 301 sends (304) the first data matrix 305 to the data decompression device 303. At 310, the data compression device 301 selects a d-dimensional subspace from the k-dimensional space spanned by the selected reference data or reference matrix, and projects the second data matrix onto the d-dimensional subspace, and subsequently compresses the projection residual. The reference matrix is ​​a submatrix of the first data matrix 305. The submatrix can be a part or all of the first data matrix 305. The data compression device 301 sends (313) the second data matrix to the data decompression device 303, specifically, the projection part or the projection result, and the compression result of the projection residual. The data compression device 301 can also send the compression result of the projection part or the projection result. The compression result of the projection part and the compression result of the projection residual can be further quantized. In this way, the correlation between the second data and the first data can be used to efficiently compress the second data, reduce the transmission bandwidth or save storage space.

[0090] Figure 3BSchematic diagram of transmitting the first data matrix and the second data matrix in an embodiment of the present application. In embodiment 320, the data compression device performs LRMA decomposition, such as matrix decomposition, on the reference matrix C 325 (m rows and k columns) in the first data matrix to obtain the product 330 of the projection matrix (s columns) and the projection coefficients (s rows). The projection matrix can represent a basis (Base) or a subspace, and the projection matrix can be a coefficient matrix. At 335, the data compression device selects a d-dimensional subspace from the k-dimensional space spanned by the reference data as the projection matrix (d columns) for the second data matrix G. This d-dimensional subspace can select d columns from the k columns of the reference matrix C, or it can select d columns from the projection matrix of s columns. The data compression device projects the second data matrix G of m rows and n columns to the selected d-dimensional subspace, or the projection matrix of d columns, and obtains a projection part or projection result 345 of d rows and n columns, as well as a projection residual or orthogonal part 350. The data compression device performs subsequent compression on the projection residual or orthogonal part 350, for example, by using the LRMA method, or dictionary compression, or transform domain compression, difference, quantization, etc., to obtain a compression result. The data compression device can also compress the projection part or projection result 345 by, for example, quantization, entropy coding, etc. to obtain a compression result. The d-dimensional subspace is selected by using the correlation between the first data matrix and the second data matrix, and the second data matrix is ​​projected. The projection result and the projection result can be used to characterize the second data matrix G 340 with a smaller error. It can be understood by those skilled in the art that in addition to matrix decomposition, other methods such as dictionary learning can also be used to obtain the projection matrix, and the present disclosure is not limited to this.

[0091] In an embodiment of the present disclosure, a second data matrix such as fine-grained data is projected on a space spanned by a first data matrix such as coarse-grained data to assist compression, and a projection operation is performed on a space spanned by a reference matrix. The second data matrix can be projected and decomposed given a projection matrix, and the projection matrix can be selected in an appropriate manner, or the projection matrix can be constructed on an optimal subspace of the space spanned by the reference matrix for projection. In this way, the second data matrix can be compressed using the correlation between the first data matrix and the second data matrix to reduce the compression error. In an embodiment of the present disclosure, coarse-grained data (corresponding to the first data matrix) can be retrieved based on spatial position or clustering, and fine-grained data can be compressed in a feedback interaction manner. The feedback can be the retrieval result or confidence of the coarse-grained data. In this way, the fine-grained data is accurately compressed using the correlation between the coarse-grained data and the fine-grained data (corresponding to the second data matrix) to reduce the compression error. The data compression device and the data decompression device can also exchange the correspondence between the coarse-grained data and the fine-grained data, as well as the indication of the reference coarse-grained data, so as to achieve synchronization between the data compression device and the data decompression device, which is conducive to accurate decompression.

[0092] Figure 4A Schematic diagram of projection compression of a given projection matrix in an embodiment of the present application. Specifically, embodiment 400 implements projection of the second data matrix G under a given projection matrix P. At least a portion of the reference matrix or a basis of the reference matrix constitutes a projection matrix P405 (dimension m*d), and its column space constitutes a projection space. The projection matrix P is used to perform projection decomposition on the second data matrix G 410 to be compressed to obtain a projection part or result part 415 (P T P) -1 P T G. According to the second data matrix G 410 and the projection part or result part: 415, the orthogonal part 420G-(P T P) -1 P T G. The orthogonal part is subsequently compressed to obtain a compression result. The subsequent compression may be any one of the following: LRMA decomposition, dictionary compression, transform domain compression, differential calculation, quantization, or other compression methods, which are not limited in the embodiments of the present disclosure. The projection part or the result part may also be compressed to obtain a compression result. In this embodiment, the first data matrix or the reference matrix may be coarse-grained, and the second data matrix G 410 may be fine-grained.

[0093] Figure 4B A flowchart for performing projection compression on a given projection matrix in an embodiment of the present application, and Figure 4A The data compression device 431 and the data decompression device 433 can be respectively Figure 2 Specific implementation of the data compression device 201 and the data decompression device 203 in the embodiment 430. In the embodiment 430, the data compression device 431 sends (434) the first data matrix 435 to the data decompression device 433. At 440, the data compression device 431 projects the second data matrix using the projection matrix P, and subsequently compresses the projection residual (orthogonal part). The data compression device 431 sends (443) the second data matrix 445 to the data decompression device 433. Specifically, the projection result matrix (P T P) -1 P T G, and the projection residual G-(P T P) -1 P T G performs subsequent compression on the compression result. The data compression device 431 may also send the compression result of the projection result matrix to the data decompression device 433.

[0094] Figure 5A Schematic diagram of generating a projection matrix by column selection or LRMA decomposition in an embodiment of the present application. Specifically, embodiment 500 may be Figure 4A The specific generation method of the projection matrix P in .

[0095] In the disclosed embodiment, the projection matrix P 520 can be selected from the d columns in the reference matrix C 505 of m rows and k columns. The reference matrix C is the first data matrix or a submatrix of the first data matrix. The selection method can be to solve a combinatorial optimization problem. When k is not large, the k columns can be traversed and searched, or the d columns can be selected from the k columns by heuristic search. The projection matrix P520 can also be selected from the d columns of the projection basis 510 (m*s matrix) when the reference data matrix C is decomposed by LRMA. For example, after performing SVD decomposition on the reference matrix, the d columns corresponding to the largest d singular values ​​of the SVD can be directly selected. In this way, two different methods can be flexibly used to obtain the projection matrix P. The method of directly selecting d columns in the reference matrix has a small amount of calculation, while the method of performing SVD decomposition on the reference matrix can reduce the projection error.

[0096] Figure 5B A flowchart of generating a projection matrix for data compression by column selection or LRMA decomposition in an embodiment of the present application, and Figure 5A The projection matrix generation method shown in corresponds to Figure 4B compared to, Figure 5BIn the embodiment 530 of the embodiment 530, the data compression device 431 also sends the first indication information, i.e., the projection matrix dimension d indication 535, the second indication information, i.e., the projection matrix source indication 540 (1 bit), and the third indication information, i.e., the projection matrix column indication 545 (k bits) to the data decompression device 433. The number of columns or dimension d of the projection matrix is ​​a parameter related to the compression rate (data volume) and can be dynamically indicated. In actual use, the data compression device 431 and the data decompression device 433 can configure some possible values ​​in advance, for example, all possible values ​​of d are 1, 2, 3, 4, 5, 6, 7, 8, and then dynamically select or indicate one of the values ​​with several bits. For example, the aforementioned 8 possible values ​​of d require 3 bits for indication. The data compression device 431 can also send a projection matrix source indication 540 (1 bit) to the data decompression device 433 to indicate whether the projection matrix P comes from the reference matrix or the basis after the reference matrix is ​​decomposed by LRMA. The projection matrix source indication 540 can be dynamically indicated each time it is transmitted, or it can be semi-statically indicated and selected once at intervals. When the projection matrix P is directly derived from the reference matrix, the projection matrix column indication 545 is used to indicate which d columns are selected from the k columns of the reference matrix as the projection matrix P, and a k-bit bitmap can be used to identify the selected d columns, at which time 1≤d≤k. When the projection matrix P is derived from the basis of the reference matrix for LRMA decomposition, the projection matrix column indication 545 is used to indicate the number of columns of the projection matrix P, at which time 1≤d≤s. In the embodiment of the present disclosure, the data compression device 431 and the data decompression device 433 can also be fixedly configured, and one of the d columns of the reference matrix and the d columns of the basis after the reference matrix LRMA decomposition is selected to compress and decompress the second data matrix. In this way, the correlation between the first data matrix and the second data matrix can be used, and the compression method can be flexibly selected to compress and decompress the second data matrix. Each indication information realizes the synchronization of the data compression device and the data decompression device, which is conducive to accurate data compression and decompression. In the embodiment of the present disclosure, the first data matrix or reference matrix C 505 may be coarse-grained, and the second data matrix may be fine-grained.

[0097] Fig. 6A Schematic diagram of subspace projection in the embodiment of the present application. In embodiment 600, the data compression device can select the best d-dimensional subspace in the column space 605 spanned by the first data matrix to project the second data matrix 610, and the selection of the d-dimensional subspace is related to the second data matrix 610, thereby reducing the projection error and further reducing the compression error.

[0098] In the disclosed embodiment, the reference matrix C (m*k dimension) referred to by the second data matrix G (m*n dimension) to be compressed is the first data matrix or a submatrix of the first data matrix, a d-dimensional subspace of the reference matrix C is selected, and a projection matrix is ​​obtained. The selection criterion is to minimize the norm (Frobenius) norm of the orthogonal part after projection, that is, the error of the projection of the second data matrix G in the subspace approaches the minimum value. The solution can be obtained by solving the optimization problem, and the specific steps are as follows:

[0099] The data compression device first selects the optimal subspace

[0100]

[0101] Where R k×d is a real space of k*d dimensional matrix, which is a floating point number. T is the transpose of the P matrix, is the Frobenius norm operation, and argmin is the minimum value optimization.

[0102] The subspace selection matrix S is obtained from the previous step * Get the optimal projection subspace, that is, get the projection matrix

[0103] P * =C×S *

[0104] The second data matrix G is projected and decomposed on the optimal projection subspace into the following two parts:

[0105] Projection part: (P *T P * ) -1 P *T G

[0106] Orthogonal part: GP(P T P) -1 P T G

[0107] Finally, the orthogonal part of the projection is compressed using methods such as LRMA compression, dictionary compression, transform domain compression, etc. The projection part can also be compressed using methods such as quantization and entropy coding.

[0108] In the embodiment of the present disclosure, the optimal subspace S * It is optimized for the second data matrix G, so that the projection error of G in this subspace is minimized. When a new second data matrix G' appears, S can be recalculated. * , thus obtaining the optimal substrate every time.

[0109] In the embodiment of the present disclosure, for the second data matrix, the data to be transmitted includes three parts: the subspace selection matrix S * , the subspace projection part of the second data matrix, and the subsequent compression result of the projection residual of the second data matrix. Figure 6B A flowchart for generating a projection matrix for data compression for subspace projection in an embodiment of the present application, and Fig. 6A And the above obtains the subspace selection matrix S * , and obtain the projection matrix corresponding to the embodiment. and Figure 5B In contrast, in sending (623) the second data matrix 625, a subspace selection matrix S is added. * In the disclosed embodiment, the subspace matrix S is selected in the data compression device 431 * Under the condition of , the projection matrix source indication 540 (1 bit) and the projection matrix column indication 545 (k bits) may no longer be sent. * The basis characteristics of the projection subspace have been fully described, and the source of the projection matrix and the column position of the projection matrix no longer need to be indicated.

[0110] In the embodiment of the present disclosure, the second data matrix G may be a fine-grained matrix, and the first data matrix and the reference matrix may be coarse-grained matrices.

[0111] In the embodiments of the present disclosure, a specific solution method for the above optimization problem is described later. The optimal subspace selection problem is to select a d-dimensional subspace in the k-dimensional column space spanned by the reference data matrix C, so that the residual error of the second data matrix G after projection on the d-dimensional subspace is as small as possible, that is,

[0112]

[0113] This problem has a closed-form solution. The detailed solution steps are as follows:

[0114] First, perform QR decomposition on the reference matrix C:

[0115] C=Q×R, The columns are orthogonal, Q is a real matrix with m rows and k columns, and R is a real matrix with k rows and k columns.

[0116] Substituting C=QR into P=CS, we get P=QRS. RS can be written as The original problem is equivalent to the following optimization problem:

[0117]

[0118] The above formula is equivalent to Find the optimal d-dimensional projection subspace. The matrix Q TG is decomposed by SVD, and we get

[0119] Q T G=U∑V T

[0120] Optimal is equal to the first d columns of matrix U, so the optimal solution to the original problem is And further get the projection matrix

[0121] P * =C×S *

[0122] To summarize the above solution process, the overall process is to project first and then perform SVD decomposition to obtain the optimal subspace. In this way, the optimal projection subspace related to the second data matrix can be obtained by a closed-form solution, reducing the projection error of the second data matrix and thus reducing the compression error.

[0123] In the embodiment of the present disclosure, the first data matrix, the reference matrix, and the second data matrix may have different granularities, for example, different spatial resolutions in a radio frequency map scenario.

[0124] Fig. 7A It is a schematic diagram of the radio frequency map (RF map) in the embodiment of the present application. The radio frequency map is a kind of air interface native data related to geographic location information, which can describe the electromagnetic propagation characteristics in the environment and can be used to assist communication and positioning. As shown in Example 700, the radio frequency map data divides the space into small grids according to a certain resolution, and each grid is represented by a location point, usually the center point of the grid. The radio frequency map data can record the electromagnetic propagation environment information at the representative position, such as multipath information (such as angle, delay, power, etc.), Scaler information (such as capacity, channel quality information CQI), whether the electromagnetic propagation between certain base stations is a line of sight (Line of Sight, LOS) or a non-line of sight (None Line of Sight, NLOS), a deterministic H matrix, the geographic location coordinates represented by the grid, etc. The radio frequency map data can be a summary of the above information at the grid point position, or a part of the above information, describing the electromagnetic propagation characteristics of the entire area. Similar information can also be used in point cloud data scenarios.

[0125] In the embodiment of the present disclosure, for example, Figure 1E In the scenario shown, coarse-grained data and fine-grained data can be divided and retrieved in two ways: location-based and clustering-based. Figure 7B shows a location-based approach, while Figure 7C A clustering-based approach is shown. Figure 7BIn embodiment 710, the coarse-grained data 715 sent by the data compression device is data obtained by spatial low-resolution sampling of the original data, corresponding to the first data matrix. The fine-grained data 725 (corresponding to the second data matrix) sent by the data compression device is obtained according to the feedback result 720 of spatial position retrieval of the coarse-grained data 715 fed back from the data decompression device. The first granularity of the coarse-grained data 715 is greater than the second granularity of the fine-grained data 725. The data decompression device can feedback the index of a subset 720 of the received coarse-grained data 715, or can use a bitmap to indicate the subset 720. The indexing method and the bitmap method can be selected from the two, and a preconfigured method can be adopted. The data compression device is based on the index of the subset 720 of the fed-back coarse-grained data, for example Figure 7B The position index in can be used to retrieve the fine-grained data 725 to be sent next. In this way, the association between the fine-grained data 725 and the coarse-grained data 715 can be established by means of a position index or a bitmap, making full use of the spatial correlation between the two types of data, performing data compression on the area of ​​interest, improving compression efficiency, and reducing data transmission resources or data storage space. In the disclosed embodiment, the coarse-grained data 715 and the fine-grained data 725 can be in vector form or in matrix form, and a vector is a special form of a matrix.

[0126] In the disclosed embodiment, the data decompression device may also use the confidence of the coarse-grained data 715 for feedback. For example, the data decompression device may divide the confidence into several levels, and then feedback the confidence level of each coarse-grained data 715. At this time, the data compression device determines which coarse-grained data 715 to retrieve based on the feedback confidence, thereby obtaining the fine-grained data 725 to be sent subsequently. The location-based method or the confidence method can be used in the radio frequency map data scenario or the point cloud data scenario. Compared with retrieving a subset, more accurate feedback can be provided through confidence.

[0127] exist Figure 7CIn embodiment 730, the data compression device clusters the original data 735, and the obtained coarse-grained data 740 (corresponding to the first data matrix) is used as the cluster center of the original data 735, and the coarse-grained data 740 is sent to the data decompression device. Points of different shapes in 735 represent original data in different classes. For example, points of shapes such as regular triangles, squares, and diamonds represent original data in different classes, respectively. In embodiment 730, the coarse-grained data 740 includes 6 cluster center points. The data decompression device retrieves 3 cluster center points 745 of interest from the coarse-grained data 740 and feeds them back to the data compression device. In the data compression device, the fine-grained data 750 (corresponding to the second data matrix) is obtained according to the retrieval result 745 of the coarse-grained data fed back by the data decompression device. The first granularity of the coarse-grained data 740 is greater than the second granularity of the fine-grained data 750. This clustering method can be used in radio frequency map data scenarios or point cloud scenarios. In the embodiments of the present disclosure, similar to Figure 7B In the scenario of the data decompression device, the confidence of the coarse-grained data 735 can also be fed back, which will not be elaborated in this disclosure. By clustering, the data transmission resources or data storage space of the coarse-grained data can be reduced. By generating fine-grained data through feedback from clustering, the spatial correlation between the coarse-grained data and the fine-grained data can be fully utilized, data compression can be performed on the area of ​​interest, the compression efficiency can be improved, and the data transmission resources or data storage space can be reduced. In the embodiment of the present disclosure, the coarse-grained data 740 and the fine-grained data 750 can be in vector form or in matrix form, and the vector is a special form of the matrix.

[0128] Fig.7D This is a flowchart of data compression based on retrieval in an embodiment of the present application. Fig.7D Process 760 in corresponds to Fig. 7A , 7B , 7C, specifically illustrate the signaling used in the above embodiments. The data compression device 761 and the data decompression device 763 may be specific implementations of the data compression device 201 and the data decompression device 203, respectively.

[0129] In process 760, the data compression device 761 sends (764) a coarse-grained data matrix 765 to the data decompression device 763. The data decompression device 763 analyzes and retrieves the coarse-grained data matrix 765, and feeds back (768) the retrieval result or confidence 770 of the coarse-grained data matrix. At 775, the data compression device 761 projects the fine-grained data matrix G and subsequently compresses the projection residual (orthogonal part). The data compression device 761 sends (778) a projection matrix dimension d indication 780, a projection matrix source indication 782 (1 bit), and a projection matrix column indication 784 (k bits) to the data decompression device 763 to configure the data decompression device 763 to achieve synchronization of data compression and decompression. The data compression device 761 sends (788) a fine-grained data matrix 790 to the data decompression device 763. The fine-grained data matrix 790 may include: a subspace selection matrix S * , the projection result matrix (P *T P * ) - 1 P * TG, and the projection residual GP(P T P) -1 P T The fine-grained data matrix 790 may also include a projection result matrix (P *T P * ) -1 P *T In this way, a complete data transmission and signaling transmission mechanism can be established between the data compression device 761 and the data decompression device 763, and the correlation between the coarse-grained data and the fine-grained data can be used to compress and decompress the data, thereby improving the compression efficiency and reducing the data transmission resources or data storage space.

[0130] It will be understood by those skilled in the art that, in addition to Figure 7A-7C As shown in the embodiment, coarse-grained data and fine-grained data of different spatial resolutions are used for radio frequency maps or point cloud scenes. The coarse-grained data and fine-grained data can also be used for example Figure 1D The frequency domain resolution of the channel data, or other time domain resolution and other scenarios are not limited in the embodiments of the present disclosure.

[0131] The disclosed embodiment also discloses compressing fine-grained data using the correspondence between fine-grained data and coarse-grained data, and referring to the indication of coarse-grained data, for example, Figure 1E , Fig. 7A , Figure 7B , Figure 7C The scene shown.

[0132] There are two possible correspondences between fine-grained data and coarse-grained data: no grouping and compression of fine-grained data, and grouping and compression of fine-grained data, respectively. Fig. 8A and Figure 8B Shown.

[0133] Fig. 8A It is a schematic diagram of unpacked data compression in an embodiment of the present application.

[0134] In embodiment 800, reference coarse-grained data C 805 and fine-grained data G 810 to be compressed are not grouped. During compression, all fine-grained data G 810 are compressed together, and the compression process corresponds to, or refers to, the same group of coarse-grained data, that is, the overall reference coarse-grained data 805. The reference coarse-grained data 805 can be in two different ways. The current reference coarse-grained data can be all the coarse-grained data fed back by the data decompression device, or it can be a subset of the coarse-grained data sent by the data compression device in the previous round. Under the condition of using a subset of the coarse-grained data, the subset can be indicated, for example, by means of a subscript index or a bit map. In this way, the fine-grained data can be compressed with reference to all the coarse-grained data or a subset of the coarse-grained data, making full use of the coarse-grained data and improving the compression effect.

[0135] Figure 8B It is a schematic diagram of packet data compression in an embodiment of the present application.

[0136] In the embodiment 820, the data compression device groups the fine-grained data 830, for example, into Figure 8B Each group of fine-grained data corresponds to a group in the coarse-grained data 825. Specifically, the fine-grained data 830 currently sent by the data compression device can be divided into multiple groups, and each group refers to a coarse-grained data fed back by the data decompression device. For example, corresponding to Figure 7C In the clustering embodiment, grouping can be performed based on clustering, and the reference coarse-grained data of each group can be the cluster center of the group. Each group of data in the fine-grained data 830 can refer to a subset of the coarse-grained data 825 sent in the previous round. In the embodiment of the present disclosure, the subset can be indicated, for example, by using a subscript index or a bitmap. In this way, the compression of the fine-grained data can refer to the part of the coarse-grained data that is most relevant to it, reducing the computational complexity and improving the compression efficiency.

[0137] Fig. 9 Flow chart of the integrated signaling interaction in the embodiment of the present application. Fig. 9 In the embodiment 900, the signaling used in the above embodiments is summarized. The data compression device 901 and the data decompression device 903 can be specific implementations of the data compression device 201 and the data decompression device 203.

[0138] Fig. 9 The first data matrix (coarse-grained data matrix) 910, the block 920 and the second data matrix (fine-grained data matrix) 960 in Figure 6B The first data matrix 405, the block 410, the second data matrix 625, and Fig.7D The coarse-grained data matrix 765, confidence 770, block 775, and fine-grained data matrix 790. Fig. 9 The retrieval results of the coarse-grained data in , the confidence level 910 corresponds to Fig.7D The coarse-grained data matrix retrieval results in will not be repeated in this disclosure.

[0139] In process 900, the sixth configuration information, or projection space mode indication 935, is used to select one of the following two ways to perform projection: using the column space of the reference matrix (reference coarse-grained data) for projection, or using the optimal subspace of the column space of the reference matrix (reference coarse-grained data) for projection. The seventh configuration information identifies the reference matrix from the first data matrix.

[0140] The mode and parameters 905 to be configured sent (904) by the data decompression device 903 to the data compression device 901 include: first configuration information, second configuration information, third configuration information, fourth configuration information and fifth configuration information. The mode and parameters 905 to be configured can also be sent (904) by the data compression device 901 to the data decompression device 903 to synchronize the data compression device 901 with the data decompression device 903.

[0141] Configuration signaling related to data partitioning and retrieval may include first configuration information, second configuration information, and third configuration information. The first configuration information or subset feedback mode configuration information is used to identify whether the data decompression device 903 uses a subscript index or a bit map when feeding back a coarse-grained data subset. The second configuration information or confidence feedback enable configuration information is used to identify that the data decompression device 903 adds confidence feedback when feeding back a coarse-grained data subset. The third configuration information or data partition mode configuration information is used to select between geographic location-based and cluster-based methods.

[0142] Configuration signaling related to data correspondence and reference coarse-grained data indication may include fourth configuration information and fifth configuration information. The fourth configuration information or data correspondence mode configuration information is used to select between group correspondence and non-group correspondence. The fifth configuration information or coarse-grained data reference mode configuration information is used to select between the following two reference modes: compressing the current fine-grained data (or a group of fine-grained data) with reference to the coarse-grained data fed back by the data decompression device 903; or compressing the current fine-grained data (or a group of fine-grained data) with reference not to the coarse-grained data fed back by the data decompression device 903, but all the coarse-grained data sent by the previous round of data compression device 901, or a subset of all the coarse-grained data.

[0143] In the embodiment of the present disclosure, the data compression device 901 and the data decompression device 903 can be well configured and synchronized through the above-mentioned signaling, so as to facilitate accurate compression and decompression.

[0144] Fig. 10A Schematic diagram of the effect of compressing radio frequency map data in the embodiment of the present application. In embodiment 1000, a comparison is made between the distortion based on LRMA in an implementation and the distortion using projection LRMA in the embodiment of the present disclosure.

[0145] The simulation corresponding to the embodiment 1000 is for radio frequency map data, using a coarse data and fine-grained data generation mode based on geographic location, corresponding to Figure 1E scene. For coarse-grained data, RF map data of 4 geographical locations at downsampled spatial resolution is used, with 10 diameters for each location. For fine-grained data, RF map data of 12 locations at original spatial resolution is used, with 10 diameters for each location. Benchmark 1005 is to directly perform LRMA compression on fine-grained data. In the disclosed embodiment, corresponding to 1010, the reference coarse-grained data uses the coarse-grained data of all 4 locations, and the projection matrix is ​​selected from the first d columns of the projection basis after LRMA decomposition of the reference coarse-grained data, and d is dynamically optimized. Curves 1005 and 1010 are rate distortion (RD) curves for compressing the pitch angle, the horizontal axis is the number of floating-point numbers after compression, and the vertical axis is the distortion. It can be seen from embodiment 1000 that, corresponding to the same vertical axis distortion, the number of floating-point numbers after compression of embodiment 1010 of the disclosed embodiment is smaller than that of benchmark 1005, achieving good data compression effect.

[0146] Fig. 10B Schematic diagram of the effect of compressing the channel matrix in the embodiment of the present application. In embodiment 1020, a comparison is made between the distortion based on LRMA in an implementation and the distortion using projection LRMA in the embodiment of the present disclosure.

[0147] The simulation corresponding to Example 1020 is mainly for the channel matrix data of a large-scale multiple input multiple output (MIMO) antenna system, corresponding to Figure 1D Scenario. The simulation environment is a single user, and data is fed back every 20 transmission time intervals (TTIs), and each TTI contains 14 orthogonal frequency division multiplexing (OFDM) symbols. 64 resource blocks (RBs)*12 subcarriers (SCs)=768 frequencies are used, and the subcarrier spacing is 3KHz. The number of antennas is 32 for reception and 1024 for transmission. Benchmark 1025 corresponds to reconstructing the data of the current TTI into a 32768*768 matrix, directly performing LRMA compression, and processing the real and imaginary parts separately. In 1030 of the embodiment of the present disclosure, the data fed back last time, for example, the data 20 TTIs ago, is used as a reference, and the projection matrix is ​​selected from the first d columns of the projection basis after the reference data is decomposed by LRMA, and d is fixed to 40. Curves 1025 and 1030 are rate distortion (RD) curves for compressing the channel transmission data H matrix, where the horizontal axis is the number of floating point numbers after compression and the vertical axis is distortion. As can be seen from embodiment 1020, corresponding to the same vertical axis distortion, the number of floating point numbers after compression of embodiment 1030 of the present disclosure is less than that of reference 1025, achieving a good data compression effect.

[0148] pass Fig. 10A and Fig. 10B It can be seen that the compression using the correlation between the coarse-grained data and the fine-grained data, or the first data matrix and the second data matrix, has achieved good results.

[0149] Fig.11Flow chart of the data compression method in the embodiment of the present application. The method can be executed by the data compression device 201. Unless otherwise specified, the data compression device 201 in the embodiment of the present application can refer to the data compression device itself (for example, implemented as a terminal device, a network device), or a component in the data compression device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can realize all or part of the functions of the data compression device. The following description is taken as an example that the execution subject is the data compression device 201. In the method 1100, at 1110, the data compression device 201 sends a first data matrix. At 1120, the data compression device 201 sends the projection result and projection residual of the second data matrix. The projection result is determined according to the projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result. In this way, the correlation between the first data matrix and the second data matrix can be fully utilized to improve the compression efficiency and save transmission bandwidth or storage resources.

[0150] In some implementations, the present disclosure also includes the following embodiments: Figure 2 Other operations performed at the data compression device 201 are described in detail with reference to FIG. 10 .

[0151] Fig.12 Flowchart of the data decompression method in the embodiment of the present application. The method can be executed by the data decompression device 203. Unless otherwise specified, the data decompression device 203 in the embodiment of the present application can refer to the data decompression device itself (for example, implemented as a terminal device, a network device), or a component in the data decompression device (for example, a processor, a chip, or a chip system, etc.), or a logic module or software that can implement all or part of the functions of the data decompression device. The following description is taken as an example that the execution subject is the data decompression device 203.

[0152] In the method 1200, at 1210, the data decompression device 203 receives a first data matrix. At 1220, the data decompression device 203 receives a projection result and a projection residual of a second data matrix. At 1230, the data decompression device 203 obtains a second data matrix or an approximate matrix of the second data matrix based on the first data matrix, the projection result and the projection residual. The projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result. In this way, the data compressed by the data compression device can be decompressed and restored, and the correlation between the first data matrix and the second data matrix can be fully utilized to improve compression efficiency and save transmission bandwidth or storage resources.

[0153] In some implementations, the present disclosure also includes the following embodiments: Figure 2The other operations performed at the data decompression device 203 are described in FIG. 10 .

[0154] Fig.13 and Fig.14 The following is a schematic diagram of the structure of possible communication devices provided in the embodiments of the present application. 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 embodiments of the present application, the communication device can be as follows: Figure 1A The terminal device 101 shown in FIG. 1 may also be Figure 1A The network device 103 shown may also be a module (e.g., a processor, a chip, or a chip system, etc.) applied to a terminal device or a network device, or may also be a logic module or software that can implement all or part of the functions of the terminal device or the network device. The terminal device or the network device may be implemented as a data compression device or a data decompression device.

[0155] like Fig.13 As shown, the communication device 1300 includes a transceiver module 1301 and a processing module 1302. The communication device 1300 can be used to implement the above Figure 2 , Fig.11 , Fig.12 The functions of the data compression device or the data decompression device in the method embodiment shown.

[0156] When the communication device 1300 is used to implement Figure 2 , Fig.11 The functions of the data compression device in the depicted method embodiment are: a transceiver module 1301, configured to send a first data matrix, and a projection result and a projection residual of a second data matrix. A processing module 1302, configured to determine a projection result according to a projection matrix determined by the second data matrix relative to the first data matrix, and to determine a projection residual according to the second data matrix and the projection result.

[0157] When the communication device 1300 is used to implement Figure 2 , Fig.12 The functions of the data decompression device in the described method embodiment are: a transceiver module 1301, which is used to receive a first data matrix, and a projection result and a projection residual of a second data matrix. A processing module 1302, which is used to obtain a second data matrix or an approximate matrix of the second data matrix based on the first data matrix, the projection result and the projection residual. The projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, and the projection residual is determined according to the second data matrix and the projection result.

[0158] like Fig.14As shown, the communication device 1400 includes a processor 1410 and an interface circuit 1420. The processor 1410 and the interface circuit 1420 are coupled to each other. It is understood that the interface circuit 1420 can be a transceiver or an input-output interface. Optionally, the communication device 1400 may also include a memory 1430 for storing instructions executed by the processor 1410 or storing input data required by the processor 1410 to execute instructions or storing data generated after the processor 1410 executes instructions.

[0159] When the communication device 1400 is used to implement the method in the above method embodiment, the processor 1410 is used to execute the function of the above processing module 1302, and the interface circuit 1420 is used to execute the function of the above transceiver module 1301.

[0160] When the above communication device is a chip applied to a terminal device, the terminal device chip implements the functions of the terminal device in the above method embodiment. The terminal device chip receives information from other modules in the terminal device (such as a radio frequency module or an antenna), and the information is sent by the network device to the terminal device; or the terminal device chip sends information to other modules in the terminal device (such as a radio frequency module or an antenna), and the information is sent by the terminal device to the network device.

[0161] When the above communication device is a chip applied to a network device, the network device chip implements the function of the network device in the above method embodiment. The network device chip receives information from other modules in the network device (such as a radio frequency module or an antenna), and the information is sent by the terminal device to the network device; or the network device chip sends information to other modules in the network device (such as a radio frequency module or an antenna), and the information is sent by the network device to the terminal device.

[0162] It is understood that the processor in the embodiments of the present application may be a central processing unit (CPU), or 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.

[0163] When the device in the embodiment of the present application is a network device, the device can be as follows Fig.15The device may include one or more radio frequency units, such as a remote radio unit (RRU) 1510 and one or more baseband units (BBU) (also referred to as digital units, digital units, DU) 1520. The RRU 1510 may be referred to as a transceiver module, which may include a transmitting module and a receiving module, or the transceiver module may be a module capable of transmitting and receiving functions. The transceiver module may be connected to Fig.13 The RRU 1510 corresponds to the transceiver module 1301 in the base station, that is, the actions performed by the transceiver module 1301 can be performed. Optionally, the transceiver module can also be called a transceiver, a transceiver circuit, or a transceiver, etc., which may include at least one antenna 1511 and a radio frequency unit 1512. The RRU 1510 part is mainly used for receiving and transmitting radio frequency signals and converting radio frequency signals into baseband signals. The BBU 1510 part is mainly used for baseband processing, controlling the base station, etc. The RRU 1510 and the BBU 1520 can be physically arranged together or physically separated, that is, a distributed base station.

[0164] The BBU 1520 is the control center of the base station, which can also be called a processing module. Fig.13 The processing module 1302 in the embodiment corresponds to the processing module 1303, which is mainly used to complete baseband processing functions, such as channel coding, multiplexing, modulation, spread spectrum, etc. In addition, the processing module can execute the actions executed by the processing module 1302. For example, the BBU (processing module) can be used to control the base station to execute the operation flow of the network device in the above method embodiment.

[0165] In one example, the BBU 1520 may be composed of one or more single boards, and multiple single boards may jointly support a wireless access network of a single access standard (such as an LTE network), or may respectively support wireless access networks of different access standards (such as an LTE network, a 5G network, or other networks). The BBU 1520 also includes a memory 1521 and a processor 1522. The memory 1521 is used to store necessary instructions and data. The processor 1522 is used to control the base station to perform necessary actions, such as controlling the base station to execute the operation process of the network device in the above method embodiment. The memory 1521 and the processor 1522 can serve one or more single boards. In other words, a memory and a processor may be separately set on each single board. It is also possible that multiple single boards share the same memory and processor. In addition, necessary circuits may also be set on each single board.

[0166] The present application embodiment provides a communication system. The communication system may include the above Figure 2The data compression device and the data decompression device involved in the embodiment shown are, for example, the terminal device 101 or the network device 103. Optionally, the terminal device and the network device in the communication system may execute Figure 2 , Fig.11 , Fig.12 Any of the communication methods shown in .

[0167] The present application also provides a circuit that can be coupled to a memory and can be used to execute a process related to a terminal device or a network device in any of the above method embodiments. The chip system may include the chip and other components such as a memory or a transceiver.

[0168] It should be understood that the processor mentioned in the embodiments of the present application may be a CPU, or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0169] It should also be understood that the memory mentioned in the embodiments of the present application 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 (doubledatarate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (synchlink DRAM, SLDRAM), and direct rambus RAM (DR RAM).

[0170] 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 in the processor.

[0171] It should be noted that the memory described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0172] It should be understood that in the various embodiments of the present application, 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 application.

[0173] It will be appreciated 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. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0174] It can be clearly understood 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.

[0175] In the several embodiments provided in the present application, 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 schematic. For example, the division of the module is a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that 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.

[0176] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0177] In addition, each functional module in each embodiment of the present application 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.

[0178] If the 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 embodiment of the present application is essentially or the part that contributes or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method of each embodiment of the present application. The aforementioned computer-readable storage medium can be any available medium that can be accessed by a computer. By way of example but 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.

[0179] 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 identical objects, and are used to distinguish the objects referred to, without implying a specific spatial order, temporal order, order of importance, etc. of the objects referred to. In some embodiments, values, processes, selected items, determined items, equipment, 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 among many available functional options, and such selections do not need to 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 cover a variety of actions. For example, "determine" can include calculation, calculation, processing, export, investigation, search (e.g., search in a table, database or another data structure), ascertainment, etc. Additionally, "determining" may include receiving (eg, receiving information), accessing (eg, accessing data in a memory), etc. Furthermore, "determining" may include resolving, selecting, choosing, establishing, etc.

Claims

1. A method comprising: sending a first data matrix; as well as Sending the projection result and projection residual of the second data matrix; wherein the projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, The projection residual is determined according to the second data matrix and the projection result.

2. A method comprising: receiving a first data matrix; receiving a projection result and a projection residual of a second data matrix; as well as Based on the first data matrix, the projection result and the projection residual, obtaining the second data matrix or an approximate matrix of the second data matrix, wherein the projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, The projection residual is determined according to the second data matrix and the projection result.

3. The method according to claim 1 or 2, wherein the projection matrix is ​​determined according to the following manner: A predetermined number of columns are selected in a reference matrix to form the projection matrix, the reference matrix being a sub-matrix of the first data matrix or the first data matrix.

4. The method according to claim 1 or 2, wherein the projection matrix is ​​determined according to the following manner: Determine a projection basis matrix based on a reference matrix; and selecting a predetermined number of columns in the projection basis matrix to form the projection matrix, The reference matrix is ​​the first data matrix or a submatrix of the first data matrix.

5. The method according to claim 4, wherein determining the projection basis matrix according to the reference matrix comprises: A matrix decomposition is performed on the reference matrix to determine a projection basis matrix.

6. The method according to claim 5, wherein the matrix decomposition comprises any one of the following: Low Rank Matrix Approximation (LRMA) decomposition, SVD decomposition, or QR decomposition.

7. The method according to any one of claims 1 or 3-6, further comprising sending at least one of the following: First indication information, used to indicate the number of columns of the projection matrix; Second indication information, used to indicate whether the projection matrix is ​​based on the reference matrix or on a projection basis matrix obtained by performing LRMA decomposition on the reference matrix; or The third indication information is used to indicate the position of the column of the projection matrix in the reference matrix or the projection base matrix.

8. The method according to any one of claims 2 to 6, further comprising receiving at least one of the following: First indication information, used to indicate the number of columns of the projection matrix; Second indication information, used to indicate whether the projection matrix is ​​based on the reference matrix or on a projection basis matrix obtained by performing LRMA decomposition on the reference matrix; or The third indication information is used to indicate the position of the column of the projection matrix in the reference matrix or the projection base matrix.

9. The method of claim 1 , wherein determining the projection matrix comprises: selecting a subspace of a predetermined number of dimensions in the column space spanned by the reference matrix, wherein the projection matrix represents the subspace, The reference matrix is ​​the first data matrix or a submatrix of the first data matrix. 10 . The method according to claim 9 , wherein the subspace is selected so as to minimize the norm of the projection residual after the second data matrix is ​​projected onto the projection matrix.

11. The method of claim 9 or 10, wherein determining the projection matrix comprises: Performing QR decomposition on the reference matrix to determine a Q matrix; Performing singular value decomposition on the product of the transpose of the Q matrix and the data matrix to obtain an eigenvector matrix; determining a subspace selection matrix based on the first predetermined number of columns of the eigenvector matrix; as well as The projection matrix is ​​determined based on the subspace selection matrix and the reference matrix.

12. The method of claim 9 or 10, wherein determining the projection matrix comprises: Perform QR decomposition on the reference data C to obtain a Q matrix; C=Q×R, The columns are orthogonal, For the matrix Q T Perform SVD decomposition on G to obtain the U matrix, where G is the second data matrix; Q T G=U∑V T Get the first d columns of matrix U, denoted as Get the subspace selection matrix Get the projection matrix P * =C×S * .

13. The method of claim 9 or 10, wherein determining the projection matrix comprises: Solve the following optimization problem, where G is the second data matrix and C is the reference matrix 14. The method according to any one of claims 1, 9-11, further comprising: Fourth indication information is sent, where the fourth indication information is used to indicate a subspace selection matrix, wherein the subspace selection matrix is ​​used to determine the projection matrix together with the reference matrix.

15. The method according to any one of claims 2, 9-11, further comprising: Fourth indication information is received, where the fourth indication information is used to indicate a subspace selection matrix, wherein the subspace selection matrix is ​​used to determine the projection matrix together with the reference matrix.

16. The method according to claim 1, further comprising: receiving feedback information for the first data matrix; as well as The second data matrix is ​​determined based on the feedback information and the first data matrix.

17. The method according to claim 1, further comprising: Sending fifth indication information for the first data matrix; as well as Based on the fifth indication information and the first data matrix, the second data matrix is ​​determined.

18. The method according to claim 2, further comprising: Feedback information for the first data matrix is ​​sent, where the feedback information is used together with the first data matrix to determine the second data matrix.

19. The method according to claim 2, further comprising: receiving fifth indication information for the first data matrix; The fifth indication information is used together with the first data matrix to determine the second data matrix.

20. The method according to any one of claims 16 to 19, wherein the feedback information or the fifth indication information comprises at least one of the following: Subset indication information, used to indicate the first data matrix or a sub-matrix of the first data matrix; or Confidence information is used to indicate the confidence of the first data matrix.

21. The method of any one of claims 1-20, wherein the first data matrix has a first granularity, the second data matrix has a second granularity, and the first granularity is greater than the second granularity.

22. The method of claim 21, wherein at least one of the following: The first data is sampled data of a geographic space using a first resolution, the second data matrix is ​​sampled data of the geographic space using a second resolution higher than the first resolution, and the subset indication information indicates spatial position information of the subset of the first data matrix; or The first data matrix is ​​a plurality of cluster centers of a plurality of data classes determined by performing clustering on the original data, the second data matrix is ​​the data contained in one or more data classes among the plurality of data classes, and the subset indication information indicates one or more cluster centers among the plurality of cluster centers corresponding to the one or more data classes.

23. The method according to any one of claims 1 to 22, wherein the reference matrix is ​​determined based on at least one of: Feedback information for the first data matrix; a submatrix of the first data matrix; or The first data matrix.

24. The method according to any one of claims 2 to 23, wherein: The second data matrix is ​​not grouped but corresponds to the reference matrix as a whole; or The second data matrix is ​​divided into a plurality of data groups, and the plurality of data groups respectively correspond to a plurality of sub-matrices of the reference matrix.

25. The method according to any one of claims 1, 3-7, 9-14, 16, 17, 20-24, further comprising sending at least one of the following: first configuration information, used to indicate whether the feedback information for the first data matrix is ​​based on an index or a bitmap; second configuration information, used to indicate whether confidence feedback for the first data matrix is ​​enabled; third configuration information, used to indicate whether the first data matrix is ​​based on geographic location or clustering; fourth configuration information, used to indicate whether the reference matrix is ​​determined based on feedback information for the first data matrix or based on the first data matrix; or The fifth configuration information is used to indicate whether the second data matrix as a whole corresponds to the reference matrix, or whether the multiple data groups divided into the second data matrix correspond to the multiple sub-matrices of the reference matrix respectively.

26. The method of claim 25, wherein at least one of the first configuration information, the second configuration information, the third configuration information, the fourth configuration information, or the fifth configuration information is transmitted before the first data matrix is ​​transmitted.

27. The method of any one of claims 2, 8, 15, 18-24, further comprising receiving at least one of: first configuration information, used to indicate whether the feedback information for the first data matrix is ​​based on an index or a bitmap; second configuration information, used to indicate whether confidence feedback for the first data matrix is ​​enabled; third configuration information, used to indicate whether the first data matrix is ​​based on geographic location or clustering; fourth configuration information, used to indicate whether the reference matrix is ​​determined based on feedback information for the first data matrix or based on the first data matrix; or The fifth configuration information is used to indicate whether the second data matrix as a whole corresponds to the reference matrix, or whether the multiple data groups divided into the second data matrix correspond to the multiple sub-matrices of the reference matrix respectively.

28. The method of claim 27, wherein at least one of the first configuration information, the second configuration information, the third configuration information, the fourth configuration information, or the fifth configuration information is received before sending the first data matrix.

29. The method of any one of claims 1, 3-7, 9-14, 16, 17, 20-26, further comprising sending at least one of the following: Sixth configuration information, used to indicate whether the projection matrix is ​​based on the column space of the reference matrix or on a subspace of the column space; or The seventh configuration information is used to identify the reference matrix from the first data matrix. 30 . The method according to claim 29 , wherein at least one of the sixth configuration information or the seventh configuration information is sent before sending the projection result and the projection residual.

31. The method of any one of claims 2, 8, 18-24, 27, 28, further comprising receiving at least one of: Sixth configuration information, used to indicate whether the projection matrix is ​​based on the column space of the reference matrix or on a subspace of the column space; or The seventh configuration information is used to identify the reference matrix from the first data matrix. 32 . The method according to claim 31 , wherein at least one of the sixth configuration information or the seventh configuration information is received before the projection result and the projection residual are sent.

33. An apparatus comprising: A first data matrix sending module, used for sending a first data matrix; as well as A result sending module, used for sending the projection result and projection residual of the second data matrix; wherein the projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, The projection residual is determined according to the second data matrix and the projection result.

34. An apparatus comprising: A first data matrix receiving module, used for receiving a first data matrix; A result receiving module, used for receiving the projection result and projection residual of the second data matrix; as well as A second data matrix acquisition module is used to acquire the second data matrix or an approximate matrix of the second data matrix based on the first data matrix, the projection result and the projection residual. wherein the projection result is determined according to a projection matrix determined by the second data matrix relative to the first data matrix, The projection residual is determined according to the second data matrix and the projection result.

35. A system comprising the apparatus of claims 33 and 34.

36. An apparatus comprising: A processor, and a memory storing instructions, wherein when the instructions are executed by the processor, the terminal device executes the method according to any one of claims 1 to 32.

37. A computer-readable storage medium storing instructions, which, when executed by an electronic device, causes the electronic device to perform the method according to any one of claims 1 to 32.

38. A computer program product, comprising instructions, which, when executed by an electronic device, cause the electronic device to perform the method according to any one of claims 1 to 32.

Images