Ai model parameter interaction method and device

Abstract

An AI model parameter interaction method and device. The method is applied to a WiFi sending end. The method comprises: after AI model parameters are embedded into an autoencoder of a WiFi sending end, sending to a WiFi receiving end a data frame with a special sequence by means of the autoencoder of the WiFi sending end, so that the WiFi receiving end recognizes the data frame on the basis of the special sequence, and decodes the data frame by means of an autoencoder of the WiFi receiving end when successfully recognizing the data frame. The autoencoder improves the coding efficiency and the decoding accuracy of a traditional WiFi transmitter system and a traditional WiFi receiver system, enhancing the WiFi network performance and the user experience.

Description

Interaction method and device for AI model parameters

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This disclosure is based on Chinese patent application CN202310810064.7, filed on June 30, 2023, entitled “Interaction method and device for AI model parameters”, and claims the priority of the patent application, and all the contents disclosed therein are incorporated into this disclosure by reference. Technical Field

[0003] The embodiments of the present disclosure relate to the field of communications, and more specifically, to a method and device for interacting with AI model parameters. Background Art

[0004] Traditional artificial intelligence / machine learning (AI / ML) technologies are based on centralized models, requiring large amounts of data to be exchanged between data sources and centralized servers. Recently, distributed AI / ML algorithms, such as federated learning, have been developed, which allow more analysis to be performed at the source and reduce the amount of data that needs to be exchanged.

[0005] Research shows that AI / ML algorithms can help improve the performance of wireless communication networks by providing better resource utilization, lower energy consumption, higher reliability, and robustness to changing environments. As these algorithms become more mature and cost-effective, wireless local area networks (WLANs) are likely to leverage AI / ML to enhance network performance and user experience.

[0006] However, there is currently no specific implementation of AI / ML for WLAN, and it is impossible to achieve the effect of enhancing network performance and user experience through AI / ML.

[0007] Summary of the Invention

[0008] The embodiments of the present disclosure provide a method and apparatus for interacting with AI model parameters to at least address the problem that related technologies have not yet applied AI / ML to specific WLAN implementations, and thus are unable to achieve the effect of enhancing network performance and user experience through AI / ML.

[0009] According to one embodiment of the present disclosure, a method for interacting with AI model parameters is provided, which is applied to a WIFI transmitter, wherein the WIFI transmitter includes an autoencoder of the WIFI transmitter, and the method includes:

[0010] After the AI ​​model parameters are set in the autoencoder of the WIFI transmitting end, a data frame with a special sequence is sent to the WIFI receiving end through the autoencoder of the WIFI transmitting end, so that the WIFI receiving end recognizes the data frame according to the special sequence, and if the recognition is successful, the data frame is decoded by the autoencoder of the WIFI receiving end.

[0011] In an exemplary embodiment, after the data frame with the special sequence is sent to the WIFI receiving end through the autoencoder of the WIFI transmitting end, the method further includes:

[0012] Monitoring the feedback message from the WIFI receiving end;

[0013] If no feedback message sent by the WIFI receiving end is monitored within the preset time, it is determined that the AI ​​model parameters of the autoencoder of the WIFI sending end need to be updated.

[0014] In an exemplary embodiment, it further includes:

[0015] When the autoencoder of the WIFI transmitting end needs to update the AI ​​model parameters, a trigger frame with a special sequence is sent to the WIFI receiving end within an interframe interval to trigger the autoencoder of the WIFI receiving end to feed back the new AI model parameters.

[0016] In an exemplary embodiment, it further includes:

[0017] Receive new AI model parameters fed back by the autoencoder of the WIFI receiving end, and send a confirmation message to the WIFI receiving end.

[0018] In an exemplary embodiment, the data frame with the special sequence carries feedback channel state information CSI content.

[0019] According to another embodiment of the present disclosure, another AI model parameter interaction method is provided, which is applied to a WIFI receiver, wherein the WIFI receiver includes an autoencoder and a classifier of the WIFI receiver, and the method includes:

[0020] After the AI ​​model parameters are set in the autoencoder of the WIFI transmitter, the classifier receives the data frame with the special sequence sent by the autoencoder of the WIFI transmitter, and identifies the data frame according to the special sequence;

[0021] If the recognition is successful, the data frame is decoded by the autoencoder of the WIFI receiving end.

[0022] In an exemplary embodiment, after decoding the data frame by the autoencoder of the WIFI receiving end, the method further includes:

[0023] When the autoencoder of the WIFI transmitter needs to update the AI ​​model parameters, receiving a trigger frame with a special sequence sent by the WIFI transmitter;

[0024] identifying the trigger frame according to the special sequence by the classifier;

[0025] If the recognition is successful, the autoencoder of the WIFI receiving end is trained to obtain new AI model parameters;

[0026] The new AI model parameters are fed back to the autoencoder of the WIFI sending end.

[0027] In an exemplary embodiment, identifying the trigger frame according to the special sequence by the classifier includes:

[0028] Matching the special sequence of the trigger frame with a locally stored special sequence;

[0029] In case of successful matching, it is determined that the recognition is successful.

[0030] According to another embodiment of the present disclosure, an AI model parameter interaction device is provided, which is applied to a WIFI transmitter, wherein the WIFI transmitter includes an autoencoder of the WIFI transmitter, and the device includes:

[0031] The first sending module is configured to send a data frame with a special sequence to a WIFI receiving end through the autoencoder of the WIFI sending end after the AI ​​model parameters are set in the autoencoder of the WIFI sending end, so that the WIFI receiving end recognizes the data frame according to the special sequence, and decodes the data frame through the autoencoder of the WIFI receiving end if the recognition is successful.

[0032] According to another embodiment of the present disclosure, another AI model parameter interaction device is provided, which is applied to a WIFI receiving end, wherein the WIFI receiving end includes a self-decoder and a classifier of the WIFI receiving end, and the device includes:

[0033] A first receiving module is configured to receive, through the classifier, a data frame with a special sequence sent by the encoder of the WIFI transmitting end after the autoencoder of the WIFI transmitting end is set with the AI ​​model parameters, and identify the data frame according to the special sequence;

[0034] The decoding module is configured to decode the data frame through the autoencoder of the WIFI receiving end when the recognition is successful.

[0035] According to another embodiment of the present disclosure, a computer-readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the steps of any one of the above method embodiments when running.

[0036] According to another embodiment of the present disclosure, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the steps in any one of the above method embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] FIG1 is a hardware structure block diagram of a mobile terminal according to an AI model parameter interaction method according to an embodiment of the present disclosure;

[0038] FIG2 is a network architecture diagram of a WIFI system according to an embodiment of the present disclosure;

[0039] FIG3 is a flow chart of the interaction of AI model parameters according to an embodiment of the present disclosure;

[0040] FIG4 is a schematic diagram of AI parameter interaction within a DIFS frame interval according to an embodiment of the present disclosure;

[0041] FIG5 is a schematic diagram of AI parameter interaction within an EIFS frame interval according to an embodiment of the present disclosure;

[0042] FIG6 is a flowchart of the interaction of AI model parameters according to another embodiment of the present disclosure;

[0043] FIG7 is a structural block diagram of an AI model parameter interaction device according to an embodiment of the present disclosure;

[0044] FIG8 is a structural block diagram of an interactive device for AI model parameters according to another embodiment of the present disclosure. DETAILED DESCRIPTION

[0045] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings and in conjunction with embodiments.

[0046] It should be noted that the terms "first", "second", etc. 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.

[0047] The method embodiments provided in the embodiments of the present disclosure can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 is a hardware structure block diagram of a mobile terminal of an interactive method of AI model parameters in an embodiment of the present disclosure. As shown in Figure 1, the mobile terminal may include one or more (only one is shown in Figure 1) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the above-mentioned mobile terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that the structure shown in Figure 1 is only for illustration and does not limit the structure of the above-mentioned mobile terminal. For example, the mobile terminal may also include more or fewer components than those shown in Figure 1, or have a configuration different from that shown in Figure 1.

[0048] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the interactive method of AI model parameters in the embodiment of the present disclosure. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implementing the above-mentioned method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0049] The transmission device 106 is used to receive or send data via a network. A specific example of the aforementioned network may include a wireless network provided by the mobile terminal's communications provider. In one embodiment, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0050] FIG2 is a network architecture diagram of a WIFI system according to an embodiment of the present disclosure. The embodiment of the present disclosure can run on the network architecture shown in FIG2 . As shown in FIG2 , the network architecture of the WIFI system according to the embodiment of the present disclosure mainly includes a WIFI transmitter and a WIFI receiver. The WIFI transmitter includes an autoencoder (AE) and a traditional WIFI transmitter, and the WIFI receiver includes an autoencoder AE, a classifier, and a traditional WIFI receiver. The functions of each module are as follows:

[0051] The autoencoder can be set at the WIFI transmitter and the WIFI receiver respectively, including an AE encoder and an AE decoder. The AE encoder takes the encoding vector as output, and the AE decoder takes the encoding vector as input. The area between the encoder input and the encoding vector, and the area between the decoder output and the decoding vector are called hidden layers. The structure of the hidden layer can be a convolutional neural network (CNN) structure, a fully connected structure, or a recurrent neural network (RNN) structure.

[0052] Among them, the AE encoder is used to encode the transmitted data, redefine the functional coding modulation with the coding modulation in the traditional WIFI transmitter, or act as a transmitter to compress and encode the transmitted content such as channel estimation data.

[0053] The AE decoder is used to decode the Channel State Information (CSI) content compressed and fed back by the encoder, or to detect and demodulate the transmitted content as a receiver.

[0054] Traditional WIFI transmitter: Traditional WIFI transmitter consists of a coded modulation RF front end, which is used to send encoded data.

[0055] Classifier: The classifier is connected to the autoencoder and traditional WiFi transmitter at the WiFi transmitter, and receives data sent by the WiFi transmitter. It is one of the most important hardware devices in this disclosure. Its main functions are twofold: identifying the trigger frame sent by the AE at the WiFi transmitter and identifying the encoding vector encoded by the AE at the WiFi transmitter.

[0056] The data field of the trigger frame is a special sequence. Identifying the trigger signal matches the special sequence on the trigger frame with a locally stored signature sequence. If the match is successful, identification is confirmed, and AE parameter interaction can continue. The AE-encoded code vector at the Wi-Fi transmitter can be scrambled with a special pseudo-noise (PN) sequence. The classifier can use the locally cached PN sequence for detection and identification. If identification is successful, the AE-encoded code vector at the Wi-Fi transmitter can be sent to the Wi-Fi receiver's AE decoder for decoding. If identification is unsuccessful, the detection process of the traditional Wi-Fi receiver begins.

[0057] Traditional WIFI receiver: used to demodulate and decode received data.

[0058] In addition, it should be noted that the feedback link is not a hardware part, but a frame interaction sequence adapted to this hardware structure. When the autoencoder AE is used for signal modulation and demodulation, the feedback link is used to feed back the parameters of the AE encoder trained by the WiFi receiver to the WiFi transmitter and place them in the AE encoder of the WiFi transmitter.

[0059] In this embodiment, a method for interacting AI model parameters running on the network architecture of the above-mentioned WIFI system is provided. FIG3 is a flow chart of the interaction of AI model parameters according to an embodiment of the present disclosure, which is applied to a WIFI transmitter, wherein the WIFI transmitter includes an autoencoder of the WIFI transmitter. As shown in FIG3 , the process includes the following steps:

[0060] In step S301, after the AI ​​model parameters are set in the autoencoder of the WIFI transmitting end, a data frame with a special sequence is sent to the WIFI receiving end through the autoencoder of the WIFI transmitting end, so that the WIFI receiving end recognizes the data frame according to the special sequence, and decodes the data frame through the autoencoder of the WIFI receiving end if the recognition is successful.

[0061] As an example, when the autoencoder of the WIFI transmitter is set into the AI ​​model parameters, the autoencoder AE of the WIFI transmitter can be used preferentially for frame transmission. When sending, a special field can be added to the data frame for frame identification.

[0062] As an example, the WIFI receiving end may identify the data frame according to the special sequence, and if the identification is successful, decode the data frame through the autoencoder of the WIFI receiving end.

[0063] In an optional embodiment, the data frame with the special sequence carries feedback channel state information CSI content.

[0064] In an optional embodiment, after sending a data frame with a special sequence to a WIFI receiving end through the autoencoder of the WIFI transmitting end, the method further includes: monitoring a feedback message from the WIFI receiving end; if no feedback message sent by the WIFI receiving end is monitored within a preset time, determining that the AI ​​model parameters of the autoencoder of the WIFI transmitting end need to be updated.

[0065] As an example, after the WIFI transmitter completes sending a data frame with a special sequence to the WIFI receiver, it enters the monitoring phase and can monitor the feedback message from the WIFI receiver; if no feedback message sent by the WIFI receiver is monitored within a preset time, it is determined that the AI ​​model parameters of the autoencoder of the WIFI transmitter need to be updated.

[0066] In an optional embodiment, when the autoencoder of the WIFI transmitting end needs to update the AI ​​model parameters, a trigger frame with a special sequence is sent to the WIFI receiving end within an inter-frame interval to trigger the autoencoder of the WIFI receiving end to feed back the new AI model parameters.

[0067] The definition of the interframe space specified in 802.11 is as follows: aSIFSTime = 16μs (OFDM System) Slottime = 9μs aDIFSTime = aSIFSTime + 2*Slottime aEIFSTime = aSIFSTime + ACKTransmissionTIME + aDIFSTime

[0068] SIFS is the short interframe space (SIFS), which is equal to the time required for a node to switch from the sending state to the receiving state and correctly decode, or the time required to switch from the receiving state to the sending state.

[0069] OFDM stands for Orthogonal frequency-division multiplexing;

[0070] DIFS stands for Distributed-Coordination-Function Interframe Space (DIFS). In the DCF protocol, a node must monitor the channel for idleness before starting to send data. If the channel is idle, the node must wait for the DIFS period before starting to send data. If the channel is detected as busy at any time during the DIFS period, the node is forced to postpone its data transmission.

[0071] EIFS stands for Extended Interframe Space (EIFS), which means that if an error occurs in the previous frame, the sending node has to delay the sending of the next frame by EIFS instead of DIFS.

[0072] As an example, when the autoencoder at the WiFi transmitter needs to update the AI ​​model parameters, a trigger frame with a special sequence is sent to the WiFi receiver within a DIFS interframe interval or an EIFS interframe interval to trigger the autoencoder at the WiFi receiver to feed back the new AI model parameters.

[0073] It should be noted that the above interaction process can be implemented through a traditional WIFI transmitter and a traditional WIFI receiver, so that the hardware such as the autoencoder AE added to the WIFI transmitting end and the WIFI receiving end can work normally.

[0074] In an optional embodiment, new AI model parameters fed back by the autoencoder of the WIFI receiving end are received, and a confirmation message is sent to the WIFI receiving end.

[0075] As an example, the WIFI transmitter can receive the new AI model parameters fed back by the autoencoder of the WIFI receiver within a SIFS inter-frame interval, and send an acknowledgment message (ACK) to the WIFI receiver to notify the WIFI receiver whether the reception is successful.

[0076] To further facilitate understanding by those skilled in the art, the following describes the interaction process within the DIFS frame interval and the interaction process within the EIFS frame interval using two embodiments:

[0077] Example 1:

[0078] Figure 4 is a schematic diagram of AI parameter interaction within the DIFS frame interval according to an embodiment of the present disclosure. As shown in Figure 4, after completing a transmission or reception, the Wi-Fi device can enter the listening phase. The listening time is the DIFS time in the figure. At this time, if the AP needs to update the AI ​​model parameters, the Wi-Fi transmitter can send a trigger frame with a special field to trigger the Wi-Fi receiver to send new AI model parameters.

[0079] After receiving the trigger frame, the Wi-Fi receiver can first match the special fields on the trigger frame with the special fields stored locally. If a match is found, it starts to construct the new AI model parameters required by the Wi-Fi sender based on the content on the trigger frame.

[0080] After a SIFS period, the Wi-Fi receiver can return the new AI model parameters to the Wi-Fi transmitter through the feedback link and place the new AI model parameters into the neural network of the Wi-Fi transmitter, completing the interaction process.

[0081] Example 2:

[0082] Figure 5 is a schematic diagram of AI parameter interaction within the EIFS frame interval according to an embodiment of the present disclosure. As shown in Figure 5, the WIFI transmitter needs to first send an error frame that cannot pass the check during the nth transmission process, and then enter the monitoring phase. The monitoring time is the EIFS time in the figure. At this time, if the WIFI transmitter needs to update the model parameters, it can send a trigger frame with a special field to trigger the WIFI receiver to send the AI ​​model parameters.

[0083] After receiving the trigger frame, the Wi-Fi receiver can first match the special fields on the trigger frame with the special fields stored locally. If a match is found, it starts to construct the new AI model parameters required by the Wi-Fi sender based on the content on the trigger frame.

[0084] When the Wi-Fi receiver feeds new AI model parameters to the Wi-Fi transmitter, it can transmit multiple frames continuously using the EIFS interval, rather than using the DIFS interval. For example, the AI ​​model parameters can be fed back to the Wi-Fi transmitter in multiple parts, or the complete AI model parameters can be fed back to the Wi-Fi transmitter multiple times. The Wi-Fi transmitter then calculates the average value based on these multiple complete AI model parameters.

[0085] After a SIFS period, the Wi-Fi receiver can return the new AI model parameters to the Wi-Fi transmitter through the feedback link and place the new AI model parameters into the neural network of the Wi-Fi transmitter, completing the interaction process.

[0086] After completing the reception, the WIFI transmitter may send an ACK to the WIFI receiver to notify the WIFI receiver of the successful reception.

[0087] In the disclosed embodiments, autoencoders are respectively provided at the Wi-Fi transmitter and the Wi-Fi receiver. After AI model parameters are set in the autoencoder at the Wi-Fi transmitter, the autoencoder at the Wi-Fi transmitter sends data frames with a special sequence to the Wi-Fi receiver. The Wi-Fi receiver identifies the data frames based on the special sequence and, if the identification is successful, decodes the data frames through the autoencoder at the Wi-Fi receiver. The autoencoder improves the coding efficiency and decoding accuracy of traditional Wi-Fi transmitter and receiver systems, thereby enhancing Wi-Fi network performance and user experience. This solves the problem in related technologies that AI / ML has not been applied to specific WLAN implementations, and therefore cannot achieve the effect of enhancing network performance and user experience through AI / ML.

[0088] In this embodiment, another method for interacting AI model parameters running on the network architecture of the above-mentioned WIFI system is provided. FIG6 is a flowchart of the interaction of AI model parameters according to another embodiment of the present disclosure, which is applied to a WIFI receiving end, wherein the WIFI receiving end includes an autoencoder and a classifier of the WIFI receiving end. As shown in FIG5 , the process includes the following steps:

[0089] Step S601: After the AI ​​model parameters are set in the autoencoder of the WIFI transmitter, the classifier receives a data frame with a special sequence sent by the autoencoder of the WIFI transmitter, and identifies the data frame according to the special sequence.

[0090] As an example, after the autoencoder of the WIFI transmitter is set into the AI ​​model parameters, the classifier can receive the data frame with a special sequence sent by the autoencoder of the WIFI transmitter and identify the data frame according to the special sequence.

[0091] As an example, the data frame may be a code vector encoded by AE of a WIFI transmitter, and may be scrambled by a special pseudo noise (PN) sequence. The classifier may use a locally cached PN sequence for detection and identification.

[0092] Step S602: If the identification is successful, the data frame is decoded by the autoencoder of the WIFI receiving end.

[0093] As an example, the data frame may be a coding vector encoded by AE of the WIFI transmitter. If the recognition is successful, the coding vector encoded by AE of the WIFI transmitter may be sent to the AE decoder of the WIFI receiver, and decoded using the AE decoder of the WIFI receiver, thereby improving the accuracy of the autoencoder.

[0094] In an optional embodiment, after the data frame is decoded by the autoencoder of the WIFI receiving end, the method further includes: when the autoencoder of the WIFI transmitting end needs to update the AI ​​model parameters, receiving a trigger frame with a special sequence sent by the WIFI transmitting end; identifying the trigger frame according to the special sequence by the classifier; if the identification is successful, training the autoencoder of the WIFI receiving end to obtain new AI model parameters; and feeding back the new AI model parameters to the autoencoder of the WIFI transmitting end.

[0095] As an example, when the autoencoder of the Wi-Fi transmitter needs to update the AI ​​model parameters, the Wi-Fi transmitter can send a trigger frame with a special sequence to the classifier, and the classifier can identify the trigger frame based on the special sequence on the trigger frame.

[0096] As an example, when the recognition is successful, the WIFI receiving end can train the autoencoder of the WIFI receiving end to obtain new AI model parameters, and the new AI model parameters can be fed back to the autoencoder of the WIFI sending end through a feedback link.

[0097] As an example, in the embodiment of the present disclosure, the WIFI receiving end undertakes the main computing work, namely the training work, and the steps are as follows:

[0098] 1. The data to be sent s=[s1,s2,...,s n ]Normalization, the normalization method is as follows:

[0099] 2. Normalize the data s through the encoder norm Mapped to the encoding vector x k x k =f i (f i-1 (...f1(s norm |θ1)|θ i-1 |θ i )

[0100] 3. The encoded vector x is decoded by the decoder k Mapping to an estimate of the original data

[0101] in, represents the estimate of the original data, θ i Represents the trainable network parameters, i.e. w i and b i ;

[0102] Basic network unit function f i (x) The general expression is as follows: f i (x) = f a (w i ⊙x+b i )

[0103] Among them, ⊙ indicates that the network operator is defined by the neural network type. Commonly used networks include fully connected networks DNN (Dense Neural Network, DNN), Convolutional Neural Network (Convolutional Neural Network, CNN), and Recurrent Neural Network (Recurrent Neural Network, RNN). a (x) represents the activation function, which can include Relu activation function, Tanh activation function, Sigmoid activation function, etc.

[0104] 4. Train the network parameters using the loss function. The loss function represents the gap between the deep learning evaluation module and the actual situation. The loss function uses the mean square error and is defined as follows:

[0105] Since the basic network unit uses normalized evaluation parameters, cross entropy can also be used as the cost function:

[0106] The WIFI receiver periodically calls the trainer for training. The WIFI receiver can use the new data collected in real time during the current cycle to train the existing network. The training object is the trainable parameter w i 、b i The training method is gradient descent, which can include RMSProp algorithm, Adam algorithm, etc. After the training is completed, the trained network parameters, i.e., w, can be converted to i and b i Feedback is sent to the WiFi transmitter. The data exchanged during the feedback process are the trainable parameters of the encoder or decoder in AE.

[0107] In an optional embodiment, identifying the trigger frame according to the special sequence by the classifier includes: matching the special sequence of the trigger frame with a locally stored special sequence; and determining that the identification is successful if the match is successful.

[0108] As an example, the data field of the trigger frame can be a special sequence. The recognition trigger signal can match the special sequence on the trigger frame with the locally stored feature sequence. If the match is successful, it can be determined that the recognition is successful and the AE parameter interaction can continue.

[0109] After the AI ​​model parameters are embedded in the Wi-Fi transmitter's autoencoder, a classifier receives data frames with a special sequence from the Wi-Fi transmitter's autoencoder and identifies the data frames based on the special sequence. If successful, the Wi-Fi receiver's autoencoder decodes the data frames. The autoencoder improves the coding efficiency and decoding accuracy of traditional Wi-Fi transmitters and receivers, enhancing Wi-Fi network performance and user experience. This addresses the issue of related technologies failing to leverage AI / ML for WLAN implementations, preventing them from achieving the desired results.

[0110] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present disclosure is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), including a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present disclosure.

[0111] In this embodiment, an interactive device for AI model parameters is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated here. As used below, the term "module" can implement a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceived.

[0112] FIG7 is a structural block diagram of an AI model parameter interaction device according to an embodiment of the present disclosure, which is applied to a WIFI transmitter. The WIFI transmitter includes an autoencoder of the WIFI transmitter. As shown in FIG7 , the device includes:

[0113] The first sending module 701 is configured to send a data frame with a special sequence to the WIFI receiving end through the autoencoder of the WIFI transmitting end after the AI ​​model parameters are set in the autoencoder of the WIFI transmitting end, so that the WIFI receiving end recognizes the data frame according to the special sequence, and decodes the data frame through the autoencoder of the WIFI receiving end if the recognition is successful.

[0114] In an optional embodiment, the method further includes:

[0115] a monitoring module configured to monitor a feedback message from the WIFI receiving end after the data frame with the special sequence is sent to the WIFI receiving end through the autoencoder of the WIFI transmitting end;

[0116] The determination module is configured to determine that the AI ​​model parameters of the autoencoder of the WIFI sending end need to be updated if no feedback message sent by the WIFI receiving end is monitored within a preset time.

[0117] In an optional embodiment, the method further includes:

[0118] The second sending module is configured to send a trigger frame with a special sequence to the WIFI receiving end within an inter-frame interval when the autoencoder at the WIFI transmitting end needs to update the AI ​​model parameters, so as to trigger the autoencoder at the WIFI receiving end to feedback new AI model parameters.

[0119] In an optional embodiment, the method further includes:

[0120] The third sending module is configured to receive the new AI model parameters fed back by the autoencoder of the WIFI receiving end, and send a confirmation message to the WIFI receiving end.

[0121] In an optional embodiment, the data frame with the special sequence carries feedback channel state information CSI content.

[0122] FIG8 is a structural block diagram of an AI model parameter interaction device according to another embodiment of the present disclosure, which is applied to a WIFI receiving end, wherein the WIFI receiving end includes an autoencoder and a classifier of the WIFI receiving end. As shown in FIG8 , the device includes:

[0123] The first receiving module 801 is configured to receive, through the classifier, a data frame with a special sequence sent by the encoder of the WIFI transmitting end after the AI ​​model parameters are set in the autoencoder of the WIFI transmitting end, and identify the data frame according to the special sequence;

[0124] The decoding module 802 is configured to decode the data frame through the autoencoder of the WIFI receiving end when the recognition is successful.

[0125] In an optional embodiment, the method further includes:

[0126] A second receiving module is configured to receive a trigger frame with a special sequence sent by the WIFI transmitting end after the data frame is decoded by the autoencoder of the WIFI receiving end and when the autoencoder of the WIFI transmitting end needs to update the AI ​​model parameters;

[0127] an identification module, configured to identify the trigger frame according to the special sequence through the classifier;

[0128] A training module is configured to train the autoencoder of the WIFI receiving end to obtain new AI model parameters when the recognition is successful;

[0129] A feedback module is configured to feed back the new AI model parameters to the autoencoder of the WiFi transmitting end.

[0130] In an optional embodiment, the identification module includes:

[0131] a matching submodule, configured to match the special sequence of the trigger frame with a locally stored special sequence;

[0132] The determination submodule is configured to determine that the recognition is successful if the match is successful.

[0133] It should be noted that the above modules can be implemented through software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0134] An embodiment of the present disclosure further provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured to execute the steps of any one of the above method embodiments when run.

[0135] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0136] An embodiment of the present disclosure further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0137] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0138] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail here.

[0139] Obviously, those skilled in the art should understand that the modules or steps of the present disclosure described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed across a network composed of multiple computing devices, they can be implemented using program code executable by the computing device, and thus, they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be performed in a different order than herein, or they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. Thus, the present disclosure is not limited to any particular combination of hardware and software.

[0140] The foregoing description is merely a preferred embodiment of the present disclosure and is not intended to limit the present disclosure. Those skilled in the art will readily appreciate that various modifications and variations of the present disclosure are possible. Any modifications, equivalent substitutions, or improvements made within the principles of the present disclosure shall be included within the scope of protection of the present disclosure.

Claims

1. An interactive method of AI model parameters, applied to a WIFI transmitter, wherein: The WIFI transmitter includes an autoencoder of the WIFI transmitter, and the method includes: After the autoencoder of the WIFI transmitting end is placed into the AI ​​model parameters, a data frame with a special sequence is sent to the WIFI receiving end through the autoencoder of the WIFI transmitting end, so that the WIFI receiving end identifies the data frame according to the special sequence, and if the identification is successful, the data frame is decoded by the autoencoder of the WIFI receiving end.

2. The method according to claim 1, wherein: After the data frame with the special sequence is sent to the WIFI receiving end through the autoencoder of the WIFI transmitting end, the method further includes: Monitoring the feedback message of the WIFI receiving end; If no feedback message sent by the WIFI receiving end is monitored within the preset time, it is determined that the AI ​​model parameters of the autoencoder of the WIFI sending end need to be updated.

3. The method according to claim 2, wherein: Also includes: When the autoencoder of the WIFI transmitting end needs to update the AI ​​model parameters, a trigger frame with a special sequence is sent to the WIFI receiving end within an inter-frame interval to trigger the autoencoder of the WIFI receiving end to feed back new AI model parameters.

4. The method according to claim 3, wherein: Also includes: Receive new AI model parameters fed back by the autoencoder of the WIFI receiving end, and send a confirmation message to the WIFI receiving end.

5. The method according to claim 1, wherein: The data frame with the special sequence carries the feedback channel state information CSI content.

6. An interactive method of AI model parameters, applied to a WIFI receiver, wherein: The WIFI receiving end includes an autoencoder and a classifier of the WIFI receiving end, and the method includes: After the autoencoder of the WIFI transmitter is placed into the AI ​​model parameters, the data frame with the special sequence sent by the autoencoder of the WIFI transmitter is received by the classifier, and the data frame is identified according to the special sequence; When the recognition is successful, the data frame is decoded by the autoencoder of the WIFI receiving end.

7. The method according to claim 6, wherein: After the data frame is decoded by the autoencoder of the WIFI receiving end, the method further includes: When the autoencoder of the WIFI transmitter needs to update the AI ​​model parameters, receiving a trigger frame with a special sequence sent by the WIFI transmitter; identifying the trigger frame according to the special sequence by the classifier; If the recognition is successful, the autoencoder of the WIFI receiving end is trained to obtain new AI model parameters; The new AI model parameters are fed back to the autoencoder of the WIFI sending end.

8. The method according to claim 7, wherein: The identifying the trigger frame according to the special sequence by the classifier includes: Matching the special sequence of the trigger frame with a special sequence stored locally; In case of successful matching, it is determined that the recognition is successful.

9. An interactive device for AI model parameters, applied to a WIFI transmitter, the WIFI transmitter comprising an autoencoder of the WIFI transmitter, the device comprising: The first sending module is configured to send a data frame with a special sequence to a WIFI receiving end through the autoencoder of the WIFI sending end after the AI ​​model parameters are placed in the autoencoder of the WIFI sending end, so that the WIFI receiving end identifies the data frame according to the special sequence, and decodes the data frame through the autoencoder of the WIFI receiving end if the identification is successful.

10. An interactive device for AI model parameters, applied to a WIFI receiving end, the WIFI receiving end comprising a self-decoder and a classifier of the WIFI receiving end, the device comprising: A first receiving module is configured to receive, through the classifier, a data frame with a special sequence sent by the encoder of the WIFI transmitting end after the autoencoder of the WIFI transmitting end is placed into the AI ​​model parameters, and identify the data frame according to the special sequence; The decoding module is configured to decode the data frame through the autoencoder of the WIFI receiving end when the recognition is successful.

11. A computer-readable storage medium having a computer program stored therein, wherein: When the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 5 are implemented, or the steps of the method described in any one of claims 6 to 8 are implemented.

12. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in any one of claims 1 to 5 or the steps of the method described in any one of claims 6 to 8 when executing the computer program.

Images