Wireless signal recognition method, system, medium and equipment based on UAV collaboration

Abstract

The present application provides a wireless signal recognition method, system, medium and device based on drone collaboration, which relates to the field of drone control technology and is applied to a master drone in a drone swarm. The master drone first performs signal recognition based on its own main edge model and obtains a first confidence level. If the confidence level is low, the collaborative slave drones in the drone swarm are used to recognize the signal and obtain a second confidence level. The recognition results and confidence levels of multiple drones are combined to determine the recognition result of the wireless signal, so as to utilize the mutual assistance advantage of the drone swarm and effectively improve the wireless signal recognition accuracy of a single drone, thereby reducing the storage and transmission requirements of the drone. At the same time, edge computing can be used to significantly reduce the computing power of the host or background server, thereby improving the working efficiency of the drone swarm.

Description

Technical Field

[0001] The present application relates to the field of drone control technology, and specifically to a wireless signal recognition method, system, medium, and device based on drone collaboration. Background Art

[0002] With the development of technologies such as artificial intelligence and networked communications, drones (UAVs) have found widespread application in many fields thanks to their low cost, large scale, and high degree of autonomy. Using UAVs for inspections, aerial photography, and rescue operations has become a widely adopted intelligent operation method. However, many complex application scenarios require the coordinated operation of multiple UAVs. Drone swarms can break down tasks into multiple subtasks, each of which can be completed collaboratively by different drones. This not only improves work efficiency, but also reduces operating costs and optimizes resource allocation.

[0003] Drone swarming is a system that uses multiple drones to inspect or monitor a target area. This system allows for rapid and accurate collection of relevant information about the target area. However, due to the complexity of the inspection environment, drones need to detect complex wireless signals. Transmitting all wireless signals back to a backend server would not only take a long time for data transmission and signal recognition, but would also require the drones to have significant storage space and transmission bandwidth. Currently, most drones use time and frequency domain recognition to identify wireless signals collected by a single drone to obtain results for that individual drone. Therefore, a wireless signal recognition method applicable to drone swarms is urgently needed. Summary of the Invention

[0004] In order to solve the above technical problems, this application proposes a wireless signal recognition method, system, medium and equipment based on drone collaboration.

[0005] According to one aspect of the present application, a wireless signal recognition method based on drone collaboration is provided, which is applied to a master drone in a drone swarm, the drone swarm including the master drone and at least one collaborative slave drone. The wireless signal recognition method based on drone collaboration includes: receiving a wireless signal; identifying a first signal type and a corresponding first confidence level of the wireless signal based on a master edge model carried on the master drone; wherein the master edge model is obtained by knowledge distillation from a signal recognition model trained by a server, and the first confidence level is determined by the confidence level of the master edge model and the probability of the first signal type appearing in the drone swarm; if the first confidence level is less than a preset confidence level threshold, obtaining a second signal type and a corresponding second confidence level of the wireless signal identified by a collaborative edge model carried on collaborative slave drones within a preset range of the master drone; wherein the collaborative edge model is obtained by knowledge distillation from the signal recognition model, and the second confidence level is the second confidence level of the collaborative edge model; based on the first confidence level and the second confidence level, comprehensively calculating a comprehensive confidence level of the first signal type; and if the comprehensive confidence level is greater than the confidence level threshold, storing the wireless signal.

[0006] In one embodiment, the comprehensive confidence of the first signal type is calculated based on the first confidence level and the second confidence level, including: obtaining a first number of cooperative slave drones within a preset range of the master drone; counting a second number of second signal types that are the same as the first signal type identified by the cooperative slave drones within the preset range of the master drone; and calculating the comprehensive confidence of the first signal type based on the first number, the second number, the second confidence level and the first confidence level.

[0007] In one embodiment, the calculation of the comprehensive confidence of the first signal type based on the first quantity, the second quantity, the second confidence and the first confidence includes: calculating the ratio of the second quantity to the first quantity to obtain the cooperation ratio of the first signal type; calculating the cooperation confidence of the cooperative slave drone based on the cooperation ratio and the second confidence; and calculating the comprehensive confidence of the first signal type based on the cooperation confidence and the first confidence.

[0008] In one embodiment, the calculating of the cooperation confidence of the cooperative slave drone based on the cooperation ratio and the second confidence includes: the calculation formula of the cooperation confidence is:

[0009] ;in, is the collaborative confidence, is the collaborative ratio, is a second confidence level of the cooperative slave UAV corresponding to a second signal type that is the same as the first signal type, The second confidence level of the coordinated slave drone within the preset range of the master drone.

[0010] In one embodiment, the calculating of the comprehensive confidence of the first signal type based on the collaborative confidence and the first confidence includes: performing a weighted average of the collaborative confidence and the first confidence to obtain the comprehensive confidence.

[0011] In one embodiment, the wireless signal recognition method based on drone collaboration further includes: respectively calculating the mean square error between the recognition results of the wireless signal by each drone and the collaborative slave drone within a preset range of the master drone; selecting the drone with the smallest mean square error as the target drone, and using the target drone to store the wireless signal.

[0012] In one embodiment, a knowledge distillation method for an edge model carried on a drone includes: closing some network channels of the signal recognition model and calculating a loss function value; if the loss function value is less than a preset value, continuing to close some network channels until the loss function value is greater than or equal to the preset value; deleting the closed network channels to obtain the edge model; and calculating the confidence of the edge model based on the loss function value of the edge model and the preset value.

[0013] According to another aspect of the present application, a wireless signal recognition system based on drone collaboration is provided, which is a master drone set in a drone group, and the drone group includes the master drone and at least one collaborative slave drone; the wireless signal recognition system based on drone collaboration includes: a wireless signal receiving module for receiving wireless signals; a preliminary signal type recognition module for identifying the first signal type and the corresponding first confidence of the wireless signal based on the main edge model carried by the main drone; wherein the main edge model is a signal recognition model obtained by server training through knowledge distillation, and the first confidence is determined by the confidence of the main edge model and the probability of the first signal type appearing in the drone group. ; A collaborative signal acquisition module, for obtaining the second signal type and corresponding second confidence of the wireless signal identified by the collaborative edge model carried by the collaborative slave UAV within the preset range of the master UAV if the first confidence is less than the preset confidence threshold; wherein the collaborative edge model is obtained by the signal recognition model through knowledge distillation, and the second confidence is the second confidence of the collaborative edge model; a collaborative signal type determination module, for comprehensively calculating the comprehensive confidence of the first signal type based on the second confidence of the second signal type that is the same as the first signal type and the first confidence; a wireless signal storage module, for storing the wireless signal if the comprehensive confidence is greater than the confidence threshold.

[0014] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and the computer program is used to execute any of the above methods.

[0015] According to another aspect of the present application, an electronic device is provided, comprising: a processor; a memory for storing instructions executable by the processor; and the processor for executing any of the above methods.

[0016] Beneficial effects: The wireless signal recognition method, system, medium and device based on drone collaboration provided by the present application are applied to the master drone in a drone group, and the drone group includes a master drone and at least one collaborative slave drone; by receiving a wireless signal; based on a main edge model carried on the master drone, a first signal type of the wireless signal and a corresponding first confidence level are identified; wherein the main edge model is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is determined by the confidence level of the main edge model and the probability of the first signal type appearing in the drone group; if the first confidence level is less than a preset confidence level threshold, the second signal type and the corresponding second confidence level of the wireless signal obtained by the collaborative edge model carried on the collaborative slave drone within the preset range of the master drone are obtained; wherein the collaborative ... The model is obtained through knowledge distillation, and the second confidence is the second confidence of the collaborative edge model; based on the first confidence and the second confidence, the comprehensive confidence of the first signal type is obtained by comprehensive calculation; if the comprehensive confidence is greater than the confidence threshold, the wireless signal is stored; that is, the master drone first performs signal recognition based on its own main edge model and obtains the first confidence. If the confidence is low, the collaborative slave drone in the drone group is used to identify the signal and obtain the second confidence. The recognition results and confidence of multiple drones are combined to determine the recognition result of the wireless signal, so as to utilize the mutual assistance advantages of the drone group and effectively improve the wireless signal recognition accuracy of a single drone, thereby reducing the storage and transmission requirements of the drone. At the same time, edge computing can be used to greatly reduce the computing power of the host or acquired server, thereby improving the work efficiency of the drone group. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1 This is a flowchart of a wireless signal recognition method based on drone collaboration provided by an exemplary embodiment of the present application.

[0019] Figure 2 It is a structural diagram of a wireless signal recognition system based on drone collaboration provided by an exemplary embodiment of the present application.

[0020] Figure 3 It is a structural diagram of an electronic device provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0021] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0022] Figure 1 This is a flow chart of a wireless signal recognition method based on UAV collaboration provided by an exemplary embodiment of the present application. The wireless signal recognition method based on UAV collaboration is applied to a master UAV in a UAV swarm, wherein the UAV swarm includes a master UAV and at least one cooperative slave UAV; Figure 1 As shown, the wireless signal recognition method based on UAV collaboration includes the following steps:

[0023] Step 110: Receive a wireless signal.

[0024] Among them, wireless signals include: wireless signals with frequencies ranging from 100MHz to 10GHz in different frequency bands for different types of drones, such as wireless signals with different modulation methods such as frequency modulation (FM), amplitude modulation (AM), phase modulation (PM), double sideband (DSB), single sideband (SSB), orthogonal frequency division multiplexing (OFDM), etc. At the same time, they also include communication, radar, satellite and other wireless signals that overlap with the commonly used frequency bands of drones. During the inspection process of the drone swarm, each drone will conduct inspections according to the pre-set inspection route and collect and receive wireless signals in real time. For the wireless signals collected by the drones, this application performs edge recognition on them to reduce the amount of data transmission.

[0025] Step 120: Based on the master edge model carried by the master UAV, a first signal type and a corresponding first confidence level of the wireless signal are identified.

[0026] Among them, the main edge model is obtained by knowledge distillation from the signal recognition model obtained by server training, and the first confidence is determined by the confidence of the main edge model and the probability of the first signal type appearing in the drone group. The present application obtains a signal recognition model through server training, and obtains an edge model with reduced complexity and computational complexity through a knowledge distillation algorithm. The edge model is mounted on a drone, and the drone performs preliminary analysis and recognition of the acquired wireless signal to reduce the data transmission volume of the drone group and the recognition data volume of the server. Since the recognition accuracy of the edge model obtained by knowledge distillation is lower than the accuracy of the signal recognition model mounted on the server, the present application simultaneously determines its confidence when recognizing the wireless signal to provide a judgment basis for subsequent accurate signal recognition.

[0027] Step 130: If the first confidence level is less than a preset confidence level threshold, a second signal type and a corresponding second confidence level of the wireless signal identified by the collaborative edge model carried by the collaborative slave UAV within the preset range of the master UAV are obtained.

[0028] The collaborative edge model is obtained by knowledge distillation of the signal recognition model, and the second confidence level is the second confidence level of the collaborative edge model. When the confidence level (first confidence level) of the master drone in identifying a wireless signal is low, the present application utilizes the collaborative slave drones in its vicinity (within a preset range) to identify the wireless signal, wherein the wireless signal can be collected and identified by the collaborative slave drones themselves, or sent by the master drone to the collaborative slave drones to determine the target signal to be identified. It should be understood that if the first confidence level is greater than or equal to the confidence level threshold, it means that the master drone's recognition accuracy is high, and the master drone's recognition result can be directly adopted at this time.

[0029] Step 140: Based on the first confidence level and the second confidence level, a comprehensive confidence level of the first signal type is obtained by comprehensive calculation.

[0030] After the collaborative slave drone identifies the signal type and confidence level of the wireless signal, the master drone's signal type and corresponding confidence level are combined to calculate a comprehensive confidence level that the wireless signal is of the first signal type. Preferably, if the master and collaborative slave drones identify different signal types for the same wireless signal, indicating that the master drone's recognition result for the wireless signal is inaccurate (because there are typically multiple collaborative slave drones, this result is likely due to a malfunction in the master drone, such as an edge model recognition anomaly), the master drone can abandon recognition of the wireless signal, allowing the collaborative slave drone to perform recognition. After completing this mission, the master drone undergoes maintenance, including hardware and software overhauls. Software overhauls can be performed by re-distilling the server's signal recognition model knowledge to obtain the master drone's edge model, or by retraining the master drone using wireless signals that the master drone cannot recognize and accurate signal types identified based on server recognition to improve its recognition accuracy.

[0031] Step 150: If the comprehensive confidence is greater than the confidence threshold, the wireless signal is stored.

[0032] When the combined confidence level calculated based on the identification results of multiple drones and the corresponding confidence levels indicates that the wireless signal is of the first type and is high, the signal type of the wireless signal can be determined, and the relevant data of the wireless signal is stored. If the calculated combined confidence level is low, the signal type of the wireless signal cannot be accurately confirmed. In this case, the wireless signal can be transmitted back to the server, which will re-identify the wireless signal to ensure the accuracy of wireless signal identification. Preferably, before transmission, the storage status and operating status of the multiple drones for which the combined confidence level is obtained are first determined, and drones with an idle operating status are preferentially selected to transmit the wireless signal back to ensure the normal operation of other drones. If all drones are in an operating state, the drone with the least storage capacity is selected to transmit the wireless signal back, or the drone closest to the server and whose storage capacity does not exceed a preset value is selected to transmit the wireless signal back.

[0033] The wireless signal recognition method based on drone collaboration provided by the present application is applied to a master drone in a drone group, and the drone group includes a master drone and at least one collaborative slave drone; by receiving a wireless signal; based on a master edge model carried on the master drone, a first signal type of the wireless signal and a corresponding first confidence level are identified; wherein the master edge model is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is determined by the confidence level of the master edge model and the probability of the first signal type appearing in the drone group; if the first confidence level is less than a preset confidence level threshold, the second signal type of the wireless signal and the corresponding second confidence level are obtained by the collaborative edge model carried on the collaborative slave drone within the preset range of the master drone; wherein the collaborative edge model is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is determined by the confidence level of the master edge model and the probability of the first signal type appearing in the drone group; if the first confidence level is less than a preset confidence level threshold, the second signal type of the wireless signal and the corresponding second confidence level are obtained by the collaborative edge model carried on the collaborative slave drone within the preset range of the master drone; wherein the collaborative edge model is a signal recognition model obtained by server training obtained by knowledge distillation The second confidence level is the second confidence level of the collaborative edge model; based on the first confidence level and the second confidence level, the comprehensive confidence level of the first signal type is obtained by comprehensive calculation; if the comprehensive confidence level is greater than the confidence level threshold, the wireless signal is stored; that is, the master UAV first performs signal recognition based on its own main edge model and obtains the first confidence level. If the confidence level is low, the collaborative slave UAV in the UAV group is used to recognize the signal and obtain the second confidence level. The recognition result of the wireless signal is determined by combining the recognition results and confidence levels of multiple UAVs, so as to utilize the mutual assistance advantages of the UAV group and effectively improve the wireless signal recognition accuracy of a single UAV, thereby reducing the storage and transmission requirements of the UAV. At the same time, edge computing can be used to greatly reduce the computing power of the host or acquired server, thereby improving the working efficiency of the UAV group.

[0034] In one embodiment, the specific implementation method of the above-mentioned step 140 can be: obtaining a first number of cooperative slave drones within a preset range of the master drone; counting a second number of second signal types that are the same as the first signal type identified based on the cooperative slave drones within the preset range of the master drone; and calculating a comprehensive confidence level of the first signal type based on the first number, the second number, the second confidence level and the first confidence level.

[0035] This application determines the number of collaborative slave drones near the master drone to count the number of drones among these collaborative slave drones that obtain the same identification results as the master drone, that is, determines the number of drones near the master drone and the number of drones among them whose identification results are the same as the master drone, and combines the corresponding recognition result confidence levels to comprehensively calculate the comprehensive confidence level of the first signal type (that is, the recognition result of the master drone).

[0036] In one embodiment, the specific implementation method of the above-mentioned step 140 can be: calculating the ratio of the second number to the first number to obtain the cooperation ratio of the first signal type; based on the cooperation ratio and the second confidence level, calculating the cooperation confidence level of the cooperative slave drone; based on the cooperation confidence level and the first confidence level, calculating the comprehensive confidence level of the first signal type.

[0037] This application calculates the proportion of drones with the same identification results as the main drone near the main drone, and calculates the collaborative confidence of the collaborative slave drone in combination with the confidence of the collaborative slave drone with the same identification results as the main drone, and then combines the collaborative confidence and the first confidence to calculate the comprehensive confidence of the first signal type.

[0038] In one embodiment, the specific implementation of step 140 may be: the calculation formula of the collaboration confidence is:

[0039] ;in, is the collaborative confidence, is the collaborative ratio, is a second confidence level of the cooperative slave UAV corresponding to a second signal type that is the same as the first signal type, The second confidence level of the coordinated slave drone within the preset range of the master drone.

[0040] This application uses the above formula to calculate the collaborative confidence of the collaborative slave drones near the master drone.

[0041] In one embodiment, the specific implementation of the above step 140 may be: performing a weighted average on the collaborative confidence and the first confidence to obtain a comprehensive confidence.

[0042] This application can perform a weighted average of the collaborative confidence and the first confidence to calculate the comprehensive confidence corresponding to the first signal type. The weights of the collaborative confidence and the first confidence can be set equal, or they can be determined based on the confidence of the edge model carried by each drone (i.e., the weight is positively correlated with the confidence of the corresponding edge model).

[0043] In one embodiment, the above-mentioned wireless signal recognition method based on drone collaboration may also include: respectively calculating the mean square error between the recognition results of the wireless signal of each drone and the collaborative slave drone within a preset range of the master drone; selecting the drone with the smallest mean square error as the target drone, and using the target drone to store the wireless signal.

[0044] The present application can calculate the mean square error between the recognition results of the master drone and each nearby drone for the wireless signal, and select the drone with the smallest mean square error as the target drone (the difference between the recognition result of the target drone and the recognition result of the collaborative slave drone is the smallest, that is, it represents the majority of the recognition results) to store the wireless signal (when the comprehensive confidence of the recognition result of the wireless signal is greater than the confidence threshold).

[0045] In one embodiment, the above-mentioned wireless signal recognition method based on drone collaboration may also include: the knowledge distillation method of the edge model carried on the drone is: closing some network channels of the signal recognition model and calculating the loss function value; if the loss function value is less than a preset value, continuing to close some network channels until the loss function value is greater than or equal to the preset value; deleting the closed network channels to obtain the edge model; based on the loss function value and the preset value of the edge model, calculating the confidence of the edge model.

[0046] Specifically, the present application reduces the model complexity and computational complexity by gradually closing some network channels of the signal recognition model, and calculates its loss function value while closing them (indicating the degree to which the recognition accuracy of the model is reduced after closing some network channels). If the loss function is small, the network channels can continue to be closed until the loss function reaches the set value, thereby obtaining an edge model. In order to further reduce the complexity of the model, the present application can directly delete the closed network channels and calculate the confidence of the edge model based on the loss function value. Preferably, the present application can close different network channels for drones with different tasks in a drone group, so as to realize that the drones retain different characteristics of the signal recognition model in the server, and then obtain edge models with different recognition focuses, so as to better adapt to the recognition of various wireless signals.

[0047] Optionally, the present application can use a feature-based knowledge distillation method to distill a signal recognition model for classifying and identifying wireless signals. Specifically, the input wireless signal data is first integrated and the one-dimensional wireless signal data is packaged into two-dimensional image data. The one-dimensional wireless signal data has the same data length, so the packaged two-dimensional image data also has a fixed size. The two-dimensional image data is used as the data input of the knowledge distillation model to facilitate the training and learning of the knowledge distillation method. The knowledge distillation method uses a loss function based on mean square error:

[0048]

[0049] in, and Respectively represent classification vectors and corresponding target vectors, k is the total number of classification vectors.

[0050] Figure 2 This is a schematic diagram of a wireless signal recognition system based on UAV collaboration provided by an exemplary embodiment of the present application. The wireless signal recognition system based on UAV collaboration is set in a master UAV in a UAV group, and the UAV group includes a master UAV and at least one cooperative slave UAV; Figure 2 As shown, the wireless signal recognition system 20 based on UAV collaboration includes: a wireless signal receiving module 21 for receiving wireless signals; a preliminary signal type recognition module 22 for identifying a first signal type of the wireless signal and a corresponding first confidence level based on a master edge model carried on a master UAV; wherein the master edge model is a signal recognition model trained by a server and obtained through knowledge distillation, and the first confidence level is determined by the confidence level of the master edge model and the probability of the first signal type appearing in the UAV group; a collaborative signal acquisition module 23 for acquiring a second signal type and a corresponding second confidence level of the wireless signal identified by a collaborative edge model carried on a collaborative slave UAV within a preset range of the master UAV if the first confidence level is less than a preset confidence level threshold; wherein the collaborative edge model is a signal recognition model obtained through knowledge distillation, and the second confidence level is the second confidence level of the collaborative edge model; a collaborative signal type determination module 24 for comprehensively calculating a comprehensive confidence level of the first signal type based on the second confidence level and the first confidence level of a second signal type that is the same as the first signal type; and a wireless signal storage module 25 for storing the wireless signal if the comprehensive confidence level is greater than the confidence level threshold.

[0051] The wireless signal recognition system based on UAV collaboration provided by the present application is provided in a master UAV in a UAV group, and the UAV group includes a master UAV and at least one collaborative slave UAV; a wireless signal is received by a wireless signal receiving module 21; a preliminary signal type recognition module 22 identifies a first signal type of the wireless signal and a corresponding first confidence level based on a master edge model carried on the master UAV; wherein the master edge model is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is determined by the confidence level of the master edge model and the probability of the first signal type appearing in the UAV group; if the first confidence level is less than a preset confidence level threshold, the collaborative signal acquisition module 23 obtains a second signal type and a corresponding second confidence level of the wireless signal obtained by the collaborative edge model carried on the collaborative slave UAV within the preset range of the master UAV; wherein the collaborative ... a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is a signal recognition model obtained by server training obtained by knowledge distillation, and the first confidence level is a signal recognition model obtained by server training obtained by The second confidence is obtained through knowledge distillation, which is the second confidence of the collaborative edge model; the collaborative signal type determination module 24 obtains the comprehensive confidence of the first signal type based on the first confidence and the second confidence; if the comprehensive confidence is greater than the confidence threshold, the wireless signal storage module 25 stores the wireless signal; that is, the master drone first performs signal recognition based on its own main edge model and obtains the first confidence. If the confidence is low, the collaborative slave drone in the drone group is used to identify the signal and obtain the second confidence. The recognition results and confidence of multiple drones are combined to determine the recognition result of the wireless signal, so as to utilize the mutual assistance advantage of the drone group and effectively improve the wireless signal recognition accuracy of a single drone, thereby reducing the storage and transmission requirements of the drone. At the same time, edge computing can be used to greatly reduce the computing power of the host or acquired server, thereby improving the working efficiency of the drone group.

[0052] In one embodiment, the above-mentioned cooperative signal type determination module 24 can be further configured to: obtain a first number of cooperative slave drones within a preset range of the master drone; count a second number of second signal types that are the same as the first signal type identified based on the cooperative slave drones within the preset range of the master drone; and calculate a comprehensive confidence level of the first signal type based on the first number, the second number, the second confidence level and the first confidence level.

[0053] In one embodiment, the above-mentioned cooperative signal type determination module 24 can be further configured to: calculate the ratio of the second number to the first number to obtain the cooperative ratio of the first signal type; based on the cooperative ratio and the second confidence level, calculate the cooperative confidence level of the cooperative slave drone; based on the cooperative confidence level and the first confidence level, calculate the comprehensive confidence level of the first signal type.

[0054] In one embodiment, the above-mentioned cooperation signal type determination module 24 can be further configured as follows: the calculation formula of the cooperation confidence is:

[0055] ;in, is the collaborative confidence, is the collaborative ratio, is a second confidence level of the cooperative slave UAV corresponding to a second signal type that is the same as the first signal type, The second confidence level of the coordinated slave drone within the preset range of the master drone.

[0056] In one embodiment, the above-mentioned cooperative signal type determination module 24 can be further configured to: perform weighted averaging on the cooperative confidence and the first confidence to obtain a comprehensive confidence.

[0057] In one embodiment, the wireless signal recognition system 20 based on drone collaboration can be further configured to: calculate the mean square error between the recognition results of the wireless signal of each drone and the collaborative slave drone within the preset range of the master drone; select the drone with the smallest mean square error as the target drone, and use the target drone to store the wireless signal.

[0058] In one embodiment, the above-mentioned wireless signal recognition system 20 based on drone collaboration can be further configured as follows: the knowledge distillation method of the edge model carried on the drone is: closing some network channels of the signal recognition model and calculating the loss function value; if the loss function value is less than the preset value, continue to close some network channels until the loss function value is greater than or equal to the preset value; delete the closed network channels to obtain the edge model; based on the loss function value and the preset value of the edge model, calculate the confidence of the edge model.

[0059] Below, reference Figure 3 The electronic device according to the embodiment of the present application is described. The electronic device may be either or both of the first device and the second device, or a standalone device independent of them, and the standalone device may communicate with the first device and the second device to receive collected input signals from them.

[0060] Figure 3 The figure shows a block diagram of an electronic device according to an embodiment of the present application.

[0061] like Figure 3 As shown, the electronic device 10 includes one or more processors 11 and a memory 12 .

[0062] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

[0063] The memory 12 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), a hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may execute the program instructions to implement the methods of the various embodiments of the present application described above and / or other desired functions. Various contents such as input signals, signal components, and noise components may also be stored in the computer-readable storage medium.

[0064] In one example, the electronic device 10 may further include an input device 13 and an output device 14 , and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0065] When the electronic device is a stand-alone device, the input device 13 may be a communication network connector, configured to receive collected input signals from the first device and the second device.

[0066] In addition, the input device 13 may also include, for example, a keyboard, a mouse, and the like.

[0067] The output device 14 can output various information to the outside, including determined distance information, direction information, etc. The output device 14 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0068] Of course, to simplify, Figure 3 Only some of the components related to the present application in the electronic device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, the electronic device 10 may further include any other appropriate components according to specific application scenarios.

[0069] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.

[0070] The computer program product may be written in any combination of one or more programming languages ​​to implement the program code for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0071] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0072] The computer-readable storage medium may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0073] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0074] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0075] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0076] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0077] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A wireless signal recognition method based on UAV collaboration, characterized in that: The invention is applied to a master drone in a drone swarm, wherein the drone swarm includes the master drone and at least one cooperative slave drone; the wireless signal recognition method based on drone cooperation includes: Receive wireless signals; Based on a master edge model carried by the master drone, a first signal type of the wireless signal and a corresponding first confidence score are identified; wherein the master edge model is a signal recognition model trained by a server and obtained through knowledge distillation, and the first confidence score is determined by the confidence score of the master edge model and the probability of the first signal type appearing in the drone swarm; If the first confidence level is less than a preset confidence level threshold, obtaining a second signal type and a corresponding second confidence level of the wireless signal identified by a collaborative edge model carried by a collaborative slave UAV within a preset range of the master UAV; wherein the collaborative edge model is obtained by knowledge distillation from the signal recognition model, and the second confidence level is the second confidence level of the collaborative edge model; Based on the first confidence level and the second confidence level, comprehensively calculate and obtain a comprehensive confidence level of the first signal type; If the comprehensive confidence is greater than the confidence threshold, the wireless signal is stored.

2. The wireless signal recognition method based on UAV collaboration according to claim 1 is characterized in that: The step of comprehensively calculating the first confidence level and the second confidence level to obtain the comprehensive confidence level of the first signal type includes: Obtaining a first number of coordinated slave drones within a preset range of the master drone; Counting a second number of second signals of the same type as the first signal that are identified by the cooperative slave drones within a preset range of the master drone; A comprehensive confidence level of the first signal type is calculated based on the first number, the second number, the second confidence level, and the first confidence level.

3. The wireless signal recognition method based on UAV collaboration according to claim 2 is characterized in that: The calculating, based on the first number, the second number, the second confidence level, and the first confidence level, to obtain a comprehensive confidence level of the first signal type includes: Calculating a ratio of the second number to the first number to obtain a synergy ratio of the first signal type; Calculating the cooperation confidence of the cooperative slave UAV based on the cooperation ratio and the second confidence; Based on the collaborative confidence and the first confidence, a comprehensive confidence of the first signal type is calculated.

4. The wireless signal recognition method based on UAV collaboration according to claim 3 is characterized in that: The step of calculating the cooperation confidence of the cooperative slave UAV based on the cooperation ratio and the second confidence includes: The calculation formula of the collaborative confidence is: ； in, is the collaborative confidence, is the collaborative ratio, is a second confidence level of the cooperative slave UAV corresponding to a second signal type that is the same as the first signal type, The second confidence level of the coordinated slave drone within the preset range of the master drone.

5. The wireless signal recognition method based on UAV collaboration according to claim 3 is characterized in that: The calculating, based on the collaborative confidence and the first confidence, a comprehensive confidence of the first signal type includes: A weighted average is performed on the collaborative confidence and the first confidence to obtain the comprehensive confidence.

6. The wireless signal recognition method based on UAV collaboration according to claim 1, characterized in that: The wireless signal recognition method based on UAV collaboration also includes: Calculating the mean square error between the recognition results of the wireless signal by each drone and the cooperative slave drone within the preset range of the master drone; The UAV with the smallest mean square error is selected as a target UAV, and the target UAV is used to store the wireless signal.

7. The wireless signal recognition method based on UAV collaboration according to claim 1 is characterized in that: The knowledge distillation methods for edge models onboard drones include: Closing some network channels of the signal recognition model and calculating the loss function value; If the loss function value is less than the preset value, continue to close some network channels until the loss function value is greater than or equal to the preset value; Deleting the closed network channels to obtain the edge model; The confidence of the edge model is calculated based on the loss function value of the edge model and the preset value.

8. The wireless signal recognition system based on UAV collaboration is characterized by: A master drone is provided in a drone swarm, wherein the drone swarm includes the master drone and at least one cooperative slave drone; the wireless signal recognition system based on drone cooperation includes: A wireless signal receiving module, used for receiving wireless signals; a preliminary signal type identification module, configured to identify a first signal type and a corresponding first confidence level of the wireless signal based on a master edge model carried by the master UAV; wherein the master edge model is a signal identification model trained by a server and obtained through knowledge distillation, and the first confidence level is determined by the confidence level of the master edge model and the probability of the first signal type appearing in the UAV swarm; a collaborative signal acquisition module, configured to, if the first confidence level is less than a preset confidence level threshold, acquire a second signal type and a corresponding second confidence level of a wireless signal identified by a collaborative edge model carried by a collaborative slave UAV within a preset range of the master UAV; wherein the collaborative edge model is derived from the signal recognition model through knowledge distillation, and the second confidence level is the second confidence level of the collaborative edge model; a collaborative signal type determination module, configured to obtain a comprehensive confidence level of the first signal type by comprehensive calculation based on a second confidence level of a second signal type that is the same as the first signal type and the first confidence level; The wireless signal storage module is configured to store the wireless signal if the comprehensive confidence is greater than the confidence threshold.

9. A computer-readable storage medium, characterized in that The storage medium stores a computer program, and the computer program is used to execute the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the method according to any one of claims 1 to 7.

Images