Communication method and corresponding device

Abstract

The embodiment of the invention provides a communication method and corresponding equipment. A method performed by a communication node may include obtaining channel state information on a second resource using a first neural network; obtaining second information by using a second neural network based on the channel state information; and updating the first neural network by using the second information.

Description

Technical Field

[0001] The present disclosure relates to the field of communication technology, and in particular, to a method executed by a communication node, a communication node, and a computer-readable storage medium. Background Art

[0002] Considering the development of wireless communication from generation to generation, these technologies are mainly developed for services targeting people, such as voice calls, multimedia services, and data services. With the commercialization of the fifth-generation (5G) communication system, the number of connected devices is expected to grow exponentially. These will be increasingly connected to the communication network. Examples of the Internet of Things can include vehicles, robots, drones, home appliances, displays, smart sensors connected to various infrastructures, construction machinery, and factory equipment. Mobile devices are expected to develop in various forms such as augmented reality glasses, virtual reality headsets, and holographic devices. Efforts have been made to develop improved 6G communication systems in order to provide various services by connecting hundreds of billions of devices and things in the sixth-generation (6G) era. For these reasons, 6G communication systems are called super 5G systems.

[0003] The 6G communication system, which is expected to be commercialized around 2030, will have a peak data rate of tera (1,000 gigabits) bps and a radio latency of less than 100 μsec, thus being 50 times the data rate of the 5G communication system and having 1 / 10 of its radio latency.

[0004] In order to achieve such high data rates and ultra-low latency, the implementation of 6G communication systems in terahertz (e.g., 95GHz to 3THz bands) has been considered. It is expected that since the path loss and atmospheric absorption in the terahertz band are more serious than those in the millimeter wave (mmWave) band introduced in 5G, the technology that can ensure the signal transmission distance (i.e., coverage) will become more critical. As the main technology to ensure coverage, it is necessary to develop radio frequency (RF) elements, antennas, new waveforms with better coverage than orthogonal frequency division multiplexing (OFDM), beamforming and massive multiple input multiple output (MIMO), full dimensional multiple input multiple output (FD-MIMO), array antennas, and multi-antenna transmission technologies such as massive antennas. In addition, new technologies have been discussed to improve signal coverage in the terahertz band, such as metamaterial-based lenses and antennas, orbital angular momentum (OAM), and reconfigurable intelligence surface (RIS).

[0005] In addition, in order to improve spectrum efficiency and overall network performance, the following technologies have been developed for 6G communication systems: full-duplex technology for enabling uplink and downlink transmissions to use the same frequency resources simultaneously; network technology that utilizes satellites, high-altitude platform stations (HAPS), etc. in an integrated manner; improved network structure to support mobile base stations, etc., and to enable network operation optimization and automation, etc.; dynamic spectrum sharing technology with conflict avoidance based on spectrum usage prediction: using artificial intelligence (AI) in wireless communications to improve overall network operations by utilizing AI from the design stage of developing 6G and internalizing end-to-end AI support functions; and next-generation distributed computing technology that overcomes the computing power limitations of user equipment (UE) through ultra-high performance communication and computing resources achievable on the network (such as mobile edge computing (MEC), cloud, etc.). In addition, attempts are continuing to strengthen connectivity between devices, optimize networks, promote software-based network entities, and increase openness of wireless communications by designing new protocols to be used in 6G communication systems, developing mechanisms for implementing hardware-based security environments and secure use of data, and developing technologies for maintaining privacy.

[0006] It is expected that the research and development of 6G communication systems including hyperconnectivity of person to machine (P2M) and machine to machine (M2M) will bring about the next hyperconnectivity experience. In particular, services such as truly immersive extended reality (XR), high-fidelity mobile holograms, and digital replicas are expected to be provided through 6G communication systems. In addition, services such as remote surgery, industrial automation, and emergency response for enhanced safety and reliability will be provided through 6G communication systems, making the technology applicable to various fields such as industry, healthcare, automobiles, and home appliances. Summary of the invention

[0007] The purpose of the present disclosure is to solve at least one of the technical defects in the existing communication methods to better meet the communication needs. In order to achieve this purpose, the technical solution proposed in the present disclosure is as follows:

[0008] According to a first aspect of an embodiment of the present disclosure, there is provided a method performed by a communication node, comprising: obtaining channel state information on a second resource using a first neural network; obtaining second information based on the channel state information using a second neural network; and updating the first neural network using the second information.

[0009] Optionally, using the first neural network to obtain channel state information on the second resource includes: receiving a first signal on the first resource to obtain first information; and based on the first information, using the first neural network to obtain channel state information on the second resource.

[0010] Optionally, the first information includes at least one of the following: the first signal on the first resource, the first signal including all signals received on the first resource; a reference signal on the first resource; a second signal after specific processing of the first signal and / or the reference signal; feedback channel state information corresponding to the first resource; channel information reversely estimated based on received data; physical resources containing the reference signal; channel state information measurement report information; first related information, the first related information including at least one of information indicating the position of the reference signal in the first resource, the reference signal, information related to the first resource and / or the second resource, and characteristic quantity constraint information of the channel in the current network environment.

[0011] Optionally, the second information includes at least one of the following: prediction information for data transmission based on the channel state information, the prediction information including at least one of the probability of successful or failed data transmission, the bit error rate of data transmission, the symbol error rate of data transmission, the bit error rate of data transmission, and the signal-to-interference ratio generated by data transmission; error evaluation information of the channel state information, the error evaluation information including at least one of the estimated error of the channel state information, the quality classification label of the channel state information estimation, the deviation measure between the characteristic information of the channel state information and the preset parameter value in the second neural network, and the probability of occurrence of the channel state information in the current network environment.

[0012] Optionally, the method further includes: obtaining fourth information for assisting data transmission using a third neural network based on the second information and / or third information related to data transmission.

[0013] Optionally, the third information includes at least one of the following: relevant parameters of data transmission, including: modulation mode, coding rate, available physical resources, amount of data to be transmitted in the buffer, quality of service requirements, and at least one of the precoding codebook sets; relevant information of the network environment, including: at least one of the empirical data of environmental interference level, degree of environmental change, and relevant time of the environment.

[0014] Optionally, the fourth information includes at least one of the following: data transmission status indication information, used to indicate whether the channel state information is used for data transmission and / or whether to redo the complete channel estimation process and / or whether to update the first neural network based on the second information; recommended information on data transmission parameters, used to indicate the data transmission parameters that the communication node should use.

[0015] Optionally, in the case where the communication node is a user equipment, the fourth information is used by the user equipment for at least one of the following: determining whether to trigger re-performing a complete channel estimation process; reporting the fourth information to the base station as reference information for the base station to schedule data transmission; wherein whether to trigger re-performing a complete channel estimation process includes whether to trigger at least one of a non-periodic channel state information measurement report and sending information to the base station regarding whether the user equipment should re-perform a complete channel estimation process.

[0016] Optionally, when the communication node is a base station, the fourth information is used by the base station for at least one of the following: determining whether to trigger re-performing a complete channel estimation process; determining whether to use the channel state information for data transmission; determining whether to configure parameters used for data transmission; wherein determining whether to use the channel state information for data transmission includes determining resource allocation between user devices for data transmission; wherein whether to trigger re-performing a complete channel estimation process includes at least one of: whether to trigger non-periodic channel state information measurement reporting, and whether to schedule the user equipment to send a reference signal.

[0017] Optionally, the method further includes: obtaining fifth information related to performing data transmission based on the channel state information; and updating at least part of the parameters of the second neural network using the second information and / or the fifth information.

[0018] Optionally, the method further includes: obtaining fifth information related to performing data transmission based on the channel state information and / or the third information and / or the fourth information; and updating at least part of the parameters of the third neural network using the fourth information and / or the fifth information.

[0019] Optionally, the fifth information includes at least one of the following: feedback information of data transmission; at least one of the cumulative bit error rate, cumulative symbol error rate, cumulative success rate, and cumulative block success rate of data transmission based on the channel state information obtained by the first neural network since the last channel estimation process.

[0020] Optionally, the first neural network, the second neural network and the third neural network are trained jointly or individually.

[0021] Optionally, at least one of the first neural network, the second neural network and the third neural network is updated based on a preset time interval, or is updated based on the number of output information, or is updated based on the triggering of a specific event, wherein the specific event includes at least one of the prediction information or error evaluation information in the second information exceeding a preset threshold, the communication node failing to decode received data, receiving a retransmission instruction or a parameter update instruction, and receiving feedback negative acknowledgment NACK information.

[0022] Optionally, using the second information to update the first neural network includes: converting the second information into an error amount; according to the back-propagation rule, back-propagating the error amount layer by layer according to each neural network layer included in the first neural network to obtain the error amount corresponding to the parameters in each neural network layer; using an optimization algorithm to calculate the update amount of the parameter based on the error amount corresponding to the parameter and update the parameter.

[0023] Optionally, the first resource includes the second resource, or the second resource includes the first resource, or the first resource and the second resource include at least part of the same resource, or the first resource and the second resource do not include the same resource.

[0024] According to a second aspect of an embodiment of the present disclosure, a communication node is provided, which may include: a transceiver; and a processor, coupled to the transceiver and configured to execute the method performed by the communication node as described above.

[0025] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising: at least one processor; and at least one memory storing computer executable instructions, wherein the computer executable instructions, when executed by the at least one processor, prompt the at least one processor to execute the communication method as described above.

[0026] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium storing instructions is provided. When the instructions are executed by at least one processor, the at least one processor is prompted to execute the communication method as described above.

[0027] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects:

[0028] The online training and updating of the parameters of the first neural network are realized to avoid the decrease of the accuracy of the acquisition of the channel state information due to the change of the external environment.

[0029] The beneficial effects brought about by the technical solutions provided by the embodiments of the present disclosure will be described below in conjunction with specific optional embodiments, or can be learned from the description of the embodiments, or can be known through the implementation of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly and easily illustrate and understand the technical solutions in the embodiments of the present disclosure, the drawings required for describing the embodiments of the present disclosure are briefly introduced below.

[0031] Figure 1 An example wireless network according to an embodiment of the present disclosure is shown;

[0032] Figure 2 An example base station according to an embodiment of the present disclosure is shown;

[0033] Figure 3 An example user device according to an embodiment of the present disclosure is shown;

[0034] Figure 4 A schematic diagram showing SRS resource configuration using two antenna ports is shown;

[0035] Figure 5 is a flow chart of a method performed by a communication node according to an embodiment of the present disclosure;

[0036] Figure 6 is a flowchart of a method performed by a communication node according to another embodiment of the present disclosure;

[0037] Figure 7 is a flowchart of a method for online updating of a second neural network according to an embodiment of the present disclosure;

[0038] Figure 8 is a flowchart of a method for online updating of a third neural network according to an embodiment of the present disclosure;

[0039] Fig. 9 A schematic diagram of the structure of a first neural network according to an embodiment of the present disclosure is shown;

[0040] Fig.10 A schematic diagram of the structure of a second neural network according to an embodiment of the present disclosure is shown;

[0041] Fig.11 A schematic diagram of the structure of a third neural network according to an embodiment of the present disclosure is shown;

[0042] Fig.12 A schematic diagram of the structure of a neural network unit according to an embodiment of the present disclosure is shown;

[0043] Fig.13A block diagram of a communication node in a communication system according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0044] Before carrying out the following specific embodiments, it may be advantageous to set forth the definitions of certain words and phrases used throughout the patent document. The term "connection" and its derivatives refer to any direct or indirect communication between two or more elements, regardless of whether those elements are in physical contact with each other. The terms "send", "receive" and "communicate" and their derivatives include both direct and indirect communication. The terms "include" and "include" and their derivatives mean unrestricted inclusion. The term "or" is compatible, meaning and / or. The phrase "associated with" and its derivatives mean including, included, interconnected with it, including, included, connected to or connected with it, coupled to or coupled with it, can communicate with it, collaborate with it, interweave, juxtapose, approach, be bound to or bound with it, have, have the property of, have to ... or have a relationship with ... etc. The term "controller" means any device, system or part thereof that controls at least one operation. Such a controller can be implemented in a hardware manner or in a combination of hardware and software and / or firmware. The functionality associated with any particular controller, whether local or remote, may be centralized or distributed. The phrase "at least one of" when used with a list of items means that different combinations of one or more of the listed items may be used, and only one item in the list may be required. For example, "at least one of A, B, and C" includes any of the following combinations: A, B, C, A and B, A and C, B and C, and only A, only B, only C. Likewise, the term "set" means one or more. Thus, a set of items may be a single item or a set of two or more items.

[0045] Moreover, various functions as described below can be implemented or supported by one or more computer programs, each of which is formed by a computer-readable program code and embodied in a computer-readable medium. The terms "application" and "program" refer to one or more computer programs, software components, instruction sets, processes, functions, objects, categories, instances, related data, or a portion thereof suitable for implementation in a suitable computer-readable program code. The phrase "computer-readable program code" includes any type of computer code, including source code, object code, and executable code. The phrase "computer-readable medium" includes any type of medium that can be accessed by a computer, such as a read-only memory (ROM), a random access memory (RAM), a hard drive, a compact disc (CD), a digital video disc (DVD), or any other type of memory. "Non-transitory" computer-readable media excludes wired, wireless, optical, or other communication links that convey instantaneous electrical signals or other signals. Non-transitory computer-readable media include media in which data can be permanently stored and media in which data can be stored and rewritten later, such as rewritable optical discs or erasable memory devices.

[0046] Definitions for certain other words and phrases are provided throughout this patent document. Those of skill in the art should understand that in many, if not most instances, such definitions apply to prior, as well as future uses of such defined words and phrases.

[0047] The figures and various embodiments used to describe the principles of the present disclosure are included herein for illustration only and should not be construed in any way as limiting the scope of the present disclosure. In addition, those skilled in the art will appreciate that the principles of the present disclosure can be implemented in any appropriately arranged wireless communication system.

[0048] The following Figures 1 to 3 Various embodiments of the present disclosure are described as being implemented in a wireless communication system. Figures 1 to 3 The description is not meant to imply physical or architectural limitations to the manner in which different embodiments may be implemented. Different embodiments of the present disclosure may be implemented in any suitably arranged communications system.

[0049] Figure 1 An example wireless network according to an embodiment of the present disclosure is shown. Figure 1 The embodiment of the wireless network shown in FIG. 1 is for illustration only. Other embodiments of the wireless network 100 may be used without departing from the scope of the present disclosure.

[0050] like Figure 1As shown, the wireless network includes a base station (next generation nodeB, gNB or gNodeB) 101, gNB 102, and gNB 103. gNB 101 communicates with gNB 102 and gNB 103. gNB 101 also communicates with at least one network 130 such as the Internet, a proprietary Internet Protocol (IP) network, or other data network.

[0051] The gNB 102 provides wireless broadband access to a network 130 for a plurality of first user equipments (UEs) within a coverage area 120 of the gNB 102. The plurality of first UEs include a UE 111 that may be located at a small business (SB); a UE 112 that may be located at an enterprise (E); a UE 113 that may be located at a WiFi hot spot (HS); a UE 114 that may be located at a first residence (R1); a UE 115 that may be located at a second residence (R2); and a UE 116 that may be a mobile device (M) such as a cellular phone, a wireless laptop, a wireless personal digital assistant (PDA), etc. The gNB 103 provides wireless broadband access to the network 130 for a plurality of second UEs within a coverage area 125 of the gNB 103. The plurality of second UEs include UE 115 and UE 116 and subscriber stations (SS, e.g., UEs) 117, 118, and 119. In some embodiments, one or more of gNBs 101-103 may communicate with each other and UEs 111-116 using existing wireless communication technologies, and one or more of UEs 111-119 may communicate directly with each other (e.g., UEs 117-119) using other existing or proposed wireless communication technologies.

[0052] Depending on the type of network, the term "base station" or "BS" may refer to any component (or collection of components) configured to provide wireless access to a network, such as a transmit point (TP), a transmit-receive point (TRP), an enhanced (or "evolved") base station (eNodeB or eNB), a 5G base station (gNB), a macro cell, a femto cell, a wireless fidelity (WiFi) access point (AP), or other wireless capable devices. A base station may provide wireless access according to one or more wireless communication protocols, such as 3GPP 5G new radio (NR), Long Term Evolution (LTE), LTE-Advanced (LTE-A), high speed packet access (HSPA), Wi-Fi 802.11a / b / g / n / ac, etc. For convenience, various names of base station type devices and functions are used interchangeably in this patent document to refer to network infrastructure components that provide wireless access to remote terminals. Also, depending on the network type, the term "user equipment" (UE) may refer to any component such as a mobile station (MS), a subscriber station (SS), a remote terminal, a wireless terminal, a reception point, or a user device, etc. For convenience, various names of user equipment type devices and functions are used interchangeably in this patent document to refer to a remote wireless device that wirelessly accesses a BS regardless of whether the UE is a mobile device (such as a mobile phone or a smart phone) or a remote wireless device that is generally considered to be a fixed device (such as a desktop computer or a vending machine).

[0053] Dashed lines illustrate the approximate extents of coverage areas 120 and 125, which are shown as generally circular for purposes of illustration and explanation only. It should be clearly understood that coverage areas associated with a gNB, such as coverage areas 120 and 125, may have other shapes, including irregular shapes, depending on the configuration of the gNB and variations in the wireless environment associated with natural and artificial obstacles.

[0054] As described in more detail below, one or more of UEs 111-119 include circuitry, programming, or a combination thereof. In certain embodiments, one or more of gNBs 101-103 include circuitry, programming, or a combination thereof.

[0055] although Figure 1 An example of a wireless network is shown, but Figure 1Various changes may be made. For example, wireless network 100 may include any number of gNBs and any number of UEs in any suitable arrangement. Moreover, gNB 101 may communicate directly with any number of UEs and provide those UEs with wireless broadband access to network 130. Similarly, each gNB 102-103 may communicate directly with network 130 and provide UEs with direct wireless broadband access to network 130. In addition, gNBs 101, 102, and / or 103 may provide access to other or additional external networks such as an external telephone network or other types of data networks.

[0056] Figure 2 An example base station according to an embodiment of the present disclosure is shown. Figure 2 The embodiment of the gNB 102 shown in FIG. 1 is for illustration only, and Figure 1 gNBs 101 and 103 may have the same or similar configurations. However, gNBs appear in a variety of configurations, and Figure 2 The scope of the present disclosure is not limited to any particular implementation of the gNB.

[0057] like Figure 2 As shown in FIG. 1 , gNB 102 includes multiple antennas 200a-200n, multiple radio frequency (RF) transceivers 201a-201n, transmit (TX) processing circuitry 203, and receive (RX) processing circuitry 204. gNB 102 also includes a controller / processor 205, a memory 206, and a backhaul or network interface 207.

[0058] The RF transceivers 201a-201n receive incoming RF signals from the antennas 200a-200n, such as signals sent by UEs in the network 100. The RF transceivers 201a-201n down-convert the incoming RF signals to generate intermediate frequency (IF) or baseband signals. The IF or baseband signals are sent to the RX processing circuit 204, which generates processed baseband signals by filtering, decoding and / or digitizing the baseband or IF signals. The RX processing circuit 204 sends the processed baseband signals to the controller / processor 205 for further processing.

[0059] TX processing circuit 203 receives analog or digital data (such as voice data, web data, email, or interactive video game data) from controller / processor 205. TX processing circuit 203 encodes, multiplexes, and / or digitizes the outgoing baseband data to generate processed baseband or IF signals. RF transceivers 201a-201n receive the outgoing processed baseband or IF signals from TX processing circuit 203 and up-convert the baseband or IF signals to RF signals that are transmitted via antennas 201a-201n.

[0060] Controller / processor 205 may include one or more processors or other processing devices that control the overall operation of gNB 102. For example, controller / processor 205 may control the reception of forward channel signals and the transmission of reverse channel signals by RF transceivers 201a-201n, RX processing circuitry 204, and TX processing circuitry 203 in accordance with well-known principles. Controller / processor 205 may also support additional functionality, such as more advanced wireless communication functionality.

[0061] For example, the controller / processor 205 may support beamforming or directional routing operations in which outgoing signals from multiple antennas 200a-200n are weighted differently to effectively steer the outgoing signals in a desired direction. Any of a variety of other functions may be supported in the gNB 102 by the controller / processor 205.

[0062] The controller / processor 205 is also capable of executing programs and other processes, such as an operating system (OS), located in the memory 206. The controller / processor 205 can move data into or out of the memory 206 as required by the executing process.

[0063] The controller / processor 205 is also connected to a backhaul or network interface 207. The backhaul or network interface 207 allows the gNB 102 to communicate with other devices or systems via a backhaul connection or via a network. The interface 207 may support communication over any appropriate (one or more) wired or wireless connections. For example, when the gNB 102 is implemented as part of a cellular communication system (such as a cellular communication system supporting 5G, LTE, or LTE-A), the interface 207 may allow the gNB 102 to communicate with other gNBs via a wired or wireless backhaul connection. When the gNB 102 is implemented as an access point, the interface 207 may allow the gNB 102 to communicate over a wired or wireless local area network or over a wired or wireless connection to a larger network (such as the Internet). The interface 207 includes any appropriate structure that supports communication over a wired or wireless connection such as an Ethernet or RF transceiver.

[0064] Memory 206 is connected to controller / processor 205. A portion of memory 206 may include random access memory (RAM), and another portion of memory 206 may include flash memory or other read-only memory (ROM).

[0065] although Figure 2 An example of gNB 102 is shown, but the Figure 2 For example, gNB 102 may include any number of Figure 2 As a specific example, the access point may include multiple interfaces 207, and the controller / processor 205 may support routing functions to route data between different network addresses. As another specific example, although shown as including a single instance of TX processing circuitry 203 and a single instance of RX processing circuitry 204, the gNB 102 may include multiple instances of each (such as one for each RF transceiver). For example, Figure 2 The various components in may be combined, further subdivided, or omitted, and additional components may be added according to specific needs.

[0066] Figure 3 An example user device according to an embodiment of the present disclosure is shown. Figure 3 The embodiment of UE 116 shown in FIG. 1 is for illustration only, and Figure 1 UEs 111-115 and 117-119 may have the same or similar configurations. However, UEs may appear in a variety of configurations, and Figure 3 The scope of the present disclosure is not limited to any particular implementation of the UE.

[0067] like Figure 3 As shown in FIG. 1 , UE 116 includes an antenna 301, a radio frequency (RF) transceiver 302, a TX processing circuit 303, a microphone 304, and a receive (RX) processing circuit 305. UE 116 also includes a speaker 306, a controller or processor 307, an input / output (I / O) interface (IF) 308, an input device 309, a touch screen display 310, and a memory 311. The memory 311 includes an OS 312 and one or more applications 313.

[0068] RF transceiver 302 receives incoming RF signals transmitted by a gNB of network 100 from antenna 301. RF transceiver 302 downconverts the incoming RF signals to generate an IF or baseband signal. The IF or baseband signal is sent to RX processing circuitry 305, which generates a processed baseband signal by filtering, decoding, and / or digitizing the baseband or IF signal. RX processing circuitry 305 sends the processed baseband signal to speaker 306 (such as for voice data) or processor 307 for further processing (such as for web browsing data).

[0069] The TX processing circuit 303 receives analog or digital voice data from the microphone 304 or other outgoing baseband data (such as web data, email, or interactive video game data) from the processor 307. The TX processing circuit 303 encodes, multiplexes, and / or digitizes the outgoing baseband data to generate a processed baseband or IF signal. The RF transceiver 302 receives the outgoing processed baseband or IF signal from the TX processing circuit 303 and up-converts the baseband or IF signal to an RF signal that is transmitted via the antenna 301.

[0070] The processor 307 may include one or more processors or other processing devices and executes the OS 312 stored in the memory 311 to control the overall operation of the UE 116. For example, the processor 307 may control the reception of forward channel signals and the transmission of reverse channel signals by the RF transceiver 302, the RX processing circuit 305, and the TX processing circuit 303 according to well-known principles. In some embodiments, the processor 307 includes at least one microprocessor or microcontroller.

[0071] The processor 307 is also capable of executing other processes and programs located in the memory 311, such as a process for CSI reporting on the uplink channel. The processor 307 can move data into or out of the memory 311 as needed to execute the process. In some embodiments, the processor 307 is configured to execute the application 313 based on the OS 312 or in response to a signal received from the gNB or operator. The processor 307 is also coupled to the I / O interface 308, which provides the UE 116 with the ability to connect to other devices such as laptops and portable computers. The I / O interface 308 is the communication path between these accessories and the processor 307.

[0072] Processor 307 is also connected to touch screen display 310. A user of UE 116 may use touch screen display 310 to enter data into UE 116. Touch screen display 310 may be a liquid crystal display, light emitting diode display, or other display capable of rendering text and / or at least limited graphics, such as from a website.

[0073] The memory 311 is connected to the processor 307. A portion of the memory 311 may include RAM, and another portion of the memory 311 may include flash memory or other ROM.

[0074] although Figure 3 An example of UE 116 is shown, but the Figure 3 Make various changes. For example, Figure 3 The various components in the embodiment may be combined, further subdivided, or omitted, and additional components may be added as required. As a specific example, processor 307 may be divided into multiple processors, such as one or more central processing units (CPUs) and one or more graphics processing units (GPUs). Furthermore, although Figure 3 The UE 116 is shown configured as a mobile phone or smart phone, but the UE may be configured to operate as other types of mobile or stationary devices.

[0075] One of the main means to ensure communication rate and reliability in current wireless communication systems is to process signals or adjust the sending and receiving decisions through channel state information (CSI) between two nodes transmitting information (two nodes are used as a sending node and a receiving node, or both nodes are used as a sending node and a receiving node). Figure 2 Before sending a signal, the gNB (as shown in FIG) precodes the transmitted signal according to its own known channel state information and adjusts the power and modulation coding mode of the transmitted signal. The node (such as Figure 3After receiving the signal, the UE shown in the figure performs channel equalization on the received signal according to its own known channel state information, selects different transmit and receive beams, etc., so as to eliminate the influence of the channel on the signal as much as possible. As an example only and not a limitation, the channel state information can be at least one of the following: channel response, reference signal receive power (RSRP), signal to interference and noise ratio (SINR), channel quality indicator (CQI), precoding matrix indication (precoding matrix), layer indicator (LI), rank indicator (RI), resource indicator (resource indicator).

[0076] At present, the booming development of smart mobile devices and related applications has stimulated users' demand for faster and lower latency wireless communications. These demands have prompted wireless communication networks to evolve towards wider bandwidths and more antennas. Accordingly, the resources and difficulty required to obtain wireless channel state information have also increased. In order to further improve the spectrum efficiency of wireless communication systems, both the industry and academia have regarded how to efficiently obtain and use channel state information as one of the key challenges for future wireless communication systems.

[0077] In an existing wireless communication system, for the above-mentioned method of ensuring communication quality by eliminating the influence of the channel on the signal, the receiving node and / or the sending node needs to know the channel state information of each antenna used for transmission at any time point and any frequency point of information transmission. Without loss of generality, the channel state information described in the present disclosure is the comprehensive influence of the amplitude and phase of the signal of the entire wireless communication link on all physical resources occupied / used by its transmission. Here, the physical resource is a resource entity that can be used for signal transmission (for example, a demodulation reference signal (DM-RS), a channel state information reference signal (CSI-RS), a sounding reference signal (SRS)) in a communication system. For example, the resource entity can be a time domain physical resource, a frequency domain physical resource, and an antenna domain physical resource. Without loss of generality, the antenna described in the present disclosure is a generalized antenna, which characterizes the spatial resources occupied by signal transmission. The antenna domain physical resources may be a physical antenna element, an antenna array including multiple physical antenna elements, a beam, an antenna port, or a precoding matrix of a transmitter, etc.; the time point represents the time domain physical resources occupied by signal transmission, which may be a duration with a time unit as the minimum granularity, wherein the time unit has a predetermined time length, wherein the time unit may be a wireless frame, a sub-frame, a slot, a sub-slot or a symbol, etc. As an example only and not a limitation, for example, the time point may be a time slot consisting of 14 OFDM symbols (the time unit is an OFDM symbol); the frequency point represents the frequency domain resources occupied by signal transmission, which is a frequency range with a frequency unit as the smallest granularity, wherein the frequency unit has a predetermined spectrum width, and may be a sub-band, a bandwidth part (BWP), a resource block (resource block), a physical resource block (PRB) or a subcarrier, etc. As an example only and not a limitation, for example, in LTE and NR, a frequency point may be a frequency range consisting of 12 subcarriers (the frequency unit is a subcarrier) spaced at 15kHz intervals.

[0078] Typically, information transmission (e.g., Physical Downlink Control Channel (PDCCH), Physical Downlink Shared Channel (PDSCH), Physical Uplink Shared Channel (PUSCH), Physical Uplink Control Channel (PUCCH), etc.) occurs on more than one physical resource (e.g., multiple time points, multiple frequency points, multiple antennas, and various combinations of the above time, frequency, and antenna). In an existing wireless communication system, a method for obtaining channel state information between two nodes is to select a portion of physical resources that can be used for information transmission between the two nodes according to a certain rule to send a reference signal (RS), and estimate the wireless channel state information accordingly. A wireless communication system may simultaneously send non-reference signals, such as data signals and / or control signals, on the remaining available physical resources. A reference signal is a sequence generated based on a specific rule and parameter, and its content (such as the rule and parameter used, what kind of reference signal) and the physical resource where it is sent (i.e., the physical resource used to transmit the reference signal) are shared by two nodes. The reference signal may also be called a pilot signal, a training signal, etc.

[0079] The most widely used wireless communication system at present is a cellular communication system based on the 3rd Generation Partnership Project (3GPP) protocol, such as 4G communication systems such as Long Term Evolution (LTE), evolved LTE (LTE-Advanced, LTE-A), and 5G communication systems such as NR. In LTE-A and NR systems, channel state information can be obtained by sending channel state information reference signals, sounding reference signals, or demodulation reference signals. Specifically, taking the SRS reference signal as an example, before sending the SRS reference signal, the base station first determines the first resource (i.e., the physical resource occupied by the SRS reference signal) and the second resource (i.e., the physical resource on which the channel state information needs to be obtained), and configures the SRS reference signal through control information, and then informs the user equipment / terminal of the configuration information (e.g., the physical resource positions such as the time domain and frequency domain and antenna that can be included in the first resource and the second resource, and the relevant parameters required to generate the reference signal, etc.). Depending on the application scenario, the communication system will determine the mode and density of the first resource according to the changing trend of the channel scenario. For example, in NR, a subband or partial bandwidth in the frequency domain needs to be configured with at least some subcarriers in a resource block (RB) to place the SRS reference signal corresponding to an antenna port, and at least one subcarrier is configured for each antenna port to place the SRS reference signal. If the system is configured with multiple RBs, the SRS reference signals of different antenna ports can also be placed on different RBs. In the time domain, the communication system uses the time slot as the minimum unit, and periodically or non-periodically configures the slot where the SRS reference signal needs to be placed. Figure 4 FIG. 1 is a schematic diagram showing an SRS resource configuration using two antenna ports (i.e., antenna port 1 and antenna port 2) in the 3GPP NR protocol. Figure 4As shown, one RB is composed of 12 subcarriers in the frequency domain, and one slot is composed of 14 OFDM symbols in the time domain. In this schematic diagram, SRS reference signal 1 uses antenna port 1 to occupy the 10th to 13th OFDM symbols in the time domain and the 1st, 3rd, 5th, 7th, 9th, and 11th subcarriers in the frequency domain; SRS reference signal 2 also occupies the 10th to 13th OFDM symbols in the time domain, and occupies the 2nd, 4th, 6th, 8th, 10th, and 12th subcarriers in the frequency domain. After determining the information such as the first resource and the second resource of the SRS reference signal, the terminal can send the reference signal on the first resource, and the base station can obtain the channel state information on the second resource through an interpolation / prediction method (such as the least square (LS) algorithm, the linear minimum mean square error (LMMSE) algorithm) based on the reference signal received on the first resource. In some cases, the base station also needs to transmit the channel state information it obtains to the terminal through the SRS resource indicator (SRI) in the control information, so that it can obtain the channel information necessary for communication. For other reference signals, such as CSI-RS reference signals, the base station and the terminal exchange behaviors and repeat the above steps so that both parties can obtain the required channel state information.

[0080] However, the above method for acquiring channel state information needs to be further improved.

[0081] In existing systems that obtain channel state information through neural networks, various parameters of the neural network are obtained through training based on a large amount of sample data. Communication nodes will incur a lot of additional overhead when obtaining these sample data. For example, in order to obtain real channel state information for training the neural network, the receiving end needs to transmit the complete and real channel state information to the transmitting end through physical resources. In order to avoid additional overhead, the neural networks in the system are obtained through offline training before the system is deployed. After the system is actually deployed, the various parameters of the neural network will no longer change. However, the actual environment will change over time, resulting in a decrease in the accuracy of the channel state information obtained by the system, which in turn affects the reliability of uplink / downlink transmission.

[0082] The present disclosure proposes a method for acquiring channel state information based on a neural network, which can realize online training and updating of neural network parameters, and avoid the decrease in the accuracy of channel state information acquisition due to changes in the external environment. At the same time, the online update process in the method disclosed in the present disclosure does not require measurement of the target channel to obtain channel state information, which effectively reduces the overhead of reference signal transmission and channel state information feedback. The method can be applied to various communication nodes of the network, such as base stations and terminals.

[0083] In the present disclosure, a communication node may be a base station, for example, a macro base station, a micro base station, a pico base station, a femto base station, a backhaul base station for unlimited backhaul, and a base station that integrates both access and backhaul functions, etc., for a wireless access network; a communication node may also be a user device, for example, a mobile phone, a laptop computer, a tablet computer, a smart watch, etc., with a wireless access function.

[0084] Figure 5 is a flowchart of a method performed by a communication node according to an embodiment of the present disclosure.

[0085] Reference Figure 5 In step S501,

[0086] The first neural network is used to obtain channel state information on the second resource.

[0087] In step S502, based on the channel state information, second information related to the channel state information is obtained using a second neural network. For example, the second information can be obtained by a second neural network based on the channel state information and the second related information.

[0088] In step S503, the first neural network is updated using the second information. Figure 5 The method shown can accurately obtain channel state information, thereby solving the problem that wireless channels have complex and changeable characteristics in the time domain, frequency domain, and space domain.

[0089] In some examples, the first neural network may include a preprocessing unit, a pretrained neural network, etc., or the first neural network may consist only of a pretrained neural network. For example, the first neural network first processes the first signal and / or the first related information on the first resource through the preprocessing unit, and then obtains the channel state information on the second resource through the pretrained neural network based on the processing result, such as Fig. 9 shown.

[0090] The first information may include at least one of the following: a first signal on a first resource, the first signal including all signals received on the first resource; a reference signal on the first resource; a second signal after a specific processing of the first signal and / or the reference signal; feedback of channel state information corresponding to the first resource; channel information estimated inversely based on received data; a physical resource containing a reference signal; channel state information measurement report information; first related information, the first related information including at least one of information indicating the position of the reference signal in the first resource, a reference signal, information related to the first resource and / or the second resource (e.g., the relative position of the second resource to the first resource, the distribution structure of the second resource, etc.), and characteristic quantity constraint information of the channel in the current network environment (e.g., variance of the channel, delay spread, correlation time). The above-mentioned first information is only exemplary, and the present disclosure is not limited thereto.

[0091] In some examples, when the first neural network is located on the base station side, the first information may be an SRS signal received by the base station, a DM-RS signal in a PUSCH or PUCCH received by the base station, or a CSI measurement report received by the base station. The first information may also be channel information reversely estimated by the base station through received data. In other examples, when the first neural network is located on the terminal side, the first information may be a CSI-RS signal received by the terminal, a DM-RS signal in a PDSCH or PDCCH received by the terminal, or an SRI indication received by the terminal from the base station. The first information may also be channel information reversely estimated by the terminal through received data.

[0092] In some examples, the first information may be a first signal received on a first resource, or a second signal after a specific processing (e.g., denoising, compression, eigenvalue decomposition, etc.) of the first signal (e.g., a denoised signal obtained by inputting the first signal into a bandpass filter). In some examples, the first information is an eigenvalue decomposition of the first signal after obtaining an intermediate signal through an MMSE algorithm, and then obtaining an eigenvector and an eigenvalue. Among them, the first resource may refer to a physical resource configured for transmitting a reference signal in a communication system, and the second resource may refer to a physical resource that a communication node needs to accurately obtain its channel state information. More specifically, the second resource may include the first resource, the first resource may include the second resource, and the first resource and the second resource may not include each other. For example, when the first neural network is located on the base station side, the second resource is a physical resource indicated by the upper layer for a certain user to transmit data (such as RB, OFDM symbol, antenna port, etc.).

[0093] The resource where the reference signal is located and the resource where the channel state information is located may be different. In this case, the channel state information on the resource without the reference signal may be estimated by the neural network disclosed in the present invention.

[0094] In some examples, when the first neural network is located at the base station side, the first information may be a sounding reference signal received by the base station, or a complete OFDM symbol including a sounding reference signal received by the base station, or channel state information corresponding to the first resource fed back by the terminal to the base station. The channel state information is the comprehensive impact of the entire wireless communication link on the amplitude and phase of the signal on all physical resources (here, the first resource) occupied by its transmission.

[0095] After obtaining the first information, the first information is input into a pre-trained neural network in the first neural network to obtain channel state information on the second resource. In some examples, the second resource may be a physical resource corresponding to the channel state information measured and / or reported by the communication node, or a physical resource where the channel state information to be obtained is located. In some examples, the second resource may include the first resource, the first resource may include the second resource, or the first resource and the second resource may not include each other. In some examples, when the first neural network is located on the base station side, the second resource is a physical resource (RB, OFDM symbol, antenna port, etc.) indicated by the upper layer for a certain user to transmit data.

[0096] Acquire second information through a second neural network based on the obtained channel state information. The second information may be an evaluation of the channel state information. In some examples, the second neural network may include a preprocessing unit, at least one pre-trained neural network or sub-neural network, etc. Alternatively, the second neural network may consist only of pre-trained neural networks or sub-neural networks.

[0097] For example, the second neural network first processes the channel state information through a preprocessing unit, and then sequentially passes the processing result and / or the second related information through each pre-trained neural network to obtain the second information, such as Fig.10 As shown. Here, the processing performed by the preprocessing unit may include feature extraction, feature decomposition, covariance processing, etc. The second relevant information may include preset threshold information, channel high-order statistics information in the target scene, the above-mentioned first information, or at least one of the first information after processing (such as compression, feature extraction). Here, the target scene may refer to the scene in which the communication node is located, that is, the second resource is in the target scene in a physical sense, and the real channel state information is determined by various objects in the target scene. The real channel state information may also be referred to as target channel state information, which is the channel state information obtained by estimating the reference on the resource where the channel state information is located. For example, the real / target channel state information of the second resource can be obtained by estimating the reference signal on the second resource.

[0098] The second information may include at least one of the following: prediction information for data transmission based on channel state information, the prediction information may include at least one of the probability of successful or failed data transmission, the bit error rate of data transmission, the symbol error rate of data transmission, the bit error rate of data transmission, and the signal-to-interference ratio generated by data transmission; error evaluation information of channel state information, the error evaluation information includes at least one of the estimated error of the channel state information, the quality classification label of the channel state information estimate, the deviation measure between the characteristic information of the channel state information and the preset parameter value in the second neural network, and the probability of occurrence of the channel state information in the current network environment.

[0099] In some examples, the second information may include prediction information (such as performance prediction information) for data transmission based on the obtained channel state information. For example, when the channel state information may include CSI-SINR information on the second resource, the second information may be prediction information for data transmission when the modulation and coding method of data transmission (e.g., PUSCH, PDSCH, etc.) is determined based on the CSI-SINR information; or, when the channel state information may include precoding matrix indicator (PMI) information on the second resource, the second information may be prediction information for data transmission when the precoding matrix of data transmission (e.g., PUSCH, PDSCH, etc.) is determined based on the PMI information. In some specific embodiments of this example, the prediction information may be at least one of the following: the probability of successful or failed data transmission (for example, the probability of the receiving node feeding back ACK when the second neural network is located at the sending node); the bit error rate of data transmission (or, equivalently, the symbol error rate, the bit error rate) (for example, when the second neural network is located at the receiving node, the ratio of the number of correct bits obtained after channel equalization and judgment of the received data to the total number of bits); the signal-to-interference ratio generated by data transmission (for example, XdB, which represents the ratio of the received signal power of the target user to the sum of the received power of non-target users in the system after precoding when the second neural network is located at the sending node).

[0100] In some other examples, the second information may also include an error evaluation of the channel state information output by the first neural network. In some specific implementations of this example, the error evaluation included in the second information may be an estimated error between the channel state information output by the first neural network and the actual channel state information. For example, when the channel state information is a channel vector, the output channel state information is recorded as And the real channel state information is Then the estimated error can be or (cosine similarity), where ‖.‖ 2In some specific implementations of this example, the error evaluation included in the second information may be a quality classification label of the channel state information estimation, that is, when the output channel state information is similar to the real channel information, it is marked as good, otherwise it is bad. For example, the output channel state information is recorded as And the real channel state information is Then the expected output label y of the second neural network is:

[0101]

[0102] Among them, α is a fixed threshold value. In some specific implementations of this example, the error evaluation included in the second information may be a deviation measure between the characteristic information of the output channel state information and the preset parameter value in the second neural network. For example, the difference between the high-order statistics such as high-order distance, high-order cumulant, high-order spectrum and the preset parameters of the output channel state information; in some specific implementations of this example, the error evaluation included in the second information may be the probability of occurrence of the output channel state information in the current network environment. For example, according to the probability distribution function preset by the system (such as Gaussian distribution function), under the condition of the first information, the probability of occurrence of the output channel state information is p(h|h pre ), where h is the output channel state information, h pre For the first information.

[0103] Each parameter in the first neural network can be updated based on the second information. For example, the second information is first converted into an error amount, and then the error amount is back-propagated layer by layer according to the back-propagation rule to obtain the error amount corresponding to each layer of the neural network parameter in the first neural network, and finally the optimization algorithm is used to calculate the update amount of the parameter based on the error amount of the parameter and update it. In some examples, the error amount is generated by calculating the distance between the second information and the preset target value, and the distance can be Euclidean distance, L1 norm, cross entropy function, DL divergence, etc.

[0104] In a specific implementation of some examples, when the second information is the probability of successful data transmission (denoted as y), the target value may be 1 or 0.95 (denoted as x), that is, the data transmission is successful with a very high probability. At this time, the error amount between the second information and the preset target value may be obtained based on the Euclidean distance, that is, ‖xy‖ 2. In another specific implementation of some examples, when the second information is a quality classification label of the channel state information estimate (denoted as y), the target value can be 1, that is, the channel state information output by the first neural network is a good estimate of the true channel state information. At this time, the error amount between the second information and the preset target value can be obtained based on the cross entropy function, that is, -logy. In some examples, the optimization algorithm can be a gradient descent algorithm, a stochastic gradient descent algorithm, a small batch gradient descent algorithm, a momentum gradient descent algorithm, a Nesterov gradient acceleration algorithm, an Adagrad method, an AdaDelta method, or an Adam algorithm. Specifically, as an example only and not a limitation, common machine learning training algorithms such as derivative clipping, regularization, model pruning, and model quantization can also be used in the process of updating the weights of network parameters.

[0105] The update of the parameters of the first neural network may be non-continuous. In some embodiments of this example, the update of the parameters of the first neural network may be periodic, for example, the first neural network updates the parameters once every 10 channel state information outputs or once every 1ms, wherein the execution period of the network parameter update may be predetermined according to the coherence time of the scenario and the channel state information estimation accuracy requirements. In other embodiments of this example, the update of the parameters of the first neural network may be based on event triggering. When the first neural network is located on the terminal side, the triggering event may be: the prediction information or error evaluation in the second information exceeds a preset threshold; the terminal fails to decode the received data; the terminal receives a retransmission instruction or parameter update instruction scheduled by the base station. When the first neural network is located on the base station side, the triggering event may be: the prediction information or error evaluation in the second information exceeds a preset threshold; the base station fails to decode the received data; the base station receives NACK information fed back by the terminal; certain events generated by the base station itself.

[0106] In addition, the first neural network and the second neural network can be implemented jointly and / or independently by the same physical entity or different physical units. The physical entity can be composed of hardware, software or a combination thereof. Those skilled in the art can configure it according to actual needs. The combination methods include but are not limited to: the first neural network and the second neural network are respectively implemented by independent entities; the first neural network and the second neural network are implemented by the same entity.

[0107] In order to obtain channel state information, the communication system needs to use part of the physical resources to transmit reference signals, resulting in a reduction in the physical resources used to transmit user data, which affects the throughput of the communication system. Figure 4The SRS occupies about 30% of the physical resources in a time slot, that is, the throughput of the communication system is reduced by about 30% compared to the theoretical upper limit of all physical resources used to transmit user data. The present disclosure can update the first neural network used to estimate the channel state information online, so that the sending node and the receiving node need to obtain the channel state information that meets the accuracy requirements, thereby ensuring the reliability of uplink / downlink transmission (for example, meeting the bit error rate requirements, etc.).

[0108] Figure 6 is a flowchart of a method performed by a communication node according to another embodiment of the present disclosure.

[0109] Reference Figure 6 In step S601, a first signal on a first resource is received, first information is obtained, and channel state information on a second resource is obtained through a first neural network based on the first information. The specific process can refer to the description of steps S501 and S502 above.

[0110] In step S602, second information is obtained through a second neural network based on the channel state information. The second information is an evaluation of the channel state information. The specific process can refer to the description of step S503 above.

[0111] In step S603, based on the second information and / or the third information related to the data transmission, the third neural network is used to obtain fourth information for assisting the data transmission. The third information is related information for the data transmission, and the fourth information is auxiliary information for the data transmission. In some examples, the third neural network may include a preprocessing unit, at least one pretrained neural network or sub-neural network, etc. Alternatively, the second neural network may be composed of only pretrained neural networks or sub-neural networks. For example, the third neural network first processes the second information and / or the third information through the preprocessing unit, and then sequentially passes it through each pretrained neural network structure to obtain the fourth information, such as Fig.11 shown.

[0112] The third information may include at least one of the following: data transmission related parameters, including: modulation mode, coding rate, available physical resources, amount of data to be transmitted in the buffer, quality of service requirements, at least one of the precoding codebook set; network environment related information, including: environmental interference level, environmental change degree, and at least one of the empirical data of the relevant time of the environment. The relevant parameters here may be parameters of current data transmission, parameters of historical data transmission, or predicted parameters.

[0113] The fourth information may include at least one of the following: data transmission status indication information, used to indicate whether the channel state information is used for data transmission and / or whether to redo the complete channel estimation process and / or whether to update the first neural network based on the second information; recommended information of data transmission parameters, used to indicate the data transmission parameters that the communication node should use. For example, the data transmission status indication information may include channel estimation indication information, network parameter update indication information, and data transmission quality indication information. The channel estimation indication information can be used to indicate whether to redo the complete channel estimation process (such as managing CSI non-periodic reporting), the network parameter update indication information can be used to indicate whether to update the first neural network based on the second information, and the data transmission quality indication information can be used to indicate whether the channel state information is used for data transmission (or the degree of recommendation for using the channel state information for data output). The data transmission quality indication information can be used to assist the base station in data transmission scheduling. The forms and names of the above information are only exemplary, and the present disclosure is not limited thereto.

[0114] According to an embodiment, when the communication node is a user equipment, the fourth information is used by the user equipment for at least one of the following: determining whether to trigger re-performing a complete channel estimation process; reporting the fourth information to a base station as reference information for scheduling data transmission by the base station; wherein whether to trigger re-performing a complete channel estimation process includes at least one of whether to trigger a non-periodic channel state information measurement report and sending information to the base station regarding whether the user equipment should re-perform a complete channel estimation process.

[0115] According to an embodiment, when the communication node is a base station, the fourth information is used by the base station for at least one of the following: determining whether to trigger re-performing a complete channel estimation process; determining whether to use the channel state information for data transmission; determining whether to configure parameters used for data transmission; wherein determining whether to use the channel state information for data transmission includes determining resource allocation between user devices for data transmission; wherein whether to trigger re-performing a complete channel estimation process includes at least one of: whether to trigger non-periodic channel state information measurement reporting, and whether to schedule the user equipment to send a reference signal.

[0116] For example, the base station may trigger the re-execution of a complete channel estimation process (for example, the base station instructs the terminal to resend the SRS (i.e., uplink channel estimation), and the base station notifies the terminal to perform non-periodic CSI measurement reporting (i.e., downlink channel estimation)). Alternatively, the base station may configure the parameters used for data transmission according to the recommended parameters in the fourth information. Alternatively, the base station may determine the resource allocation between user devices for data transmission based on the data transmission quality indication information (for example, user devices with good quality share resources, and user devices with poor quality will not be scheduled by the base station for data transmission). Alternatively, the base station may decide whether to trigger non-periodic CSI measurement reporting or schedule the user to send an SRS signal based on the channel estimation indication information. Alternatively, the base station may decide to configure the parameters (modulation order, coding rate, etc.) used by each user device scheduled for data transmission based on the recommended information on the data transmission parameters.

[0117] In some examples, the third information may include relevant parameters of the current data transmission, such as the modulation mode and coding rate of the data transmission, the available physical resources, the amount of data to be transmitted in the buffer area, and the QoS requirements of the data transmission. In some embodiments of this example, the relevant parameters of the data transmission may be historical, that is, the relevant parameters used for the data transmission from a certain moment in the past to the current time, and the third neural network may obtain empirical information of the data transmission parameter configuration from it. In some embodiments of this example, the relevant parameters of the data transmission may be current, and the third neural network may evaluate the rationality of the current data transmission parameters. In some embodiments of this example, the relevant parameters of the data transmission may be predicted, and the third neural network may evaluate whether the expected data transmission parameters are feasible and give instructions in the fourth information to make the communication node prepare accordingly. In other examples, the third information may include relevant information of the current environment or relevant information of the environment, such as the interference level of the current environment, whether the current environment has a sudden change, and empirical data of the relevant time of the environment.

[0118] In some examples, the fourth information may be data transmission status indication information, which indicates the behavior of the communication node where the third neural network is located, including whether the channel state information output by the first neural network should be used for data transmission, whether the complete channel estimation process should be repeated, and whether the first neural network should be updated based on the second information. In some specific implementations of this example, the fourth information includes the degree of recommendation for using the channel state information output by the first neural network for data transmission. For example, when the second information indicates that the cosine similarity between the channel state information and the true channel state information is 0.8, the degree of recommendation given by the fourth information may be 0.75. In the case where the second information indicates that the cosine similarity between the channel state information and the true channel state information is 0.7, when the third information indicates that the data transmission uses a higher-order modulation method (such as 256QAM), the degree of recommendation given by the fourth information may be 0.5, and when the third information indicates that the data transmission uses a lower-order modulation method (such as QPSK), the degree of recommendation given by the fourth information may be 0.75. In other specific implementations of this example, the fourth information may be channel estimation indication information, which indicates that the communication node should repeat the complete channel estimation process. For example, when the second information indicates that the channel state information contains a large estimation error (cosine similarity is less than 0.5), or when the third information indicates that the environment in which the communication node is located has changed dramatically, the fourth information indicates that the communication node should resend the reference signal according to the channel estimation process in the existing protocol to perform a complete channel estimation. In some other specific implementations of this example, the fourth information may be indication information for updating network parameters, which indicates that the first neural network should perform parameter updates. For example, when the second information indicates that the channel estimation quality label of the channel state information output by the first neural network is poor, the fourth information instructs the present structure to execute step S604 to update the first neural network.

[0119] In other examples, the fourth information may also be recommendation information for data transmission parameters. In some specific implementations of this example, the above-mentioned recommendation information may be a combination of a modulation mode and a coding rate obtained based on the above-mentioned channel state information and the amount of data to be transmitted in the buffer. For example, when the quality evaluation of the channel state information in the second information is good and the third information indicates that the amount of data to be transmitted in the buffer is large, the recommendation information contained in the fourth information may be a combination of a higher modulation mode and a coding rate that the system can adopt. In other specific implementations of this example, the above-mentioned recommendation information may be a recommendation to use or abandon certain physical resources for data transmission and / or channel estimation, that is, to give a pattern of physical resources occupied by non-uniformly distributed reference signals. For example, based on the error information of the channel state in the second information, it can be concluded that the estimation error of certain physical resources in the second resource is very low, then these physical resources may no longer be used to transmit reference signals or transmit important information (such as DCI). On the contrary, if the second information indicates that the channel estimation error of some physical resources is very high, then reference signal transmission should be increased near these physical resources. For example, assuming that there are 10 second resources, numbered 0-9, and data transmission is currently performed on resources 0 and 5, if the third information indicates that the channel estimation error near resources 7, 8, and 9 is very high, then the reference signal is transmitted on resources 0, 5, and 8 instead. For another example, in the data transmission of the PDSCH channel of the NR system, the second resource contains 10 OFDM symbols, numbered 0-9. The third information indicates that the current data transmission on the PDSCH channel is performed on resources 3-8, and reference signals are transmitted on resources 2 and 9 (i.e., the pos1 format of mapping type A in the NR standard). If the second information indicates that the channel estimation error near resources 5, 6, and 7 is very high, the fourth information recommends that the reference signal be transmitted on resources 2, 6, and 9, and data transmission is performed on resources 3, 4, 5, 7, and 8 (i.e., the pos2 format of mappingtype A in the NR standard). In other specific implementations of this example, the above-mentioned recommendation information may be a scheduling suggestion for a data transmission user. For example, when the second information indicates that the estimation error of the channel state information is low, the base station may increase the number of MU-MIMO users.

[0120] The configuration of the physical resources occupied by the reference signal in the existing system is not intelligent enough, and it is difficult to make timely and flexible adjustments to the dynamic changes in the physical environment, thereby achieving a dynamic balance between the reliability of uplink / downlink transmission and the system throughput. For example, when a large vehicle enters the scene, the system should promptly increase the physical resources occupied by the reference signal to ensure transmission performance, and when the large vehicle leaves the scene, the system should promptly reduce the physical resources occupied by the reference signal to obtain better throughput, which is difficult to do in current communication systems. The present disclosure uses a third neural network to obtain fourth information for assisting data transmission, and makes timely and flexible adjustments to the dynamic changes in the physical environment, thereby achieving a dynamic balance between the reliability of uplink / downlink transmission and the system throughput.

[0121] In some examples, when the third neural network is located on the base station side, the third information may be a scheduling decision made by the base station itself, a measurement function of the base station itself, a scheduling request reported by the terminal received by the base station, or a service quality constraint received by the base station from an upper layer protocol or a core network. In some implementations of this example, the channel estimation indication information in the fourth information may be used to manage non-periodic CSI measurement reporting. For example, when the channel estimation indication information indicates that the channel estimation process needs to be repeated, the base station sends DCI information to trigger the CSI measurement reporting process of the terminal. In some implementations of this example, the data transmission quality indication information in the fourth information may be used to assist the base station in data transmission scheduling. For example, when the data transmission quality indication information indicates that the data transmission recommendation level of user 1 is high, while the data transmission recommendation level of user 2 is low, the base station may schedule user 1 for data transmission only through DCI information, and wait until the data transmission recommendation level of user 2 returns to high before simultaneously scheduling user 1 and user 2 for data transmission. In some other implementations of this example, the recommended information of the data transmission parameters in the fourth information can be used as a reference for the base station to configure user data transmission. For example, when the recommended information of the data transmission parameters indicates that the modulation mode of data transmission recommends using 64QAM modulation, the base station can indicate in the DCI that the terminal uses 64QAM modulation for subsequent transmission. Here, the complete channel estimation process can be an existing channel estimation process.

[0122] In other examples, when the third neural network is located on the terminal side, the third information may be the DCI information made by the terminal receiving the base station, the terminal's own environmental perception function, and the service quality constraints of the terminal's own upper layer protocol. In some implementations of this example, the channel estimation indication information in the fourth information may be used to trigger the terminal to re-perform the complete channel estimation process. For example, the terminal may receive the CSI-RS signal located in the second resource for channel estimation to obtain the true channel state information and correct / update the first neural network; the terminal may also report the event (i.e., the channel estimation indication information indicates that the complete channel estimation process needs to be re-performed) to the base station and process it according to the instructions of the base station. In other implementations of this example, the recommended information of the data transmission parameters in the fourth information may be reported to the base station, thereby improving the accuracy of base station scheduling.

[0123] In step S604, each parameter in the first neural network is updated based on the second information and / or the fourth information. The specific process can refer to the description of step S504. In some examples, the execution of step S604 can be controlled by the fourth information. In some specific implementations of this example, the fourth information may include whether to directly indicate whether to update the first neural network through the second information. In other specific implementations of this example, whether step 604 is executed or not depends on whether the data transmission status indication information in the fourth information exceeds a preset threshold. For example, when the data transmission status indication information is lower than the first threshold, step 604 is not executed; when the data transmission status indication information is higher than the first threshold but lower than the second threshold, step 604 is executed; when the data transmission status indication information is higher than the second threshold, the complete channel estimation process is repeated. The above examples are merely exemplary, and the present disclosure is not limited thereto.

[0124] according to Figure 6 The method shown can accurately obtain channel state information and effectively use the channel state information to improve transmission performance, and can solve the problem that wireless channels have complex and changeable characteristics in the time domain, frequency domain, and space domain. Figure 7 is a flowchart of a method for online updating of a second neural network according to an embodiment of the present disclosure.

[0125] Reference Figure 7 In step S701, second information is obtained through a second neural network based on the channel state information. The second information is an evaluation of the channel state information. The specific process can refer to the description of step S503.

[0126] In some examples, the channel state information on the second resource is currently transmitted through Figure 5 or Figure 6The method is obtained from the output of the first neural network. In other examples. The channel state information on the second resource may be channel state information obtained by the communication node performing a traditional channel estimation process (e.g., a CSI measurement reporting process completed by a CSI-RS). In other examples, the channel state information on the second resource may be channel state information used for data transmission in the past.

[0127] In step S702, data transmission is performed based on the channel state information on the second resource. In some examples, after obtaining the channel state information, the sending node calculates the modulation order, coding rate, precoding matrix and other parameters used for data transmission according to the channel state information, and then performs channel coding, modulation, precoding and other processes in sequence before sending the data to the receiving node.

[0128] In step S703, fifth information related to data transmission based on channel state information is obtained, and at least part of the parameters of the second neural network are updated based on the second information and / or the fifth information. The fifth information may be result information generated by data transmission based on the channel state information on the second resource. For example, when updating the parameters of the second neural network, the second information and the fifth information are first converted into error amounts, and then the error amounts are back-propagated layer by layer according to the back-propagation rule to obtain the error amounts corresponding to the parameters of each layer of the neural network of the second neural network, and finally the update amount of the parameter is calculated based on the error amount of the parameter using the optimization algorithm and updated.

[0129] In some examples, when the second neural network is located at the sending node, the fifth information may be feedback information (ACK or NACK) of data transmission. For example, in an NR system including a second neural network, the base station obtains the channel state information on the physical resources of the PDSCH through the first neural network based on the CSI measurement information fed back by the terminal, and performs precoding operation on the data based on the channel state information before sending it. After receiving the data, the terminal sends ACK feedback information through PUCCH to indicate that the data transmission is successful. At this time, the fifth information is ACK feedback information. In other examples, the fifth information may be the cumulative bit error rate (which can be equivalently replaced by symbol error rate, success rate, block success rate, etc.) of data transmission based on the channel state information obtained by the first neural network since the last channel estimation process. For example, starting from the channel estimation at time t0, the sending node continuously sends data at times t1, ..., tn, and the receiving node continuously receives data based on the channel state information generated by the first neural network. Then the fifth information is the proportion of the number of correctly received bits to the total number of bits.

[0130] In some examples, the amount of error can be generated by calculating the distance between the second information and the fifth information, and the distance can be Euclidean distance, L1 norm, cross entropy function, DL divergence, etc. In some specific implementations of this example, the second information is a quality classification label of the channel state information estimation (denoted as y, when y=1, it means that the quality of the channel state information output by the first neural network is good), and the fifth information is feedback information of data transmission (denoted as x, when x=1, it means that the data transmission is successful). At this time, the amount of error between the second information and the fifth information can be obtained by the cross entropy function, that is, -xlogy-(1-x)log(1-y). In some examples, the optimization algorithm can be a gradient descent algorithm, a stochastic gradient descent algorithm, a small batch gradient descent algorithm, a momentum gradient descent algorithm, a Nesterov gradient acceleration algorithm, an Adagrad method, an AdaDelta method, and an Adam algorithm. Specifically, as an example only and not a limitation, common machine learning training algorithms such as derivative clipping, regularization, model pruning, and model quantization can also be selected in the process of updating the weights of network parameters.

[0131] In some examples, when the structure of the second neural network is as shown in Figure 10 below, that is, when the second information output by the second neural network is composed of multiple sub-information (such as the probability of successful data transmission, error evaluation information, etc.), multiple sub-information in one or more of the second information can be selected and combined with the fifth information to obtain the error amount, and the neural network parameters corresponding to the error amount are updated. For example, the second information is composed of two parts: prediction information of the probability of successful data transmission and the estimated error between the output channel state information and the actual channel state information. When the fifth information is feedback information for data transmission, the probability of successful data transmission in the second information can be selected and the fifth information can be used to calculate the error amount. In this case, only the parameters in the first sub-neural network and the second sub-neural network in the second neural network will be updated.

[0132] Figure 8 4 is a flowchart of a method for online updating of a third neural network according to an embodiment of the present disclosure.

[0133] Reference Figure 8 In step S801, second information is obtained through a second neural network based on the channel state information. The second information is an evaluation of the channel state information. The specific process can refer to the description of step S503.

[0134] In step S802, based on the second information and / or the third information, the fourth information is obtained through the third neural network. The third information is related information for data transmission, and the fourth information is auxiliary information for data transmission. The specific process can refer to the description of step S603.

[0135] In step S803, data transmission is performed based on the channel state information on the second resource and / or the fourth information and / or the third information. In some examples, after the sending node obtains the channel state information, a part of the modulation order, coding rate, precoding matrix and other parameters used for data transmission is obtained through the fourth information, and the remaining part is obtained through calculation of the channel state information, and then the channel coding, modulation, precoding and other processes are performed in sequence before the data is sent to the receiving end. In other examples, after the sending node obtains the channel state information, a part of the modulation order, coding rate, precoding matrix and other parameters used for data transmission is obtained through the fourth information, and the remaining part is obtained through the third information, and then the channel coding, modulation, precoding and other processes are performed in sequence before the data is sent to the receiving end. The above process of determining parameters is only exemplary, and the present disclosure is not limited to this.

[0136] For example, channel state information is used during data transmission, and the parameters used for data transmission (such as modulation mode, coding rate, etc.) can be determined by the third information or the fourth information. For example, in multi-user downlink transmission, the fourth information modifies the modulation mode and coding rate used by each user, while the third information determines which users transmit simultaneously.

[0137] In step S803, the fifth information is obtained, and at least part of the parameters of the third neural network are updated based on the fourth information and / or the fifth information. The fifth information is the result information generated by data transmission based on the channel state information on the second resource and / or the third information and / or the fourth information. For example, the fifth information is the result information generated by data transmission based on the channel state information on the second resource and the data transmission parameters determined by the third signal and / or the fourth information. When updating the parameters of the third neural network, the fourth information and the fifth information are first converted into error amounts, and then the error amounts are back-propagated layer by layer according to the back-propagation rule to obtain the error amounts corresponding to the parameters of each layer of the neural network of the third neural network, and finally the update amount of the parameter is calculated based on the error amount of the parameter using the optimization algorithm and updated.

[0138] In some examples, when the third neural network is located at the sending node, the fifth information may be feedback information (ACK or NACK) of data transmission. For example, in the NR system, the base station obtains the channel state information on the physical resources of the PDSCH through the first neural network based on the CSI measurement information fed back by the terminal, and precodes the data based on the channel state information and sends it. After receiving the data, the terminal sends ACK feedback information through PUCCH to indicate that the data transmission is successful, and the fifth information is ACK feedback information. In other examples, the fifth information may be the cumulative bit error rate (which can be equivalently replaced by symbol error rate, success rate, block success rate, etc.) of data transmission based on the channel state information obtained by the first neural network since the last channel estimation process. For example, starting from the channel estimation at time t0, the sending node continuously sends data at times t1, ..., tn, and the receiving node continuously receives data based on the channel state information generated by the first neural network, then the fifth information is the proportion of the number of correctly received bits to the total number of bits.

[0139] In some examples, the error amount can be generated by calculating the distance between the fourth information and the fifth information, and the distance can be Euclidean distance, L1 norm, cross entropy function, DL divergence, etc. In some specific implementations of this example, the fourth information is channel estimation indication information (denoted as y, when y=0, it means that the channel estimation process needs to be repeated), and the fifth information is feedback information of data transmission (denoted as x, when x=1, it means that the data transmission is successful). At this time, the error amount between the fourth information and the fifth information can be obtained by the cross entropy function, that is, -xlogy-(1-x)log(1-y). In some examples, the optimization algorithm can be a gradient descent algorithm, a stochastic gradient descent algorithm, a small batch gradient descent algorithm, a momentum gradient descent algorithm, a Nesterov gradient acceleration algorithm, an Adagrad method, an AdaDelta method, and an Adam algorithm. Specifically, as an example and not a limitation, common machine learning training algorithms such as derivative clipping, regularization, model pruning, and model quantization can also be selected in the process of updating the weights of network parameters.

[0140] In some examples, when the structure of the third neural network is as shown in Figure 11 below, that is, when the fourth information output by the third neural network is composed of multiple sub-information (such as channel estimation indication information and modulation mode recommendation information), multiple sub-information in one or more of the fourth information can be selected in combination with the fifth information to obtain the error amount, and the neural network parameters corresponding to the error amount can be updated. For example, the fourth information is composed of two parts: channel estimation indication information and modulation mode recommendation information. When the fifth information is feedback information for data transmission, the channel estimation indication information in the fourth information and the fifth information can be selected to calculate the error amount. In this case, only the parameters in the first sub-neural network and the second sub-neural network in the third neural network will be updated.

[0141] Fig. 9 A schematic diagram of the structure of a first neural network according to an embodiment of the present disclosure is shown.

[0142] Reference Fig. 9 , the first neural network is composed of a preprocessing unit and a neural network (also referred to as a neural network unit) connected in series, wherein the preprocessing unit can obtain the first information based on the first signal and / or related information (i.e., the first related information) on the first resource, and obtain the channel state information on the second resource through a pre-trained neural network based on the first information and / or related information. For example, the preprocessing unit obtains the first information based on the first signal on the first resource and the reference signal contained in the related information, and then inputs the first information into the subsequent neural network. In some examples, the preprocessing unit may first input the first signal into a bandpass filter to obtain a denoised signal as part of the first information, and then obtain an intermediate signal through an MMSE algorithm based on the reference signal in the related information, and finally perform eigenvalue decomposition on the intermediate signal to obtain an eigenvector matrix and eigenvalues ​​as another part of the first information and input them into the neural network.

[0143] When the communication node is a base station, the relevant information including the reference signal can be determined by the base station itself, and when the communication node is a user equipment, the relevant information including the reference signal can be configured by the base station. In addition, the pre-processing unit can also directly input its input (i.e., the reference signal and the first signal on the first resource) to the neural network unit without additional processing. The neural network unit can be a pre-trained neural network, and the specific structure is shown below Fig.12 , which can generate channel state information on the second resource through the first information generated by the pre-processing unit and the additional input related information. Fig. 9 The structure of the first neural network shown is merely exemplary, and the present disclosure is not limited thereto.

[0144] Fig.10 A schematic diagram of the structure of a second neural network according to an embodiment of the present disclosure is shown.

[0145] Reference Fig.10 , the second neural network can be composed of a preprocessing unit and three sub-neural networks (also called neural network units). Among them, the preprocessing unit can extract feature information (for example, variance of channel coefficients, delay spread, inter-antenna correlation matrix, etc.) from the input channel state information and input it to the first sub-neural network, or it can directly input the input channel state information to the first sub-neural network without additional processing. The first sub-neural network, the second sub-neural network, and the third sub-neural network can use the same or different neural network structures, and their various parameters can be first obtained through offline training. After the second neural network is deployed, through Figure 7The method shown is updated online. The first sub-neural network can process feature information and related information (i.e., second related information, such as a pre-written threshold value or a high-order channel statistic in a target scenario), and the information generated will be generated by the subsequent second sub-neural network and the third sub-neural network respectively to generate the probability of successful data transmission and the information state information estimation error in the second information. In some examples, the first sub-neural network can be formed using a transformer structure. In some examples, the second sub-neural network can be formed using an MLP-MIX network. In some examples, the third neural network can use a DNN network. In some examples, the first sub-neural network, the second sub-neural network, and the third sub-neural network can be jointly and / or independently implemented by the same physical entity or different physical units, and the physical entity can be composed of hardware, software, or a combination thereof. Those skilled in the art can configure it according to actual needs, and its combination includes but is not limited to: the first sub-neural network, the second sub-neural network, and the third sub-neural network are respectively implemented by independent entities; the first sub-neural network, the second sub-neural network, and the third sub-neural network are implemented by the same entity; the first sub-neural network and the third sub-neural network are implemented by the same entity, and the second sub-neural network is implemented by another entity; the first sub-neural network and the second sub-neural network are implemented by the same entity, and the third sub-neural network is implemented by another entity.

[0146] The number of sub-neural networks following the first sub-neural network in the second neural network can be determined based on the number of sub-information included in the second information. Multiple sub-neural networks can be used to obtain the corresponding sub-information in the second information. Fig.10 The structure of the second neural network shown is merely exemplary, and the present disclosure is not limited thereto.

[0147] Fig.11 A schematic diagram of the structure of a third neural network according to an embodiment of the present disclosure is shown.

[0148] Reference Fig.11 , the third neural network can be composed of a preprocessing unit and three sub-neural networks. Among them, the preprocessing unit can convert the third information into at least one of a vector, a matrix, and a tensor for neural network layer calculation, and then input it into the first sub-neural network, or it can directly input the third information into the first sub-neural network without additional processing. In some examples, the preprocessing unit predefines the vectors corresponding to each modulation mode and coding rate combination, and converts the modulation mode and coding rate combination in the third information into the corresponding vector by table lookup or calculation. The first sub-neural network, the second sub-neural network, and the third sub-neural network can adopt the same or different neural network structures, and their various parameters can be first obtained through offline training. After the third neural network is deployed, through Figure 8The method shown is updated online. The first sub-neural network processes the second information and the converted third information, and the information generated by it will generate the probability of successful data transmission and the information state information estimation error in the fourth information through the subsequent second sub-neural network and the third sub-neural network. In some examples, the first sub-neural network can use an MLP-MIX network. In some examples, the second sub-neural network can be a two-classification network constructed by a 3-layer DNN. In some examples, the third sub-neural network can use a multi-classification network constructed by a multi-layer CNN, and each label corresponds to a modulation mode and coding rate combination. In some examples, the first sub-neural network, the second sub-neural network, and the third sub-neural network can be jointly and / or independently implemented by the same physical entity or different physical units. The physical entity can be composed of hardware, software, or a combination thereof. Those skilled in the art can configure it according to actual needs, and its combination includes but is not limited to: the first sub-neural network, the second sub-neural network, and the third sub-neural network are respectively implemented by independent entities; the first sub-neural network, the second sub-neural network, and the third sub-neural network are implemented by the same entity; the first sub-neural network and the third sub-neural network are implemented by the same entity, and the second sub-neural network is implemented by another entity; the first sub-neural network and the second sub-neural network are implemented by the same entity, and the third sub-neural network is implemented by another entity.

[0149] The number of sub-neural networks after the first sub-neural network in the third neural network can be determined based on the number of sub-information included in the fourth information. Multiple sub-neural networks can be used to obtain the corresponding sub-information in the fourth information. Fig.11 The structure of the third neural network shown is merely exemplary, and the present disclosure is not limited thereto.

[0150] Fig.12 A schematic diagram of the structure of a neural network unit according to an embodiment of the present disclosure is shown.

[0151] Fig.12 An example is given of the structure of the neural network units or sub-neural networks appearing in the first neural network, the second neural network, and the third neural network of the present disclosure.

[0152] Reference Fig.12 , a neural network unit may include an information processing unit and multiple neural network layers. The information processing unit can process the input data (such as Fig.12 The relevant information and input information in the neural network are converted into at least one of a data vector, a matrix, and a tensor that can be used by the neural network to calculate. The neural network layer can be Fig.12At least one of the input layer, hidden layer / intermediate layer, and output layer shown in . In addition, each neural network layer can be composed of multiple neurons n, and at least one activation function can be used to activate the neurons, for example, tanh function, ReLU function, eLU function, seLU function, ceLU function, preLU function, geLU function, LeakyReLU function, Sigmoid function, Softmax function, Softplus function, etc. Each neural network layer can operate on its input data to achieve functions such as matrix transformation, data dimensionality reduction, data feature extraction, data feature combination, etc. As an example and not a limitation, multiple neural network layers can be connected in series or in parallel to form different neural network structures, including but not limited to multi-layer perceptron (MLP), multi-layer perceptron mixer (MLP-mixer), convolutional neural network (CNN), deep neural network (DNN), recursive neural network (RNN), restricted Boltzmann machine (RBM), deep belief network (DBN), bidirectional recursive deep neural network (BRDNN), generative adversarial network (GAN), transformation network (Transformer), etc.

[0153] The information processing unit can process the relevant information (such as the first relevant information or the second relevant information mentioned above) (for example, the merging, position encoding, etc. described below) so that the neural network will obtain the prior information implicit in the relevant information (for example, the relative position of the second resource and the first resource, the distribution structure of the second resource, etc.) to assist the neural network in improving the accuracy and flexibility of obtaining channel state information. Here, as an example and not a limitation, the relevant information may be at least one of the position of the first resource and the position of the second resource. The relevant information may also be a certain characteristic quantity constraint of the channel in the current scenario, such as the variance, delay spread, correlation time, etc. of the channel. The information processing unit may also convert the relevant information into at least one of a vector, a matrix, and a tensor for use in neural network layer calculations.

[0154] In some examples, the information processing unit may convert the input information (such as the first information, the channel state information output by the first neural network, the second information, or the third information) into at least one of a vector, a matrix, and a tensor for neural network layer calculation. As an example only and not a limitation, the set of known channel state information is defined as C RS First, the known channel state information is processed by a numerical mapping method (e.g., normalization, standardization, etc.) to make its numerical scale or distribution suitable for the calculation of the neural network. More specifically, taking the minimum and maximum (Min-Max) standardization as an example, C RS The maximum value of all elements in max(C RS ) and the minimum value min(CRS ), then C RS Each element in Perform the following processing After the above numerical mapping is performed, the processed elements can be The collection composed of Convert to at least one of a vector, a matrix, and a tensor corresponding to the input size of the neural network layer.

[0155] In another example, the information processing unit may further position encode the input information, and convert the position-encoded information into at least one of a vector, a matrix, and a tensor for neural network calculation. Optionally, the position encoding may be a process of adding and / or multiplying the input information with a codeword in a predetermined codebook. As an example only and not a limitation, taking the output of known channel state information by the preprocessing unit (i.e., the preprocessing unit outputs the channel state information on the first resource according to the input first signal) as an example, position encoding using a codebook may include the following processing: First, a set C consisting of known channel state information is mapped according to a numerical mapping method. RS Processing to obtain a collection It contains several elements with numerical mappings. Then, the position-encoded codebook is obtained according to the following equation.

[0156]

[0157] Where n represents the physical resource (such as time domain, frequency domain and spatial domain) where the channel state information is located, d represents the physical resource dimension of the channel state information, and N is greater than the set is any positive number containing the number of elements, and i represents the index of the physical resource dimension. Then, through the equation [f(n) (0) ,f(n) (1) ,…,f(n) (d) ] for known channel state information elements Perform position encoding to obtain the set Finally, the processed elements The collection composed of The information processing unit may also directly input the input to the neural network unit without any additional processing.

[0158] The operations in the above-mentioned neural network unit are merely exemplary and the present disclosure is not limited thereto.

[0159] Fig.13 A block diagram of a communication node in a communication system provided by an embodiment of the present disclosure is shown. Fig.13 , the communication node 1300 may include a transceiver 1310 and a processor 1320, wherein the processor 1320 is coupled to the transceiver 1310 and is configured to perform the method performed by the communication node described above. Although the transceiver 1310 and the processor 1320 are shown as separate entities, they can be implemented as a single entity, such as a single chip. The transceiver 1310 and the processor 1320 can be electrically connected or coupled to each other. The transceiver 1310 can send signals to other network nodes and receive signals from other network nodes, such as UE, base station, system simulator (e.g., simulated base station) or core network node. The processor 1320 may include one or more processing units and may control the communication node to perform operations and / or functions according to at least one of the above embodiments.

[0160] For details on the operation of the method performed by the above communication node, please refer to the above Figures 5 to 12 The description will not be repeated here.

[0161] According to an embodiment of the present disclosure, an electronic device may also be provided, comprising: at least one processor; and at least one memory storing computer executable instructions, wherein the computer executable instructions, when executed by the at least one processor, cause the at least one processor to execute any one of the above methods.

[0162] According to an embodiment of the present disclosure, a computer-readable storage medium storing instructions may also be provided, wherein when the instructions are executed by at least one processor, the at least one processor is prompted to execute any one of the above methods according to the exemplary embodiments of the present disclosure. Examples of computer-readable storage media here include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), card storage (such as, multimedia card, secure digital (SD) card or extreme digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk and any other device, any other device is configured to store computer programs and any associated data, data files and data structures in a non-transitory manner and provide the computer programs and any associated data, data files and data structures to a processor or computer so that the processor or computer can execute the computer program. The instructions or computer programs in the above-mentioned computer-readable storage medium can be run in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc. In addition, in one example, the computer program and any associated data, data files and data structures are distributed on a networked computer system, so that the computer program and any associated data, data files and data structures are stored, accessed and executed in a distributed manner by one or more processors or computers.

[0163] At least one of the above-mentioned multiple modules can be implemented by an AI model. Functions associated with AI can be executed by a non-volatile memory, a volatile memory, and a processor.

[0164] The processor may include one or more processors. In this case, the one or more processors may be general-purpose processors, such as a central processing unit (CPU), an application processor (AP), etc., a processor used only for graphics (such as a graphics processing unit (GPU), a visual processing unit (VPU) and / or an AI-specific processor (such as a neural processing unit (NPU)).

[0165] One or more processors control the processing of input data according to predefined operating rules or artificial intelligence (AI) models stored in non-volatile memory and volatile memory. The predefined operating rules or artificial intelligence models can be provided by training or learning. Here, providing by learning means that by applying a learning algorithm to multiple learning data, a predefined operating rule or AI model with desired characteristics is formed. Learning can be performed in the device itself that executes AI according to an embodiment, and / or can be implemented by a separate server / device / system.

[0166] A learning algorithm is a method of using multiple learning data to train a predetermined target device (e.g., a robot) to enable, allow, or control the target device to make a determination or prediction. Examples of learning algorithms include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0167] The artificial intelligence model can be obtained through training. Here, "obtained through training" means training a basic artificial intelligence model with multiple training data through a training algorithm to obtain a predefined operating rule or artificial intelligence model, which is configured to perform the desired features (or purposes).

[0168] As an example, the artificial intelligence model may include multiple neural network layers. Each of the multiple neural network layers includes multiple weight values, and the neural network calculation is performed by calculating between the calculation result of the previous layer and the multiple weight values. Examples of neural networks include, but are not limited to, convolutional neural networks (CNNs), deep neural networks (DNNs), recurrent neural networks (RNNs), restricted Boltzmann machines (RBMs), deep belief networks (DBNs), bidirectional recurrent deep neural networks (BRDNNs), generative adversarial networks (GANs), and deep Q networks.

[0169] It should be noted that the terms "first", "second", "third", "fourth", "1", "2", etc. (if any) in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than that shown or described in words.

[0170] It should be understood that, although the flowchart of the embodiment of the present disclosure indicates each operation step by arrows, the implementation order of these steps is not limited to the order indicated by the arrows. Unless clearly stated herein, in some implementation scenarios of the embodiment of the present disclosure, the implementation steps in each flowchart can be executed in other orders according to demand. In addition, some or all of the steps in each flowchart may include multiple sub-steps or multiple stages based on the actual implementation scenario. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage in these sub-steps or stages can also be executed at different times. In scenarios with different execution times, the execution order of these sub-steps or stages can be flexibly configured according to demand, and the embodiment of the present disclosure does not limit this.

[0171] The above is only an optional implementation method for some implementation scenarios of the present disclosure. It should be pointed out that for ordinary technicians in this technical field, without departing from the technical concept of the scheme of the present disclosure, other similar implementation methods based on the technical ideas of the present disclosure are also within the protection scope of the embodiments of the present disclosure.

[0172] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method performed by a communication node, comprising: Using the first neural network to obtain channel state information on the second resource; Based on the channel state information, obtaining second information using a second neural network; The first neural network is updated using the second information.

2. The method according to claim 1, characterized in that Obtaining channel state information on a second resource using a first neural network includes: Receiving a first signal on a first resource to obtain first information; Based on the first information, a first neural network is used to obtain channel state information on a second resource.

3. The method according to claim 2, characterized in that The first information includes at least one of the following: the first signal on the first resource, the first signal comprising all signals received on the first resource; a reference signal on the first resource; a second signal obtained by performing specific processing on the first signal and / or the reference signal; Feedback channel state information corresponding to the first resource; Channel information estimated inversely based on received data; a physical resource comprising the reference signal; Channel state information measurement report information; First related information, the first related information includes at least one of information for indicating the position of the reference signal in the first resource, the reference signal, information related to the first resource and / or the second resource, and characteristic quantity constraint information of the channel in the current network environment.

4. The method according to claim 1, characterized in that: The second information includes at least one of the following: Prediction information for data transmission based on the channel state information, the prediction information comprising at least one of a probability of successful or failed data transmission, a bit error rate of data transmission, a symbol error rate of data transmission, a bit error rate of data transmission, and a signal-to-interference ratio generated by data transmission; The error evaluation information of the channel state information includes at least one of an estimated error of the channel state information, a quality classification label of the channel state information estimate, a deviation measure between characteristic information of the channel state information and a preset parameter value in the second neural network, and an occurrence probability of the channel state information in the current network environment.

5. The method according to claim 1, further comprising: Based on the second information and / or third information related to data transmission, fourth information for assisting data transmission is obtained using a third neural network.

6. The method according to claim 5, characterized in that The third information includes at least one of the following: Parameters related to data transmission, the parameters related to data transmission including: modulation mode, coding rate, available physical resources, amount of data to be transmitted in a buffer, quality of service requirements, and at least one of a precoding codebook set; The relevant information of the network environment includes at least one of: environmental interference level, environmental change degree, and experience data of relevant time of the environment.

7. The method according to claim 5, characterized in that The fourth information includes at least one of the following: Data transmission state indication information, used to indicate whether to use the channel state information for data transmission and / or whether to re-perform a complete channel estimation process and / or to indicate whether to update the first neural network based on the second information; The recommended information of data transmission parameters is used to indicate the data transmission parameters recommended for use by the communication node.

8. The method according to claim 7, characterized in that In the case where the communication node is a user equipment, the fourth information is used by the user equipment for at least one of the following: determining whether to trigger a complete re-estimation of the channel estimation process; Reporting the fourth information to the base station as reference information for scheduling data transmission by the base station; Whether to trigger re-performing a complete channel estimation process includes at least one of whether to trigger non-periodic channel state information measurement reporting and sending information to the base station about whether the user equipment re-performs a complete channel estimation process.

9. The method according to claim 7, characterized in that: In the case where the communication node is a base station, the fourth information is used by the base station for at least one of the following: determining whether to trigger a complete re-estimation of the channel estimation process; determining whether to use the channel state information for data transmission; Determining whether to configure parameters used for data transmission; Wherein, determining whether to use the channel state information for data transmission includes determining resource allocation between user equipments for data transmission; Whether to trigger a complete re-estimation of the channel estimation process includes: whether to trigger at least one of aperiodic channel state information measurement and reporting, and whether to schedule the user equipment to send a reference signal.

10. The method according to claim 1, further comprising: obtaining fifth information related to performing data transmission based on the channel state information; At least part of parameters of the second neural network are updated using the second information and / or the fifth information.

11. The method according to claim 4, further comprising: Obtaining fifth information related to performing data transmission based on the channel state information and / or the third information and / or the fourth information; At least part of parameters of the third neural network are updated using the fourth information and / or the fifth information.

12. The method according to claim 10 or 11, characterized in that: The fifth information includes at least one of the following: Feedback information of data transmission; At least one of a cumulative bit error rate, a cumulative symbol error rate, a cumulative success rate, and a cumulative block success rate of data transmission performed based on the channel state information obtained by the first neural network since the last channel estimation process.

13. The method according to any one of claims 1 to 12, characterized in that The first neural network, the second neural network and the third neural network are trained jointly or individually; and / or At least one of the first neural network, the second neural network and the third neural network is updated based on a preset time interval, or is updated based on the number of times information is output, or is updated based on the triggering of a specific event, wherein the specific event includes at least one of the prediction information or error evaluation information in the second information exceeding a preset threshold, the communication node failing to decode received data, receiving a retransmission instruction or a parameter update instruction, and receiving feedback negative acknowledgment NACK information.

14. The method according to claim 1, characterized in that Using the second information, updating the first neural network includes: converting the second information into an error amount; According to the back propagation rule, the error amount is back propagated layer by layer according to each neural network layer included in the first neural network to obtain the error amount corresponding to the parameters in each neural network layer; An optimization algorithm is used to calculate the update amount of the parameter based on the error amount corresponding to the parameter and to update the parameter.

15. The method according to any one of claims 1 to 14, characterized in that The first resource includes the second resource, or the second resource includes the first resource, or the first resource and the second resource include at least part of the same resource, or the first resource and the second resource do not include the same resource.

16. A communication node, comprising: Transceiver; as well as A processor, coupled to the transceiver, and configured to perform the method according to any one of claims 1 to 15.

Images