WO2024022007A1 - Method and apparatus for communication in wireless local area network - Google Patents

Abstract

The present application relates to the field of communications, and in particular, relates to a method and apparatus for communication in a WLAN. The solution may be applied to WLAN systems that support IEEE 802.11ax next-generation Wi-Fi protocols, such as 802.11be, Wi-Fi 7 or EHT, and 802.11be next-generation, Wi-Fi 8 and other 802.11 series protocols, and the solution may also be applied to UWB-based wireless personal area network systems, such as 802.15 series of standards, or applied to sensing systems, such as 802.11bf series of standards. The method comprises: a request site requests information of a neural network from a response site by means of a request, so that the response site may send the requested information of the neural network to the request site according to the request, wherein the information of the neural network is associated with manufacturer information. Thus, a site can obtain suitable information of a neural network for carrying out a communication decision, and the communication performance of the site is ensured.

Description

A method and device for communication in a wireless local area network

This application claims priority to the Chinese patent application with the filing date of July 26, 2022, the application number is 202210885655.6, and the application title is "A method and device for communication in a wireless local area network", the entire content of which is incorporated into this application by reference. middle.

Technical field

The present application relates to the field of communication technology, and more specifically, to a method and device for communication in a wireless local area network.

Background technique

With the rapid development of wireless communications, new wireless technologies, new terminals, and new applications emerge one after another, making wireless networks unprecedentedly complex. The application of artificial intelligence (AI) in wireless networks can help more accurately predict channels, business characteristics, user behaviors, etc. Its advantages in wireless networks have become a consensus in the industry. Among them, AI can be applied to wireless networks through neural networks (NN) to improve the communication performance of wireless networks.

In wireless local area network (WLAN), due to the high mobility of most sites, the wireless LAN network environment where they are located often changes. For example, if a site has been dormant for a period of time, the wireless network environment where it is located may have changed. . For another example, when a non-access point station switches from the current access point to a new access point, the wireless network environment in which it is located may also change. It is difficult for a set of neural networks to be applied to all scenarios. In view of the changing wireless network environment, if an unupdated or other inappropriate neural network is used, it will affect the communication decision-making of the site, for example, affecting the site to select inappropriate channels or transmission rate, etc., which will further affect the communication performance of the site.

Contents of the invention

This application provides a method and device for communication in a wireless local area network, which associates neural network information with manufacturer information, so that the site can obtain appropriate neural network information for communication decisions, and ensure that the site can ensure the site's performance in a changing wireless network environment. communication performance.

In a first aspect, a communication method in a wireless local area network is provided, and the method is executed by a station (station, STA). The site may be a terminal, or a chip, circuit or module configured in the terminal, which is not limited in this application.

As an example, the site could be the requesting site.

The method includes: the requesting site sends a request, the request is used to request neural network information; the requesting site receives a response from the responding site, the response includes the requested neural network information, and the neural network information is associated with the manufacturer information.

Based on the above solution, the requesting site can request neural network information from the responding site through a request, and then the responding site can send the requested neural network information to the requesting site, and the neural network information is associated with the manufacturer information. In this way, This enables the site to obtain appropriate neural network information for communication decisions and ensure the site's communication performance.

On the other hand, this method avoids the site from obtaining inappropriate neural network information from the cloud or server, and also prevents the site from spending a long time training the neural network, which helps reduce communication delays.

In addition, this method avoids the site from continuously training the neural network, which helps to reduce the power consumption of the site, thereby helping the site save energy.

For example, the requesting site may be an access point (AP) or a non-AP site (non-AP STA).

Similarly, the responding site can be a non-AP site or an AP.

Combined with the first aspect, in some implementations, the request includes manufacturer information, or includes identification information of the neural network.

Based on the above solution, the requesting site can obtain the corresponding neural network information from the responding site based on the manufacturer information or the identification information of the neural network, which has higher communication efficiency.

Combined with the first aspect, in some implementations, the response includes manufacturer information.

Based on the above solution, the responding site can send the neural network information and the manufacturer information associated with the neural network information to the requesting site, which helps the responding site make communication decisions based on the vendor information.

Optionally, the response also includes identification information of the neural network.

In some examples, the information of the neural network may include parameters of the neural network and may also include the structure of the neural network.

In some examples, the vendor information includes multiple vendor information.

Specifically, for example, the request may include multiple vendor information, and the multiple vendors may include vendors to which the requested site belongs, or may include vendors supported by the requested site. For another example, the response may include multiple vendor information, and the multiple vendors may include vendors to which the responding site belongs, or vendors supported by the responding site. In this way, devices from manufacturers that support the same neural network can quickly exchange neural network information, achieving higher communication efficiency.

For example, the manufacturer information is the information of the manufacturer corresponding to the equipment manufacturer.

In other words, the device manufacturer of the requesting site may be included in the request or the device manufacturer of the responding site may be included in the response.

Combined with the first aspect, in some implementations, the request includes identification information of a basic service set (BSS), and the neural network information in the response is associated with the identification information of the BSS.

In this way, the site can obtain the neural network information of the target BSS more accurately.

Optionally, the identification information of the basic service set BSS included in the request is used to identify the BSS to which the requesting site belongs.

In conjunction with the first aspect, in some implementations, the request includes a preset condition of the requested neural network, and the request is used to request information of a neural network that satisfies the preset condition.

Based on the above solution, the requesting site can send neural network information that meets the preset conditions to the responding site. In this way, the requesting site terminal can obtain more suitable neural network information, which helps to achieve better communication decisions.

In some examples, the preset condition includes at least one of the following: generation time of the neural network, accuracy of the neural network, and model size of the neural network.

In this application, "the generation time of the neural network" and "the generation time of the information of the neural network" indicate the same meaning, and they can be replaced with each other without limitation. "The accuracy of the neural network" and "the accuracy of the information of the neural network" "Degree" indicates the same meaning, which are interchangeable without limitation.

Combined with the first aspect, in some implementations, the response includes information from multiple neural networks.

In this way, the requesting site can select the information of one neural network from the information of multiple neural networks.

In this application, multiple neural networks with the same structure but different parameters can be understood as multiple information of one neural network, multiple "information of neural networks", or multiple neural networks.

Based on the above solution, the responding site can send multiple neural network information to the requesting site. This allows the requesting site to select more suitable neural network information, which helps to achieve better communication decisions.

In some examples, the response also includes attribute information of the multiple neural networks, and the attribute information includes the generation time of the multiple neural networks, or the accuracy of the multiple neural networks, or the model sizes of the multiple neural networks.

Optionally, the requesting site can select information of one neural network from information of multiple neural networks based on the above attribute information.

Based on the above solution, the attribute information includes the generation time and accuracy of the neural network, etc., which helps the requesting site select the information of the neural network with a closer generation time and better accuracy, and helps to achieve better communication decisions.

Combined with the first aspect, in some implementations, the triggering conditions for the requesting site to send the request include: the information of the neural network stored by the requesting site has not been updated for more than a preset time; or the accuracy of the neural network stored by the requesting site is less than a threshold; Or, the requesting site does not store neural network information or does not store any neural network information; or, the neural network information related to manufacturer information stored by the requesting site has not been updated for more than a preset time; or, the requesting site does not store information related to the manufacturer. Information related to neural networks.

Based on the above solution, the requesting site can send a request under any of the above trigger conditions to obtain the required information of the neural network.

Combined with the first aspect, in some implementations, the triggering conditions for the requesting station to send the request include: the requesting station wakes up after sleeping; or the network environment of the wireless local area network of the requesting station changes.

Based on the above solution, the requesting site can send a request under any of the above trigger conditions to obtain the required information of the neural network.

The second aspect provides a communication method in a wireless local area network. The method can be executed by a station. The station can be a terminal, or a chip, circuit or module configured in the terminal. This application is not limited to this.

As an example, the site could be a responsive site.

The method includes: the response site receives a request from the request site, the request is used to request information of the neural network; the response site sends a response to the request site according to the request, the response includes the requested information of the neural network, the information of the neural network and Manufacturer information association.

In some examples, the information of the neural network includes parameters of the neural network and/or the structure of the neural network.

In some examples, the vendor information includes multiple vendor information.

For example, the manufacturer information is the information of the manufacturer corresponding to the equipment manufacturer.

Combined with the second aspect, in some implementations, the request includes manufacturer information or identification information of the neural network.

Optionally, the response also includes identification information of the neural network.

Combined with the second aspect, in some implementations, the request includes identification information of the basic service set BSS, and the neural network information in the response is associated with the identification information of the BSS.

Optionally, the request includes a preset condition of the requested neural network, and the request is used to request information of a neural network that satisfies the preset condition.

The method may further include: the access point selects information of one neural network from information of multiple neural networks according to preset conditions.

In some examples, the preset condition includes at least one of the following: generation time of the neural network, accuracy of the neural network, and model size of the neural network.

Combined with the second aspect, in some implementations, the response includes manufacturer information.

Optionally, the response includes information from multiple neural networks.

In some examples, the response also includes attribute information of one or more neural networks, and the attribute information includes the generation time of the multiple neural networks, or the accuracy of the multiple neural networks, or the model size of the multiple neural networks.

It should be understood that the beneficial effects of the above-mentioned second aspect and its various implementations can be referred to the first aspect and its various implementations.

Combined with the second aspect, in some implementations, the triggering conditions for the requesting site to send the request include: the information of the neural network stored by the requesting site has not been updated for more than a preset time; or the accuracy of the neural network stored by the requesting site is less than the threshold; Or, the requesting site does not store neural network information or does not store any neural network information; or, the neural network information related to manufacturer information stored by the requesting site has not been updated for more than a preset time; or, the requesting site does not store information related to the manufacturer. Information related to neural networks.

Combined with the second aspect, in some implementations, the triggering conditions for the requesting station to send the request include: the requesting station wakes up after sleeping; or the network environment of the wireless local area network of the requesting station changes.

A third aspect provides a communication device, which has the function of implementing the method of any possible implementation of the first aspect and the second aspect. The functions described can be implemented by hardware, or can be implemented by hardware executing corresponding software. The hardware or software includes one or more units corresponding to the above functions.

In a fourth aspect, a communication device is provided, including a processor and a memory. Optionally, a transceiver may also be included. Wherein, the memory is used to store computer programs, and the processor is used to call and run the computer programs stored in the memory, and control the transceiver to send and receive signals, so that the communication device performs any possible implementation manner of the above-mentioned first aspect and second aspect. method in.

In a fifth aspect, a communication device is provided, including a processor and a communication interface. The communication interface is used to receive data and/or information and transmit the received data and/or information to the processor. The processing The processor processes the data and/or information, and the communication interface is also used to output the data and/or information processed by the processor, so that the method of any possible implementation of the above first aspect and the second aspect be executed.

In a sixth aspect, a computer-readable storage medium is provided. Computer instructions are stored in the computer-readable storage medium. When the computer instructions are run on a computer, any one of the above first and second aspects is possible. The methods in the implementation are executed.

In a seventh aspect, a computer program product is provided. The computer program product includes computer program code. When the computer program code is run on a computer, any possible implementation of the first aspect and the second aspect is achieved. The method in is executed.

An eighth aspect provides a wireless communication system, including the requesting site in the first aspect and the responding site in the second aspect.

Description of drawings

FIG. 1 is a schematic diagram of a system architecture 100 and a schematic diagram of the structure of a device provided by an embodiment of the present application.

Figure 2 is a schematic diagram of the structure of a neural network.

Figure 3 is a schematic diagram of a neuron calculating output based on input.

Figure 4 shows a schematic diagram of changes in the wireless network environment of the site.

Figure 5 shows a schematic diagram of a method for updating neural network parameters.

Figure 6 is a schematic flow chart of a communication method 200 in a wireless local area network provided by an embodiment of the present application.

Figure 7 is a schematic flow chart of a communication method 300 in a wireless local area network provided by an embodiment of the present application.

Figure 8 is a schematic flow chart of a communication method 400 in a wireless local area network provided by an embodiment of the present application.

Figure 9 is a schematic flow chart of a communication method 500 in a wireless local area network provided by an embodiment of the present application.

FIG. 10 is a schematic diagram of a communication device 600 provided by an embodiment of the present application.

Figure 11 is a schematic structural diagram of a communication device 700 provided by an embodiment of the present application.

Figure 12 is a schematic structural diagram of a communication device 800 provided by an embodiment of the present application.

Detailed ways

The technical solutions in this application will be described below with reference to the accompanying drawings.

The technical solutions provided by the embodiments of this application can be applied to wireless local area network (WLAN) scenarios, for example, supporting Institute of Electrical and Electronics Engineers (Institute of Electrical and Electronics Engineers, IEEE) 802.11 related standards, such as 802.11a/b /g standard, 802.11n standard, 802.11ac standard, 802.11ax standard, IEEE 802.11ax next-generation Wi-Fi protocol, such as 802.11be, Wi-Fi 7, extremely high throughput (EHT), 802.11ad, 802.11ay or 802.11bf, such as 802.11be next generation, Wi-Fi 8, etc., can also be applied to wireless personal area network systems based on ultra wide band (UWB), such as the 802.15 series of standards, or for sensing ( sensing) system, such as the 802.11bf series standards. Among them, 802.11bf includes two major categories of standards: low frequency (sub7GHz) and high frequency (60GHz). The implementation of sub7GHz mainly relies on standards such as 802.11ac, 802.11ax, 802.11be and the next generation. The implementation of 60GHz mainly relies on standards such as 802.11ad, 802.11ay and the next generation.

Although the embodiments of the present application are mainly explained by taking the deployment of WLAN networks, especially networks applying the IEEE 802.11 system standard, as an example, those skilled in the art can easily understand that all aspects involved in the embodiments of the present application can be extended to use various standards or protocols. Other networks, such as high performance wireless local area network (HIPERLAN), wireless wide area network (WWAN), wireless personal area network (WPAN) or other now known or hereafter developed network. Therefore, regardless of the coverage and wireless access protocol used, the various aspects provided by the embodiments of the present application can be applied to any suitable wireless network.

The technical solutions of the embodiments of the present application can also be applied to various communication systems, such as: WLAN communication systems, wireless fidelity (Wi-Fi) systems, long term evolution (LTE) systems, LTE frequency division dual Frequency division duplex (FDD) system, LTE time division duplex (TDD), universal mobile telecommunication system (UMTS), global interoperability for microwave access (WiMAX) communication system, fifth generation (5th generation, 5G) system or new radio (NR) system, future sixth generation (6th generation, 6G) system, Internet of things (IoT) network or vehicle Internet (vehicle) to x, V2X) and other wireless LAN systems.

The above-mentioned communication systems applicable to the present application are only examples. The communication systems applicable to the present application are not limited to these and will be explained uniformly here, and will not be described in detail below.

FIG. 1 is a schematic diagram of a system architecture 100 and a schematic diagram of the structure of a device provided by an embodiment of the present application. Among them, (a) of FIG. 1 is an example of the system architecture 100 suitable for the embodiment of the present application. As shown in (a) of Figure 1, the system 100 includes multiple stations (stations, STAs), where the stations can be access points (access points, AP) 110 and access points AP 120, or they can be access points. The non-AP sites (non-AP STA) associated with the entry point AP110, for example, non-AP STA111, non-AP STA 112, and non-AP STA 113, and the non-AP sites associated with the access point AP2, for example, non-AP STA 121, non-AP STA AP STA 122, non-AP STA 123. Among them, AP110, non-APSTA111, non-AP STA 112, and non-AP STA 113 constitute the basic service set (BSS) 1, and AP 120, non-AP STA 121, non-AP STA 122, and non-AP STA 123 constitute BSS2.

It should be understood that in this application, unless otherwise specified, a site refers to a broad site, which includes AP and non-AP STAs.

As an example, the system architecture shown in (a) of Figure 1 can be applied to the Internet of Things industry, Internet of Vehicles industry, banking industry, corporate offices, sports venues and exhibition halls, concert halls, hotel rooms, dormitories, wards, classrooms, shopping malls and supermarkets , squares, streets, production workshops and warehouses, etc.

Among them, the access point can be an access point for a terminal (for example, a mobile phone) to enter a wired (or wireless) network, which is mainly deployed in homes, Inside the building and the park, the typical coverage radius is tens to hundreds of meters. Of course, it can also be deployed outdoors. The access point is equivalent to a bridge connecting the wired network and the wireless network. Its main function is to connect various wireless network clients together, and then connect the wireless network to the Ethernet.

Specifically, the access point can be a terminal or network device with a Wi-Fi chip. The network device can be a router, a relay station, a vehicle-mounted device, a wearable device, a network device in a 5G network, and a network device in a future 6G network. Or network equipment in a public land mobile communication network (public land mobile network, PLMN), etc., which are not limited by the embodiments of this application. The access point can be a device that supports the 802.11be standard. The access point can also be a device that supports multiple WLAN standards of the 802.11 family such as 802.11ax, 802.11ac, 802.11n, 802.11g, 802.11b, 802.11a, and 802.11be next generation. The access point in this application can be a highly efficient (HE) AP or an extremely high throughput (EHT) AP, or it can be an access point suitable for a certain future generation of Wi-Fi standards.

Non-AP sites can be wireless communication chips, wireless sensors or wireless communication terminals, etc., and can also be called users, user equipment (UE), access terminals, user units, user stations, mobile stations, mobile stations, and remote stations. , remote terminal, mobile device, user terminal, terminal, wireless communications device, user agent or user device. Non-AP sites can be cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), devices with wireless communications Functional handheld devices, computing devices or other processing devices connected to wireless modems, vehicle-mounted devices, Internet of Things devices, wearable devices, terminal devices in 5G networks, terminal devices in future 6G networks or terminal devices in PLMN, etc., The embodiments of the present application are not limited to this. Non-AP sites can support the 802.11be standard. Non-AP sites can also support multiple WLAN standards of the 802.11 family such as 802.11ax, 802.11ac, 802.11n, 802.11g, 802.11b, 802.11a, and 802.11be next generation.

As an example, the access points or non-AP sites in this application can be sensor nodes in smart cities, such as smart water meters, smart electricity meters, and smart air detection nodes, or smart devices in smart homes, such as smart cameras and projectors. Instruments, display screens, televisions, stereos, refrigerators, washing machines, etc. It can also be entertainment terminals, such as wearable devices such as virtual reality (VR) and augmented reality (AR), or smart offices. The smart devices in the Internet, such as printers, projectors, loudspeakers, speakers, etc., can also be infrastructure in daily life scenes, such as vending machines, self-service navigation stations in supermarkets, self-service checkout equipment, self-service ordering machines, etc. , it can also be the Internet of Vehicles equipment in the Internet of Vehicles, nodes in the Internet of Things, and equipment in large sports and music venues, etc.

Among them, access points and non-AP sites have certain artificial intelligence (AI) capabilities and can use neural networks for reasoning and decision-making. Non-AP sites and/or access points can also perform neural network training.

Figure 1(b) is a schematic structural diagram of a device provided by an embodiment of the present application. Among them, the device can be an access point or a non-AP site. The internal functional modules of the device include a central processor, a media access control (media access control, MAC) processing module, a transceiver, an antenna, and a neural network processing Unit (neural network processing unit, NPU). Among them, the transceiver includes a physical layer (PHY) processing module, the NPU includes an inference module, and optionally, the NPU also includes a training module. For example, for some IoT terminals, the training module is optional. Among them, the training module is used to train the neural network and output the neural network parameters. The trained neural network parameters will be fed back to the inference module. The NPU can act on various other modules of the device, including the central processor, MAC processing module, transceiver and antenna. The NPU can be responsible for the decision-making tasks of each module. For example, it interacts with the transceiver. The decision-making switch of the transceiver is used to save energy. For example, it interacts with the antenna and controls the orientation of the antenna. For example, it interacts with the MAC processing module and controls channel access. , channel selection and spatial multiplexing decisions, etc. The solution of this application can obtain appropriate neural network information for communication decision-making. The neural network information can be applied to the communication decision-making of the MAC processing module or the communication decision-making of the transceiver. This application does not limit this. Among them, the communication decision of the transceiver includes the communication decision of the PHY processing module. It can be understood that the schematic diagram of the device provided in Figure (b) is an example and does not constitute a limitation on the device of the present application.

With the rapid development of wireless communications, new wireless technologies, new terminals, and new applications emerge one after another, making wireless networks unprecedentedly complex. Artificial intelligence (AI) is an effective tool to improve wireless network performance, and its advantages in wireless networks have become a consensus in the industry. Specifically, the advantages of AI include the following four aspects:

1. Solve complex network problems without mathematical models;

2. Solve the problem of wireless network management with large search space;

3. Cross-layer and cross-node network-level global optimization;

4. Actively optimize wireless network parameters through AI’s prediction capabilities.

AI can be applied to channel access, rate adaptation, channel aggregation or channel prediction, etc.

The operations of traditional wireless networks, such as channel prediction, are determined based on rules, such as predicting the channel through an algorithm or function, expressed as f(·). The operation rules of each step of f(·) are determined, for example, y=f(x). The calculation from input x to output y is a clear rule and applies to all wireless network environments. After the introduction of AI, f(·) is no longer rule-based, but is described by a neural network (NN). For example, it can be described by the neural network structure and neural network parameters, expressed as f(θ,·) , where θ represents the neural network parameters. Neural network training for actual wireless network environments can naturally better meet the needs of corresponding wireless networks and further improve the performance of wireless networks.

The following is a brief introduction to neural networks.

Neural network is a machine learning technology that simulates the neural network of the human brain in order to achieve artificial intelligence. A neural network can include 3 layers, an input layer, at least one intermediate layer (also called a hidden layer), and an output layer, or more layers. Deeper neural networks may contain more hidden layers between the input and output layers. The following takes a neural network as an example to illustrate it.

Figure 2 is a schematic diagram of the structure of a neural network. As shown in Figure 2, the neural network is a fully connected neural network. The neural network includes 3 layers, namely the input layer, the hidden layer and the output layer. The input layer has 3 neurons and the hidden layer has 4 neurons. , the output layer has 2 neurons, and the neurons in each layer are fully connected to the neurons in the next layer. Each connection between neurons corresponds to a weight, and each neuron in the hidden layer and output layer can also correspond to a weight. bias. Neural network includes the structure of neural network and the parameters of neural network. Among them, the structure of the neural network refers to the number of neurons contained in each layer and how the output of the previous neurons is input to the following neurons, that is, the connection relationship between neurons. The parameters of the neural network indicate the weights and biases. As can be seen from Figure 2, each neuron may have multiple input connections, and each neuron calculates an output based on the input.

Figure 3 is a schematic diagram of a neuron calculating output based on input. As shown in Figure 3, this neuron contains 3 inputs, 1 output, and 2 calculation functions. The output calculation formula can be expressed as:
Output = activation function (input ₁ * weight ₁ + input ₂ * weight ₂ + input ₃ * weight ₃ + bias)…………(1-1)

The symbol "*" represents the mathematical operation "multiply" or "multiply by", which will not be described in detail below.

Each neuron may have multiple output connections, and the output of one neuron serves as the input of the next neuron. It should be understood that the input layer only has output connections, each neuron of the input layer is the value input to the neural network, and the output value of each neuron is directly used as the input of all output connections. The output layer only has input connections, and the output is calculated using the calculation method of the above formula (1-1). Optionally, the output layer does not need to calculate the activation function, which means that the aforementioned formula (1-1) can be transformed into: output = input ₁ * weight ₁ + input ₂ * weight ₂ + input ₃ * weight ₃ + bias.

For example, a k-layer neural network can be expressed as:
y＝f _k (f _k-1 (...(f1(w ₁ *x+b ₁ ))))…………………………………………(1-2)

Among them, x represents the input of the neural network, y represents the output of the neural network, w _i represents the weight of the i-th layer neural network, b _i represents the bias of the i-th layer neural network, and f _i represents the activation function of the i-th layer neural network. . i=1,2,…,k.

In a wireless local area network (WLAN), since most sites are highly mobile, the wireless LAN network environment where they are located often changes. For example, if a site is dormant for a period of time, the wireless network connected to the site may have changed. For another example, if a non-AP station switches from the current access point to a new access point, the wireless network it is connected to may also change. The wireless network connected to the non-AP site is the wireless network environment in which the non-AP site is located. It is difficult to apply a set of neural networks or inappropriate neural networks to all scenarios. In response to the changing wireless network environment, neural network information needs to be updated.

Figure 4 shows a schematic diagram of changes in the wireless network environment of a non-AP site. As shown in Figure 4, non-AP site 1 moves and switches from AP1 to AP2, and its wireless network environment changes. Since the neural network used by non-AP site 1 does not have information about the wireless network environment where AP2 is located, it cannot achieve optimal communication performance. Therefore, the neural network needs to be updated, for example, the neural network parameters are updated or the neural network is replaced.

Figure 5 shows a schematic diagram of a neural network update method. As shown in Figure 5, non-AP STA can send an update request to the cloud or server through the AP. The cloud or server obtains the updated neural network or neural network parameters, and then sends it to the non-AP STA through the AP.

However, this method requires wireless network access to the Internet. In wireless networks, in many cases it is not guaranteed that the wireless network can access the Internet. In addition, sending neural network parameters through the cloud or server may cause a large delay, thereby affecting the communication performance of the site. In addition, this method is difficult to achieve refined optimization and configuration. For example, it is difficult for a neural network stored in the cloud to adapt to a refined wireless environment, that is, a certain BSS.

In another neural network update method, non-AP STAs are trained in real time, and network-side training and distribution are not required. However, in reality Training will incur a large overhead. Some non-AP STAs have insufficient power and are inconvenient for training. Some non-AP STAs have limited computing power and cannot be trained. This method cannot be universally applicable. In addition, if the site is dormant, it will take some time to learn a neural network with better performance. In other words, this method will bring a large delay and affect the communication performance of the site.

In view of this, this application provides a communication method in a wireless LAN, which associates neural network information with manufacturer information, so that sites in the wireless LAN (the site can be an AP or a non-AP site) can obtain Using appropriate neural network information for communication decisions can ensure the communication performance of the site in a changing wireless network environment.

Figure 6 is a schematic flowchart of a communication method 200 in a wireless local area network provided by an embodiment of the present application. The method 200 may include the following steps.

S210, the requesting site sends a request, which is used to request information about the neural network.

For example, the requesting site may be a non-access point site (non-AP STA) or an AP, which is not limited in this application.

Accordingly, the responding site receives the request from the requesting site.

Optionally, the request may include one or more of the following: manufacturer information, identification information of the neural network, identification information of the basic service set, generation time of the neural network, accuracy of the neural network, model size of the neural network, Used to obtain information that better meets the needs of the requested site or to obtain a more appropriate neural network. It can be understood that the request may also include other information related to the information requested by the neural network, which is not limited in the embodiment of the present application. For the content and triggering conditions of the request sent by the requesting site, please refer to the description of the embodiments shown in Figures 7 to 9 below, and will not be described again here.

For example, the responding site may be a non-AP site or an AP, which is not limited in this application.

S220: The responding site sends a response to the requesting site according to the request. The response includes the requested neural network information, and the neural network information is associated with the manufacturer information.

Accordingly, the requesting site receives the response from the responding site.

Optionally, the response may also include one or more of the following related to the neural network information: manufacturer information, identification information, identification information of the basic service set, generation time, accuracy, model size, etc., In order to facilitate the requesting site to further determine the appropriate neural network based on the information of the neural network. It can be understood that the response may also include other information related to the neural network information, which is not limited in the embodiment of the present invention. For the content of the response sent by the responding site, please refer to the description of the following embodiments and will not be described again here.

The communication method in the wireless LAN provided by this application can be applied to the communication between non-AP sites and APs, the communication between non-AP sites and non-AP sites, and the communication between APs. This application There are no restrictions on application.

Taking the communication between non-AP stations and APs as an example, Figures 7-9 illustrate some specific embodiments of the communication method 200 in the wireless local area network provided by this application. The relevant contents of the following embodiments can be applied to this application. The communication method of wireless LAN will not be described here.

Figure 7 is a schematic flow chart of a communication method 300 in a wireless local area network provided by an embodiment of the present application. In this method, the requesting site is a non-AP site as an example. In the method 300, the requesting site is called the first site, and the responding site is an access point as an example. It should be noted that the examples in this embodiment can also be applied to other embodiments, such as the solutions shown in Figure 8 or Figure 9 . The method 300 may include the following steps.

S310: The first station sends a first request to the access point, and accordingly, the access point receives the first request.

Among them, the first request is used to request information about the neural network.

Specifically, the first request may be a model request, used to request neural network-related information required by the first site. It should be understood that the terms "neural network" and "neural network model" are interchangeable and have consistent meanings in this application.

As an example, the first request may be a management frame, for example, a probe request (probe request) or an association request (Association Request). The first request may also be a control frame, for example, a request to send (RTS) or Block Acknowledgment Request (Block AcknowledgementRequest, BlockAckReq). The first request can also be carried in the header of any message. In addition, the first request may also be other management frames or control frames. The embodiments of the present application do not limit this.

The information of the neural network can also be called the information of the neural network model, or the model information of the neural network. In this embodiment of the present application, the neural network required by the first site may be called the first neural network. The information of the first neural network can be understood as the information of the neural network required by the first site, and can also be called the target information of the neural network, or the target neural network information. For example, the information of the first neural network may include parameters of the first neural network and/or the structure of the first neural network. Wherein, the parameters of the first neural network include weights and/or biases of the first neural network. Wherein, the structure of the first neural network may include the number of neurons of the first neural network, the number of layers of the neural network, the number of neural networks in each layer, the number of hidden layers, and the connection relationship between the neurons. one or more items of information.

It should be understood that the "information of the first neural network" in this application can also be other forms of information related to the first neural network, or other forms of information used to reflect the calculation method of the first neural network. This application does not limit this. The first station can obtain the first neural network to make communication decisions based on the information of the first neural network.

Wherein, the first request may include manufacturer information, where "include" may be an explicit inclusion or an implicit inclusion, such as an implicit indication through the default relationship between other information carried in the first request and the manufacturer information. . For example, the manufacturer information includes the manufacturer's identification information, and the manufacturer's identification information can be used to distinguish different manufacturers. The identification information of the manufacturer may be the manufacturer ID, for example, 1, 2, 3, 4, 5, etc., or it may be the name of the manufacturer. Optionally, the manufacturer information can also be carried in other interaction frames/messages. For example, the interaction information before the first site sends the request already contains the manufacturer information.

Among them, the manufacturer information in the first request is the manufacturer associated with the first site, which can be called the first manufacturer. The first neural network association required by the first manufacturer and the first site.

Vendors can also be called AI suppliers. For example, the relationship between manufacturers and neural networks may include at least the following situations. It can be understood that the association between manufacturers and neural networks described here can be applied to other embodiments of the present application, and other embodiments will not be described in detail.

Case 1. The manufacturer corresponds to the equipment manufacturer, and the neural network is the neural network provided by the equipment manufacturer. For example, the first manufacturer is equipment manufacturer #1, and the first neural network is the neural network provided by equipment manufacturer #1 corresponding to the first site.

As an example, when the manufacturer corresponds to the equipment manufacturer, the manufacturer's identification can be indicated by the organization ID (organization identifier) in the standard. The organization ID is the globally unique identity information in the IEEE Registration Authority (IEEE Registration Authority) and is used to identify the manufacturer.

Case 2: The manufacturer corresponds to the chip manufacturer, and the neural network is the neural network provided by the chip manufacturer. For example, the first manufacturer is chip manufacturer #1, and the first neural network is the neural network provided by chip manufacturer #1.

Case 3: The manufacturer corresponds to the AI operator, and the neural network is provided by the AI operator. For example, the first manufacturer is a telecommunications operator, and the first neural network is a neural network provided by the telecommunications operator. It should be understood that the AI operator in this application generally refers to an operator that can provide AI-related services. It can be a telecommunications operator, such as China Mobile, China Unicom, China Telecom, etc., or it can also be other AI-related operators. , such as operators engaged in AI-related services, which are used to engage in services such as interoperability authentication of neural networks.

In an implementation manner, the association between the manufacturer and the site may at least include: the vendor refers to the vendor to which the site belongs, or the vendor refers to the vendor supported by the site. For example, the manufacturer refers to the manufacturer to which the site belongs. The manufacturer may refer to the equipment manufacturer, chip manufacturer or AI operator corresponding to the site. Or, for example, the manufacturer refers to the manufacturer supported by the site, that is, the manufacturer corresponding to the neural network supported by the site. In one implementation, different vendors can use the same neural network structure. For example, some vendors certify each other's neural network structures in the form of an alliance. In this case, the first request can carry the neural network structure supported by the first site. Manufacturer's information. For example, if the first site is a device of manufacturer #1 and supports the neural network of manufacturer #1, it also supports the neural network of manufacturer #2. Then the manufacturer information carried in the first request can be the identification of manufacturer #1, or it can It is manufacturer #2, and it can also carry the logos of manufacturer #1 and manufacturer #2.

Optionally, the first site may support one or more vendors' neural networks. Optionally, the first request may include a plurality of vendor information, and one or more of the plurality of vendor information is associated with the first neural network.

In an implementation manner, the first request may further include identification information of the first neural network, and the identification information of the first neural network may be used to distinguish different neural network models, or to distinguish different types of neural networks. For example, the identification information of the first neural network may include a model index (model index), a model identification (model ID), a model name, etc. In one implementation, one or more of the model index, model ID, or model name can uniquely identify any neural network. In another implementation, one or more of the model index, model ID, or model name may identify the category of the neural network. For example, a certain vendor has multiple neural network models, which are used for different tasks or functions. The tasks can be rate selection, channel access, channel state information compression, etc. In this case, model indexes can be used to distinguish these different categories of neural network models. For example, the model indexes of these neural networks are 1, 2, and 3 respectively. Model names can also be used to distinguish these neural networks with different tasks or functional categories. For example, the model names of these neural networks are rate selection neural network, channel access neural network, and channel state information compression neural network.

Optionally, the identification information of the first neural network may also include the version number (version) of the first neural network. For example, corresponding to the same model index, there will be different versions of the neural network, for example, version 1, version 2, etc. . At this point, the neural network can be identified by the model index and version number.

In this embodiment of the present application, the timing or triggering conditions for the first station to send the first request to the access point are not limited. For example, it may be that the first station moves to the BSS covered by the access point and when a network switch occurs, it sends the first request to the access point, or it may be that the first station ends the dormant state and sends the first request to the access point. , or it can be when the first site has a need for a new neural network, such as an updated neural network or a neural network with better accuracy, etc., it can send a first request to the access point, or it can be that the first site has learned that the access point When the entry point has a new or more suitable neural network or the first site periodically, the first request is sent to the access point.

Usually, if the first station is within the BSS covered by the access point, the first station sends the first request, which by default obtains the neural network corresponding to the BSS. In this case, the BSS identifier does not need to be carried. Optionally, the first site can also include the relevant identification of the target BSS in the first request, so that the access point corresponding to the target BSS can further confirm the corresponding neural network and send it to the first site, or it can facilitate other non-target BSSs. The access point is not responding or forwarding to the target access point, etc.

S320: The access point sends a first response to the first station, and accordingly, the first station receives the first response.

Wherein, the first response includes information of the first neural network. Information association between the first manufacturer and the first neural network.

For example, the information of the first neural network may include parameters of the first neural network and/or the structure of the first neural network.

The access point can send part of the information of the first neural network to the first site, for example, the parameters of the first neural network, and the first site can obtain the first neural network through the parameters of the first neural network; the access point can also send the first neural network to the first site. All the information of the first neural network is taken as a whole and all the information of the first neural network is sent to the first station.

The information of the first neural network may also be other forms of information related to the first neural network, or other forms of information used to reflect the calculation method of the first neural network, which is not limited in this application.

Wherein, the access point may send a first response to the first station according to the first request.

The access point may store a correspondence between the first manufacturer's information and the first neural network's information. For example, the access point searches for the neural network corresponding to the first manufacturer based on the identification information of the first manufacturer corresponding to the first request, and determines the information of the first neural network. Optionally, the access point may also search for the neural network corresponding to the first manufacturer based on the identification information of the first manufacturer and the identification information of the neural network included in the first request.

Based on the solution of the above embodiment, the first site can request the information of the neural network from the access point through the first request, and then the access point can search for the information of the neural network requested by the first site according to the first request, for example, the first The neural network information requested by the site is the first neural network information, and the first neural network information is sent to the first site. Since the neural network information is associated with the manufacturer information, the first site can obtain information from the neural network based on the manufacturer information. The access point obtains the information of the neural network. In this way, the site can obtain appropriate neural network information for communication decisions and ensure the communication performance of the site.

On the other hand, this method avoids the site from obtaining inappropriate neural network information from the cloud or server, and also prevents the site from spending a long time training the neural network, which helps reduce communication delays.

In addition, this method avoids the site from continuously training the neural network, which helps to reduce the power consumption of the site, thereby helping the site save energy.

Optionally, the method 300 further includes: the first station uses the information of the first neural network to make communication decisions.

For example, the first station can update the first neural network according to the parameters of the first neural network, and use the updated information of the first neural network to make communication decisions, such as channel access, rate adaptation, channel aggregation, and channel prediction. Decision-making on other communication tasks.

In an implementation scenario of the above embodiment, the first request in S310 may also include other information, which is used to further request the information of the first neural network that meets the requirements of these other information. This information may also be called first preset conditions, or matching conditions.

The first preset condition may be the generation time of the information of the first neural network, the accuracy of the first neural network, the model size of the first neural network, etc.

Specifically, the first preset condition may be the generation time of the information of the first neural network, or the generation time of the first neural network. For example, the first preset condition indicates that the generation time of the information of the first neural network should be After time point #A; for another example, the first preset condition indicates that the time difference between the generation time of the information of the first neural network and time point #B should be less than the preset value #A. In this case, the first request Used to request information about the first neural network that satisfies this generation time.

Specifically, the first preset condition may be the accuracy of the first neural network. For example, the first preset condition indicates that the accuracy of the first neural network should be greater than the preset value #B. In this case, the first request Information used to request the first neural network that meets this accuracy.

Specifically, the first preset condition may be the model size of the first neural network. For example, the first preset condition indicates that the model size of the first neural network should be smaller than the preset value #C, and the first request is used to request that the model size be satisfied. Information about the model size of the first neural network.

Optionally, in this implementation scenario, the method 300 further includes: the access point selects the information of the first neural network from the information of the multiple neural networks.

Specifically, the access point may maintain information for multiple neural networks. For example, the access point can store the correspondence between multiple neural network information, manufacturer information, model index, etc. In addition, it can also include generation time, accuracy, model size, etc. The access point may select the information of a neural network that satisfies the first preset condition from the information of multiple neural networks, for example, select the information of a neural network whose generation time, accuracy or model size satisfies the request. Taking the information of a neural network including the parameters and/or structure of the neural network as an example, "information of a neural network" is a set of information including the parameters and/or structure of the neural network.

In this application, the information of multiple neural networks can be understood as an information library or information set of information of neural networks. The access point can select the information of the first neural network required by the first site, that is, the target information of the first neural network.

Taking the information of the first neural network including parameters of the first neural network as an example, the access point can select parameters of multiple neural networks corresponding to the first manufacturer (or identification of the first manufacturer and the first neural network) ( (can also be called alternative parameters), select the parameters that meet the first preset condition as the parameters of the neural network requested by the first site, and carry the parameters, that is, the information of the first neural network, in the first response.

Based on the solution of the above implementation scenario, the access point can send the first neural network information that meets the preset conditions to the first station. In this way, the first station can obtain more suitable neural network information, which contributes to better Implement communication decisions.

In yet another implementation scenario of the above embodiment, the first neural network in the first response in S320 includes information of one or more neural networks. Optionally, the first response may also include attribute information corresponding to the one or more neural networks, such as one or more of generation time, accuracy, model size, etc. The generation time can be an absolute time, for example, time point #C, or the generation time can also be a relative time, for example, time difference #A, which represents the time difference between the generation time and time point #D, time point #D can be the default time reference point of the transceiver and receiver.

In this implementation scenario, the method 300 may further include: the first station selects the information of one neural network from the information of multiple neural networks.

For example, the first site can select one piece of information that satisfies its needs from the information of multiple neural networks and based on the attribute information of the multiple neural networks. For example, the first site selects information from multiple neural networks whose generation time or accuracy meets its own needs. Another example is that the first site selects information from multiple neural networks whose model size meets its own needs.

In any of the above implementation scenarios, the method 300 may also include: the access point obtains information of multiple neural networks. Wherein, the information of the plurality of neural networks obtained by the access point may include information of the first neural network sent to the first site.

In this way, the access point can obtain the neural network information in advance or in real time. When the first site requests the neural network from the access point, it can provide a neural network that meets the needs of the first site, or provide a more appropriate neural network to the first site. The network provides support for better communication decisions at the first site.

The access point can obtain the information of multiple neural networks when receiving a request from the first site, or it can obtain it according to certain preset conditions, such as a certain interval of time, or when there is a new demand. Types of sites are added, etc., which are not limited in the embodiments of this application.

Among them, the access point can obtain the information of the neural network in various ways, such as obtaining it from the second site, or obtaining it from the cloud or server, etc. specifically:

Method 1: The access point obtains neural network information from one or more second sites.

In one implementation manner, one or more second stations associated with the access point may send neural network information to the access point. One or more stations can send neural network information to the access point one or more times. The one or more second sites may be access points or non-AP sites.

Specifically, one or more second stations may, after receiving the second request from the access point, respond to the second request from the access point and send a second response to the access point, where the second response includes the information of the neural network. . Optionally, one or more second sites can also actively send neural network information to the access point, and one or more second sites can also send neural network information to the access point based on a certain time or predetermined rules. ,wait. The embodiments of the present application do not limit this. For a method for the access point to send a request to one or more second sites to obtain information about the neural network, please refer to the implementation shown in Figure 8 and its corresponding introduction, which will not be described again here. Here, the neural network that transmits one or more second stations to the access point is called a "second neural network". It should be understood that "first" and "second" in the embodiments of this application are only descriptive distinctions for ease of understanding and do not have any technical limitations.

Method 2: The access point obtains neural network information from the cloud or server.

Specifically, the cloud or server stores neural network information, such as parameters of the neural network, the structure of the neural network, and other information, and the corresponding manufacturer information. The access point can obtain the neural network information from the cloud or server.

Based on the above solution, since the cloud or server stores a large amount of neural network information, the access point can obtain multiple neural network information from the cloud or server. The neural network information is relatively comprehensive and can avoid multiple acquisitions, so it can reduce Small communication overhead, and provide suitable neural networks for different non-AP sites or APs.

In any of the above methods, the method 300 may further include: the access point may store information on multiple neural networks that satisfy the first site request, which may also be referred to as multiple candidate information. The access point may send multiple candidate information to the first station for selection by the first station, or may select one neural network information from the multiple candidate information and send it to the first station. Optionally, when there is only one neural network that satisfies the first site's request, the access point directly sends the information of the neural network, and there is no need to select.

Figure 8 is a schematic flow chart of a communication method 400 in a wireless local area network provided by an embodiment of the present application. Among them, taking the requesting site as an access point as an example, and taking the responding site as a non-AP site as an example, the responding site in method 400 is called the second site. It should be noted that the relevant solutions in this embodiment can also be applied to the embodiment shown in FIG. 7 or FIG. 9 , and the relevant content that has been described in detail in other embodiments will not be described again here. The method 400 may include the following steps.

S410: The access point sends a second request to the second station, and accordingly, the second station receives the second request.

The second request is used to request information about the neural network.

As described in the foregoing embodiments, the second request sent by the access point may include one or more of manufacturer information, identification information of the neural network, identification information of the basic service set, generation time, accuracy, model size, etc. Optionally, the second request sent by the access point may not include the above information to obtain information about all neural networks of the second site or information about the negotiated neural networks, etc. In this embodiment, the neural network requested by the access point from the site, or the neural network sent by the site to the access point, is called the second neural network, and the manufacturer associated with the second neural network is called the second manufacturer.

As mentioned above, the access point can send a second request to the second site when receiving a neural network request from a non-AP site, or it can send a second request to the second site based on predetermined rules or other requirements. The embodiments of the present application do not limit this. For example, the triggering conditions for the access point to send the second request to the second station may at least include the following situations:

Case 1: If the neural network information stored by the access point has not been updated for a predetermined time, the access point may send a second request to the second site to obtain updated neural network information.

Optionally, the neural network information stored by the access point may refer to the stored information of a certain neural network, such as the information of the second neural network. If the information of the second neural network has not been updated for a long time, the access point A second request is sent to the second site to obtain updated information about the neural network. In an implementation manner, the second request sent by the access point to the second site carries the identification information of the neural network, or the second request sent by the access point to the second site carries the identification information of the neural network and the corresponding manufacturer information.

Optionally, the neural network information of a certain manufacturer may not be updated, and the access point may send a second request to the second site of the corresponding manufacturer to obtain the latest neural network information of the manufacturer. In one implementation manner, the second request sent by the access point to the second site carries manufacturer information.

The neural network information stored by the access point has not been updated for more than a predetermined time. This can be based on the generation time of the neural network information and has not been updated for more than a predetermined time. It can also be based on the time that the access point has stored the neural network information. The scheduled time has not been updated.

In one implementation manner, the access point sends a second request to obtain neural network information to the second site, and may also carry the time or accuracy requirements of the neural network, so as to further obtain appropriate neural network information.

Case 2: If the accuracy of the stored neural network information is low, for example, less than a threshold, the access point sends a second request to the second station to obtain higher-precision neural network information.

For example, if the access point finds that the accuracy of the parameters of a certain manufacturer's neural network is lower than a threshold, the access point sends a second request to the second site of the manufacturer to request the manufacturer's higher accuracy neural network. parameter.

For another example, if the access point finds that the accuracy of the parameters of a specific neural network is lower than a threshold, the access point sends a second request to the second site to request higher-precision neural networks related to the specific neural network. Network parameters.

Case 3: When the access point does not store information about a certain manufacturer's neural network, the access point sends a second request to the second site to request information about the neural network associated with the manufacturer.

Optionally, the second request includes information about the manufacturer and is used to request information about the neural network of the manufacturer.

For example, when a non-AP station of the manufacturer joins the BSS to which the access point belongs, or sends information to the access point, but the access point does not store the information of the manufacturer's neural network, the access point can send information to the manufacturer. The non-AP site sends a second request to obtain the vendor's neural network information.

Optionally, the access point sends a second request to all or part of the non-AP sites in the BSS to which it belongs. If the second request is received, If the requested non-AP station belongs to the manufacturer or supports the neural network of the manufacturer or stores the neural network of the manufacturer, a second response will be sent to the access point, and the second response includes the information of the neural network of the manufacturer. If the non-AP station that receives the second request does not belong to the manufacturer or does not support the neural network of the manufacturer or does not store the neural network of the manufacturer, the second response is not sent to the access point.

Case 4: When the access point does not store information about a specific neural network, the access point sends a second request to the second site to request information about the neural network.

Optionally, the second request includes identification information of the second neural network requested by the access point.

In an optional implementation, the identification information of the second neural network can be used to identify the type of the neural network, or to identify the tasks that the neural network can perform, etc.

For example, if the access point finds that the neural network used to perform task #A is missing from its stored neural network information, the access point can send a second request to the second access point to request the neural network used to perform task #A. Neural network information. At this time, the second request includes identification information of the neural network used to perform task #A.

Case 5: When the access point does not store the information of the neural network, the access point sends a second request to the second site to request the information of the neural network.

As an implementation method of case 5, the second request is used to trigger the second site to report the information of the neural network it has trained. The second request may not include manufacturer information, or may not include time and/or accuracy requirements, neural network Any of the identifiers. That is to say, the second request does not need to be used to request information about a specific neural network. The second site can report information about all the neural networks it has trained according to the second request, and carry the manufacturer information, neural network information, etc. in the second response. One or more of the attribute information and the identity of the neural network.

For example, the access point sends a second request to all or part of the non-AP stations in the BSS to which it belongs, and the non-AP stations that receive the second request send the information of their trained neural network and the neural network to the access point. The network corresponding vendor, including the second station, sends a second response to the access point.

For example, if the second site belongs to vendor #A, which is a new vendor, and the access point does not have information about the neural network of vendor #A, the access point sends a second request to the second site for requesting Information about the neural network associated with vendor #A.

Optionally, as another implementation manner of case 5, the second request includes identification information of the BSS to which the access point belongs. That is, the second request may be used to request information about the neural network generated in the BSS to which the access point belongs.

Specifically, the second station may send the information of the neural network generated by the second station in the BSS to which the access point belongs to the access point according to the second request.

As an example, the identification information of the BSS may be the BSS ID.

In this case, the neural network information in the second response can be understood as the neural network information associated with the identification information of the BSS.

It should be noted that the example of the access point sending the second request to the second station in this embodiment can also be applied to other embodiments, such as the embodiments shown in Figures 7 and 9, and will not be described again.

S420: The second station sends a second response to the access point, and accordingly, the access point receives the second response.

Wherein, the second response includes the information of the second neural network, and the information of the second neural network is associated with the manufacturer information.

Specifically, after receiving the second request from the access point, one or more second stations may respond to the access point's request and send a second response to the access point, where the second response includes neural network information.

Optionally, one or more second sites can also actively send neural network information to the access point, and one or more second sites can also send neural network information to the access point based on a certain time or predetermined rules. ,wait. For example, after obtaining the information of the second neural network through training, the second station actively sends the information of the second neural network to the access point. For example, the information of the second neural network includes the parameters of the second neural network.

In addition, when the second station sends the information of the second neural network to the access point, it may also include manufacturer information, and the manufacturer information is associated with the second neural network. The second vendor may be a vendor to which the second site belongs, or the second vendor may be a vendor supported by the second site. Optionally, the manufacturer information in the second response includes the manufacturer's identification information. The manufacturer information may include one or more manufacturer information. Regarding the manufacturer information, the second neural network, and the correlation between the two, please refer to the relevant description of the manufacturer information, the first neural network, and their correlation in S310 above, which will not be described again here.

In one implementation, the information of the second neural network sent by the second station to the access point can be used for the information of the first neural network sent by the access point to the first station in the first response of S320. The second neural network The information of the network can be the same as the information of the first neural network, The second manufacturer information may also be the same as the first manufacturer information.

Exemplarily, the second response also includes attribute information corresponding to the second neural network, such as one or more of generation time, accuracy, model size, etc.

For example, the second response also includes identification information of the second neural network, which is used to identify a specific neural network or identify the type of neural network.

The information of the neural network in the second response may be the information of one or more neural networks. For example, the second response includes the information of the plurality of neural networks and the identification information of the BSS to which the second site belongs when generating the information of the plurality of neural networks. Optionally, the access point can save all the neural network information of the second response, or choose to save the neural network information related to the BSS to which it belongs.

Optionally, the second response includes attribute information of one or more neural networks, for example, attribute information such as generation time, accuracy, model size, etc. of one or more neural networks.

Based on the above solution, the access point can obtain the information of the neural network during information interaction with the second site. Since the access point and the second site can interact with information in real time, it has greater flexibility.

Optionally, the method 400 further includes: the access point storing the correspondence between the information of the second neural network and the manufacturer information.

Specifically, when acquiring the information of the second neural network, the access point can acquire the manufacturer information associated with the information of the second neural network, and the access point can store the correspondence between the neural network and the manufacturer information.

Among them, the corresponding relationship between the neural network and the manufacturer information can also be called the neural network-manufacturer table. The access point stores the corresponding relationship between the neural network and the manufacturer information, that is, the access point maintains the neural network-manufacturer table.

In addition, the neural network-manufacturer table may also include identification information, attribute information, etc. of the above-mentioned neural network. For example, in the neural network-manufacturer table, the information of the neural network may be the parameters and/or structure of the neural network, or other forms of related information of the neural network that can be obtained.

Based on the solution of the above embodiment, the access point can obtain the information of the second neural network from the second site. The information of the second neural network is associated with the manufacturer information, so that the access point can maintain the information of the neural network and the manufacturer information, thereby Provide support for other non-AP sites or APs to obtain more suitable neural network information, that is, provide support for better communication decisions for the site.

Figure 9 is a schematic flow chart of a communication method 500 in a wireless local area network provided by an embodiment of the present application. Method 500 may be a specific implementation based on the above-mentioned method 200, method 300, and method 400. The contents of the above-mentioned embodiments are applicable to method 500 and will not be described again here. Among them, the information of the neural network is explained by taking the parameters of the neural network as an example.

Assume that AP#1, STA#1, STA#2, and STA#3 belong to BSS#1.

S501: STA#1 (an example of the second site) and STA#2 (another example of the second site) train neural network model #1 (an example of the second neural network).

For example, STA#1 trains neural network model #1 in real time and obtains parameter #0 and parameter #1 of neural network model #1 successively.

Among them, parameter #0 is trained in BSS#0, parameter #1 is trained in BSS#1, and BSS#0 is the BSS to which STA#1 belongs before moving to BSS#1.

STA#2 trains neural network model #1 in real time and obtains parameter #2 of neural network model #1.

Among them, parameter #2 is trained in BSS#1.

Parameter #0, parameter #1, and parameter #2 are examples of parameters of the neural network respectively.

For example, the value of parameter #0 is weights#0, the value of parameter #1 is weights#1, and the value of parameter #2 is weights#2.

In addition, the equipment manufacturer of STA#1 and STA#2 is manufacturer #A (an example of a manufacturer).

S502, AP#1 (an example of the access point in method 300 and also an example of the access point in method 400) sends request #1 (second request example) to STA#1 and STA#2 to request neural parameters of the network model.

Among them, AP#1 belongs to BSS#1.

For example, when AP#1 determines that AP#1 does not have the model parameters of vendor #A (an example of a trigger condition for sending a request), AP#1 sends request #1 to the associated STA belonging to vendor #A within its BSS#1. .

For example, AP#1 sends request #1 to STA#1 and STA#2 respectively.

Optionally, request #1 includes information about the manufacturer, that is, manufacturer #A, indicating that information about the neural network associated with manufacturer #A is requested.

Optionally, request #1 includes identification information of the BSS where AP#1 is located, that is, BSS#1.

S503, STA#1 sends response #1 (an example of the second response) to AP#1.

Among them, response #1 includes the information of neural network model #1 generated in BSS#1, that is, parameter #1 of neural network model #1. Response #1 also includes the information of the manufacturer to which STA#1 belongs, that is, manufacturer #A. .

For example, response #1 may also include an identification of neural network model #1. For example, the identification of neural network model #1 is model #1.

For example, response #1 may also include the generation time of parameter #1 (an example of attribute information). For example, parameter #1 is generated at time #1.

For example, response #1 may also include the accuracy of parameter #1 (yet another example of attribute information). For example, the precision of parameter #1 is value #1.

S504, STA#2 sends response #2 (another example of the second response) to AP#1.

Similarly, response #2 includes the information of neural network model #1 generated in BSS#1, that is, parameter #2 of neural network model #1. Response #2 also includes the information of the manufacturer to which STA#2 belongs, that is, manufacturer #. A.

For example, response #2 may also include the identification of neural network model #1. For example, the identification of neural network model #1 is model #1.

For example, response #2 may also include the generation time of parameter #2. For example, parameter #2 is generated at time #2.

For example, response #2 may also include the accuracy of parameter #2. For example, the precision of parameter #2 is value #2.

S505, AP#1 stores the correspondence between the manufacturer information and the parameters of the neural network model #1.

For example, the corresponding relationship stored in AP#1 is as shown in Table 1. Table 1 includes the corresponding attribute information of the neural network parameters, such as generation time and accuracy. For example, the same model identifier in Table 1 can correspond to the parameters of multiple neural networks.

Table 1

S506: STA#3 (example of the first site) sends request #2 (example of the first request) to AP#1 to request parameters of the neural network model #1.

Among them, request #2 may include the identification of model #1.

Among them, request #2 may include information about manufacturer #A.

For example, the device manufacturer of STA #3 is manufacturer #A (example of manufacturer), so request #2 includes information of manufacturer #A (example of manufacturer information).

For another example, the equipment manufacturer of STA#3 is manufacturer #B, and STA#3 supports the neural network structure of manufacturer #A, so request #2 includes the information of manufacturer #A and manufacturer #B.

As an example, request #2 includes time information (an example of the first preset condition) for indicating the time requirement of STA #3 on the parameters of neural network model #1.

As yet another example, request #2 includes accuracy information (yet another example of the first preset condition) for indicating STA #3's accuracy requirements for the parameters of neural network model #1.

S507, AP#1 determines parameters (example of parameters of the first neural network).

AP#1 searches for the corresponding parameters in Table 1 based on the information of vendor #A and the identification of model #1 included in request #2. For example, AP#1 finds parameters #1 and #1 based on vendor #A and model #1. Parameter #2.

Corresponding to the above example, AP#1 selects parameter #1 that meets the time requirement from parameter #1 and parameter #2 based on the time information in request #2. At this time, weights#1 is the parameter selected by AP#1. .

Corresponding to another example above, AP#1 selects parameter #2 that meets the accuracy requirements from parameter #1 and parameter #2 based on the accuracy information in request #2. At this time, weights#2 is selected by AP#1. parameters.

S508, AP#1 sends response #3 (an example of the first response) to STA#3.

Among them, response #3 includes parameters selected by AP#1.

As another specific implementation manner, request #1 in S502 of method 500 does not include BSS#1, but response #1 includes BSS#0, parameter #0, and BSS#1, parameter #1. Response #2 includes BSS#1, parameter #2. Further, in S505, the AP chooses to store parameter #1, parameter #2 and their attribute information according to BSS#1 to which it belongs, while discarding parameter #0 generated in BSS#0.

As another specific implementation manner, in S507, the parameters determined by AP#1 include multiple parameters. AP#1 can send multiple parameters to STA#3, and STA#3 chooses which one to use for subsequent decisions. Or, AP#1 selects one of them and sends it to STA#3. For example, AP#1 can select one of the parameters in the stored order, or AP#1 can also select a parameter randomly.

As another specific implementation manner, the correspondence between the manufacturer information stored in AP#1 and the parameters of the neural network model #1 can be as shown in Table 2. The attribute information corresponding to the parameters of the neural network in Table 2 does not include time and accuracy information. Based on this, the above-mentioned requests and responses of the neural network may not include the generation time and accuracy. In S507, AP#1 can select the parameter as weights#1 according to the stored order, or AP#1 can randomly select one from parameter #1 and parameter #2.

Table 2

As another specific implementation manner, the correspondence between the manufacturer information stored in AP#1 and the parameters of the neural network model #1 can be as shown in Table 3. For the same model identifier, AP#1 can only store the parameters of one neural network. For example, the stored parameters of a neural network may be the latest received parameters of the neural network, or the parameters of the most accurate neural network, etc. This is not limited in the embodiment of the present invention.

table 3

It should be understood that model #2 in Table 1, Table 2, and Table 3 may be information of other neural networks that AP#1 has previously stored.

As another specific implementation manner, the correspondence between the manufacturer information stored in AP#1 and the parameters of the neural network model #1 can be as shown in Table 4. Table 4 shows examples of neural network information from multiple manufacturers. For example, BSS#1 may also include other non-AP stations, such as STA#4 and STA#5, and STA#4 and STA#5 are associated with AP#1. In S502, AP#1 may send request #1 to part or all of all STAs associated with BSS#1. In this case, in addition to receiving response #1 and response #2, AP#1 will also receive information from STA#4 and STA#5, including parameters of the neural network models trained by STA#4 and STA#5. For example, the equipment manufacturer of STA#4 is manufacturer #B in Table 4, and the equipment manufacturer of STA#5 is manufacturer #C in Table 4.

Table 4

For example, Vendor #B and Vendor #C also support Model #1. And STA#4 trains neural network model #1 and obtains neural network model #1 Parameter #3, the value of parameter #3 is weights#3, the generation time of parameter #3 is time #3, and the accuracy of parameter #3 is value #3. STA#5 trains neural network model #1 and obtains parameter #4 of neural network model #1. The value of parameter #4 is weights#4, the generation time of parameter #4 is time #4, and the accuracy of parameter #4 is value #4.

It should be understood that weights#1, weights#2, weights#3, and weights#4 in this application all represent the specific values of the parameters of the neural network. Weights can be the specific value of the weight of the neural network, or it can also be the specific value of the neural network. The specific values of weights and biases.

It can be understood that the above Tables 1 to 4 are some examples of the correspondence between AP#1's stored manufacturer information and the parameters of neural network model #1, but are not limited to the contents of Tables 1 to 4, and can also be other, The embodiment of the present invention does not limit this.

The method embodiments of the embodiments of the present application are described above, and the corresponding device embodiments are introduced below. It should be understood that the description of the device embodiments corresponds to the description of the method embodiments. Therefore, the parts not described in detail can be referred to the previous method embodiments.

Figure 10 is a schematic diagram of a communication device provided by an embodiment of the present application. As shown in Figure 10, the device 600 may include a transceiver unit 610 and/or a processing unit 620. The transceiver unit 610 can communicate with the outside, and the processing unit 620 is used to process data/information. The transceiver unit 610 may also be called a communication interface or a communication unit.

In a possible implementation, the device 600 may be the requesting site in the above method 200, the first site in the method 300, or the access point in the method 400, or may be used to implement the above method 200. A chip that functions as the requesting site, the first site in method 300, or the access point in method 400. Specifically, the device 600 can implement a process corresponding to the process performed by the requesting site in the above method 200, method 300 or method 400, wherein the transceiving unit 610 is used to perform operations related to the sending and receiving of the requesting site in the above method process.

Optionally, in this implementation, the device 600 further includes a processing unit 620, which is configured to perform operations related to processing of the requesting site in the above method flow.

Exemplarily, the transceiver unit 610 is used to send a request, the request is used to request the information of the neural network; the transceiver unit 610 is also used to receive a response from the responding site, the response includes the information of the neural network, the information of the neural network associated with the manufacturer information.

The information of the neural network may include parameters of the neural network and/or the structure of the neural network.

Among them, the manufacturer information may include multiple manufacturer information.

For example, the manufacturer information is the information of the manufacturer corresponding to the equipment manufacturer.

In one implementation, the request may include vendor information or identification information of the neural network.

Optionally, the request may also include the identification information of the BSS, and the response neural network information is associated with the identification information of the BSS.

Optionally, the request may also include a preset condition of the requested neural network, and the request is used to request information about a neural network that satisfies the preset condition. As an example, the preset condition includes at least one of the following: generation time of the neural network, accuracy of the neural network, and model size of the neural network.

In one implementation, the response may include vendor information or identification information of the neural network.

Optionally, the response may also include at least one of the following: generation time of the neural network, accuracy of the neural network, and model size of the neural network.

Optionally, the response may also include information from multiple neural networks. Further, the response may also include attribute information of multiple neural networks. As examples, the attribute information includes generation times of multiple neural networks, or accuracies of multiple neural networks, or model sizes of multiple neural networks.

The processing unit 620 may be configured to: select information of one neural network from information of multiple neural networks according to the attribute information.

In one implementation, the trigger conditions for the transceiver unit 610 to send a request include: the neural network information stored by the device 600 has not been updated for more than a preset time; or the accuracy of the neural network stored by the device 600 is less than a threshold; or , the device 600 does not store neural network information or does not store any neural network information; or the neural network information related to the manufacturer information stored by the device 600 has not been updated for more than a preset time; or the device 600 does not store any neural network information. Neural network information related to manufacturer information.

Optionally, the triggering conditions for requesting the station to send the request include: the device 600 wakes up after the station sleeps; or the network environment of the wireless local area network of the device 600 changes.

It should be understood that the above content is only taken as an example, and the device 600 can also implement other steps, actions or methods related to requesting the site in the above method 200, 300 or 400, which will not be described again here.

In a possible implementation, the device 600 may be the response site in the above method 200, the access point in the method 300, or the second site in the method 400, or may be used to implement the above method 200. A chip that functions as a responsive station, an access point in method 300, or a second station in method 400. Specifically, the device 600 can implement the response corresponding to the above method 200, 300 or 400. The process executed by the response site, wherein the transceiver unit 610 is used to perform operations related to the transceiver of the response site in the above method process.

Optionally, in this implementation, the device 600 further includes a processing unit 620, which is configured to perform operations related to processing of the response site in the above method flow.

Exemplarily, the transceiver unit 610 is configured to receive a request from the requesting site, where the request is used to request information about the neural network; the transceiving unit 610 is also configured to send a response to the requesting site according to the request, where the response includes the information about the neural network. The information of the neural network is associated with the manufacturer's information.

The information of the neural network may include parameters of the neural network and/or the structure of the neural network. The manufacturer information may include one or more manufacturer information. For example, the manufacturer information is the information of the manufacturer corresponding to the equipment manufacturer.

In one implementation, the request may include vendor information or identification information of the neural network.

Optionally, the request may also include the identification information of the BSS, and the response neural network information is associated with the identification information of the BSS.

Optionally, the request may also include a preset condition of the requested neural network, and the request is used to request information of a neural network that satisfies the preset condition.

As an example, the preset condition includes at least one of the following: generation time of the neural network, accuracy of the neural network, and model size of the neural network.

The processing unit 620 may be used to select information of one neural network from information of multiple neural networks according to preset conditions.

In one implementation, the response may include vendor information or identification information of the neural network.

Optionally, the response may also include at least one of the following: generation time of the neural network, accuracy of the neural network, and model size of the neural network.

Optionally, the response may also include information from multiple neural networks.

Optionally, the response may also include attribute information of multiple neural networks.

As an example, the attribute information includes the generation time of multiple neural networks, or the accuracy of multiple neural networks, or the model sizes of multiple neural networks.

In one implementation, the trigger conditions for the transceiver unit 610 to send a request include: the neural network information stored by the device 600 has not been updated for more than a preset time; or the accuracy of the neural network stored by the device 600 is less than a threshold; or , the device 600 does not store neural network information or does not store any neural network information; or the neural network information related to the manufacturer information stored by the device 600 has not been updated for more than a preset time; or the device 600 does not store any neural network information. Neural network information related to manufacturer information.

In one implementation, the triggering conditions for the transceiver unit 610 to send a request include: the device 600 wakes up after sleeping; or the network environment of the wireless local area network of the device 600 changes.

It should be understood that the above content is only taken as an example, and the device 600 can also implement other steps, actions or methods related to responding to the site in the above method 200, 300 or 400, which will not be described again here.

It should be understood that the device 600 here is embodied in the form of a functional unit. The term "unit" as used herein may refer to an application specific integrated circuit (ASIC), an electronic circuit, a processor (such as a shared processor, a proprietary processor, or a group of processors) used to execute one or more software or firmware programs. processor, etc.) and memory, merged logic circuitry, and/or other suitable components to support the described functionality.

The above device 600 has the function of realizing the corresponding steps performed by the requesting site in the above method 200, 300 or 400, or the above device 600 has the function of realizing the corresponding steps performed by the responding site in the above method 200, 300 or 400. The functions described can be implemented by hardware, or can be implemented by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions; for example, the transceiver unit can be replaced by a transceiver (for example, the sending unit in the transceiver unit can be replaced by a transmitter, and the receiving unit in the transceiver unit can be replaced by a receiving unit. (machine replacement), other units, such as processing units, etc., can be replaced by processors to respectively perform the sending and receiving operations and related processing operations in each method embodiment.

In addition, the above-mentioned transceiver unit may also be a transceiver circuit (for example, it may include a receiving circuit and a transmitting circuit), and the processing unit may be a processing circuit. In this embodiment of the present application, the device in Figure 10 can be the request site or the response site in the previous embodiment, or it can be a chip or a chip system, such as a system on chip (SoC). The transceiver unit may be an input-output circuit or a communication interface. The processing unit is a processor or microprocessor or integrated circuit integrated on the chip. No limitation is made here.

Figure 11 is another schematic structural diagram of a communication device provided by an embodiment of the present application. As shown in FIG. 11 , the communication device 700 includes: at least one processor 710 and a transceiver 720 . The processor 710 is coupled to the memory and is used to execute instructions stored in the memory to control the transceiver 720 to send signals and/or receive signals. Optionally, the communication device 700 further includes a memory 730 for storing instructions.

It should be understood that the above-mentioned processor 710 and the memory 730 can be combined into one processing device, and the processor 710 is used to execute the program code stored in the memory 730 to implement the above functions. During specific implementation, the memory 730 may also be integrated in the processor 710 or independent of the processor 710 .

It should also be understood that the transceiver 720 may include a receiver and a transmitter. The transceiver 720 may further include an antenna, and the number of antennas may be one or more. The transceiver 1020 may be a communication interface or an interface circuit.

When the communication device 700 is a chip, the chip includes a transceiver unit and a processing unit. The transceiver unit may be an input-output circuit or a communication interface; the processing unit may be a processor, microprocessor, or integrated circuit integrated on the chip.

An embodiment of the present application also provides a processing device, including a processor and an interface. The processor may be used to execute the method in the above method embodiment.

It should be understood that the above processing device may be a chip. For example, the processing device may be a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or a system on chip (SoC), or It can be a central processing unit (CPU), a network processor (NP), a digital signal processing circuit (DSP), or a microcontroller unit , MCU), it can also be a programmable logic device (PLD) or other integrated chip.

During the implementation process, each step of the above method can be completed by instructions in the form of hardware integrated logic circuits or software in the processor. The steps of the methods disclosed in conjunction with the embodiments of the present application can be directly implemented by a hardware processor for execution, or can be executed by a combination of hardware and software modules in the processor. The software module can be located in random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers and other mature storage media in this field. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it will not be described in detail here.

Figure 12 is another schematic structural diagram of a communication device provided by an embodiment of the present application. As shown in Figure 12, the device 800 includes a processing circuit 810 and a transceiver circuit 820. The processing circuit 810 and the transceiver circuit 820 communicate with each other through internal connection paths. The processing circuit 810 is used to execute instructions to control the transceiver circuit 820 to send signals and/or receive signals.

Optionally, the device 800 may also include a storage medium 830, which communicates with the processing circuit 810 and the transceiver circuit 820 through internal connection paths. The storage medium 830 is used to store instructions, and the processing circuit 810 can execute the instructions stored in the storage medium 830 .

In a possible implementation, the device 800 is configured to implement the process corresponding to the requesting site in the above method embodiment.

In another possible implementation, the device 800 is configured to implement the process corresponding to the response site in the above method embodiment.

According to the method provided by the embodiment of the present application, the present application also provides a computer program product. The computer program product includes: computer program code. When the computer program code is run on a computer, it causes the computer to execute the embodiment shown in Figure 3 method in.

According to the method provided by the embodiment of the present application, the present application also provides a computer-readable medium. The computer-readable medium stores program code. When the program code is run on a computer, it causes the computer to execute the steps in the above method embodiment. method.

According to the method provided by the embodiments of this application, this application also provides a system, which includes the aforementioned request site and response site.

The term "at least one of..." or "at least one of..." as used herein means all or any combination of the listed items, for example, "at least one of A, B and C", It can mean: A exists alone, B exists alone, C exists alone, A and B exist simultaneously, B and C exist simultaneously, and A, B and C exist simultaneously. "At least one" in this article means one or more. "Multiple" means two or more.

The term "and/or" in this article is just an association relationship that describes related objects, indicating that three relationships can exist. For example, A and/or B can mean: A exists alone, A and B exist simultaneously, and they exist alone. B these three situations. In addition, the character "/" in this article generally indicates that the related objects are an "or" relationship.

It should be understood that in various embodiments of the present application, "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B only based on A. B can also be determined based on A and/or other information. The terms “including,” “includes,” “having,” and variations thereof all mean “including but not limited to,” unless otherwise specifically emphasized.

It should be understood that in the various embodiments of the present application, the first, second and various numerical numbers are only used for convenience of description and are not used to limit the scope of the embodiments of the present application. For example, distinguish different information, etc.

Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may implement the described functionality using different methods for each specific application, but such implementations should not be considered beyond the scope of this application.

Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be described again here.

In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or can be integrated into another system, or some features can be ignored, or not implemented. On the other hand, the coupling or direct coupling or communication connection between each other shown or discussed may be through some interfaces, and the indirect coupling or communication connection of the devices or units may be in electrical, mechanical or other forms.

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

In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, each unit can exist physically alone, or two or more units can be integrated into one unit.

If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the existing technology or the part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk and other media that can store program code. .

The above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application. should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims (34)

1. A communication method in a wireless local area network, characterized by including:

   The requesting site sends a request for requesting information from the neural network;

   The requesting site receives a response from the responding site, where the response includes the requested neural network information, and the neural network information is associated with manufacturer information.
2. The method of claim 1, wherein the request includes the manufacturer information, or the request includes identification information of the neural network.
3. The method of claim 1 or 2, wherein the response includes the manufacturer information.
4. The method according to any one of claims 1-3, characterized in that the manufacturer information includes multiple manufacturer information.
5. The method according to any one of claims 1 to 4, characterized in that the manufacturer information is the information of the manufacturer corresponding to the equipment manufacturer.
6. The method according to any one of claims 1 to 5, wherein the response includes identification information of the neural network.
7. The method according to any one of claims 1 to 6, characterized in that the information of the neural network includes parameters of the neural network and/or the structure of the neural network.
8. The method according to any one of claims 1 to 7, wherein the request includes identification information of a basic service set (BSS), and the information of the neural network in the response is associated with the identification information of the BSS.
9. The method according to any one of claims 1 to 8, wherein the request includes at least one of the following conditions of the requested neural network: generation time of the neural network, accuracy of the neural network, model of the neural network size.
10. The method according to any one of claims 1 to 9, characterized in that the response further includes at least one of the following: the generation time of the neural network, the accuracy of the neural network and the model of the neural network size.
11. The method according to any one of claims 1 to 10, wherein the response includes information from a plurality of neural networks.
12. The method of claim 11, wherein the response further includes the generation time of the plurality of neural networks; or,

    the accuracy of the plurality of neural networks; or,

    The model size of the multiple neural networks.
13. The method according to any one of claims 1 to 12, characterized in that the triggering conditions for the requesting site to send a request include:

    The neural network information stored by the requesting site has not been updated for more than the preset time; or,

    The accuracy of the neural network stored by the requesting site is less than a threshold; or,

    The requesting site does not store information about the neural network or does not store information about any neural network; or,

    The neural network information related to the manufacturer information stored by the requesting site has not been updated for more than a preset time; or,

    The requesting site does not store information about the neural network related to the vendor information.
14. The method according to any one of claims 1 to 13, characterized in that the triggering conditions for the requesting site to send a request include:

    The requesting site wakes up after sleeping; or,

    The network environment of the wireless local area network of the requesting site changes.
15. A communication method in a wireless local area network, characterized by including:

    The responding site receives a request from the requesting site, the request being used to request information from the neural network;

    The responding site sends a response to the requesting site according to the request, where the response includes information about the requested neural network, and the information about the neural network is associated with manufacturer information.
16. The method of claim 15, wherein the request includes the manufacturer information, or the request includes identification information of the neural network.
17. The method of claim 15 or 16, wherein the response includes the manufacturer information.
18. The method according to any one of claims 15 to 17, characterized in that the manufacturer information includes multiple manufacturer information.
19. The method according to any one of claims 15 to 18, characterized in that the manufacturer information is the information of the manufacturer corresponding to the equipment manufacturer.
20. The method according to any one of claims 15 to 19, wherein the response includes identification information of the neural network.
21. The method according to any one of claims 15 to 20, characterized in that the information of the neural network includes the neural network parameters and/or structure of the neural network.
22. The method according to any one of claims 15 to 21, characterized in that the request includes identification information of a basic service set (BSS), and the information of the neural network in the response is associated with the identification information of the BSS.
23. The method according to any one of claims 15 to 22, wherein the request includes at least one of the following conditions for the requested neural network information: generation time of the neural network, accuracy of the neural network, neural network model size.
24. The method according to any one of claims 15-23, characterized in that the response further includes at least one of the following: the generation time of the neural network, the accuracy of the neural network and the accuracy of the neural network. Model size.
25. The method according to any one of claims 15 to 24, wherein the response includes information from multiple neural networks.
26. The method of claim 25, wherein the response further includes the generation time of the plurality of neural networks; or,

    the accuracy of the plurality of neural networks; or,

    The model size of the multiple neural networks.
27. The method according to any one of claims 15 to 26, characterized in that the requesting site sends trigger conditions for the request, including:

    The neural network information stored by the requesting site has not been updated for more than the preset time; or,

    The accuracy of the neural network stored by the requesting site is less than a threshold; or,

    The requesting site does not store information about the neural network or does not store information about any neural network; or

    The neural network information related to the manufacturer information stored by the requesting site has not been updated for more than a preset time; or,

    The requesting site does not store information about the neural network related to the manufacturer information.
28. The method according to any one of claims 15 to 27, characterized in that the requesting site sends trigger conditions for the request, including:

    The requesting site wakes up after sleeping; or,

    The network environment of the wireless local area network of the requesting site changes.
29. A communication device, characterized by comprising: a unit or module for implementing the method as described in any one of claims 1 to 14, or for implementing the method as described in any one of claims 15 to 28 The unit or module of the method.
30. A communication device, characterized by including:

    Memory, used to store computer instructions;

    A processor, configured to execute computer instructions stored in the memory, causing the communication device to perform the method as claimed in any one of claims 1 to 14, or to cause the communication device to perform the method as claimed in any one of claims 15 to 28. any of the methods described.
31. A chip, characterized in that it includes: a processor and an interface for calling and running a computer program stored in the memory to execute the method according to any one of claims 1 to 14, or A method as claimed in any one of claims 15 to 28 is performed.
32. A computer-readable storage medium, characterized in that it is used to store a computer program, the computer program includes instructions for implementing the method as claimed in any one of claims 1 to 14, or includes instructions for implementing the method as claimed in any one of claims 1 to 14. Instructions for the method of any one of claims 15 to 28.
33. A computer program product, characterized in that it includes computer program code. When the computer program code is run on a computer, it causes the computer to implement the method as claimed in any one of claims 1 to 14, or to implement the method as claimed in claim 1. The method described in any one of 15 to 28.
34. A communication system, characterized in that it includes a requesting site for implementing the method as described in any one of claims 1 to 14, and a response site for implementing the method as described in any one of claims 15 to 28 site.