Cooperative communication interference detection method and system in unmanned aerial vehicle ad hoc network

Abstract

The invention provides a cooperative communication interference detection method and system in an unmanned aerial vehicle ad hoc network, and the method specifically comprises the following steps: a central server firstly initializes gray wolf algorithm parameters and an interference detection model, and transmits the gray wolf algorithm parameters and the interference detection model to unmanned aerial vehicles participating in training; each unmanned aerial vehicle locally uses the interference data set to train a model, after the number of iterations is reached, a score is calculated according to model performance such as the minimum loss value or the maximum accuracy, and the optimal score is sent to the central server; and the central server selects the model with the highest score as the optimal model of the round, and sends the weight parameters of the model to all the unmanned aerial vehicles. And each unmanned aerial vehicle continues to train based on the updated weight, the above process is repeated until the model accuracy is converged, and finally the optimal interference detection model is returned. According to the method, the communication overhead is reduced, the detection accuracy is improved, and the adaptability to an unstable network is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of interference detection in unmanned aerial vehicle (UAV) ad-hoc networks, and in particular to a method and system for collaborative communication interference detection in UAV ad-hoc networks. Background Art

[0002] In recent years, with the rapid development of UAV technology, UAVs and UAV ad-hoc networks have been increasingly applied to various civilian, military, and scientific fields, such as performing tasks like reconnaissance, disaster management, and space exploration. However, with the sharp increase in the number of UAVs, the interference problem for UAV ad-hoc networks has gradually emerged, posing challenges to flight safety, data security, and social stability. Therefore, the research on interference detection in UAV ad-hoc networks is particularly important. Interference detection is the basis for ensuring the robustness of the communication process and a key component in the construction of the entire communication system.

[0003] Considering the problems of network bandwidth limitation and data loss during UAV ad-hoc network communication, it is necessary to reduce the communication overhead and improve the detection performance in an unstable network environment during the actual interference detection process. Currently, there are many algorithms for researching UAV collaborative communication interference detection, but most of them have a large communication overhead. Interference detection based on federated learning can reduce the communication overhead. However, in an unstable network environment, the classic federated learning framework will exclude some UAVs with limited network bandwidth or access restrictions from the training rounds, that is, the global model is not sent to these UAVs for local optimization. This simple processing method will greatly affect the accuracy of interference detection. Summary of the Invention

[0004] Aiming at the problems existing in the prior art, the present invention provides a method for collaborative communication interference detection in UAV ad-hoc networks. Based on the interference detection technology of federated learning and combined with the gray wolf optimization algorithm, it can reduce the communication overhead, improve the detection accuracy, and enhance the adaptability to unstable networks.

[0005] To solve the above technical problems, the present invention provides the following technical solution: A method for collaborative communication interference detection in UAV ad-hoc networks, comprising the following steps:

[0006] S1. For a number of UAVs in the area, the central server initializes the parameters of the gray wolf algorithm and the interference detection model, and sends the initial interference detection model to the UAVs participating in this round of training;

[0007] S2. Use the FedGWO weights to train the model on the UAVs participating in the training until the preset number of iterations is reached to obtain the best model;

[0008] S3. Calculate the score according to the lowest loss value or the highest accuracy, and each UAV sends the best score value to the central server;

[0009] S4. On the central server, update the best model according to the UAV with the highest score, and send the best model of this round back to each UAV;

[0010] S5. Return to execute step S2 until the interference detection accuracy rate converges, and output the final best model for realizing interference detection.

[0011] Furthermore, in the aforementioned step S1, the central server initializes the parameters of the grey wolf algorithm as follows: initialize the number of UAVs as the number of grey wolves n; initialize the search range, the maximum number of iterations w0, and the optimal solution α; initialize the vector Its expression is:

[0012]

[0013] where, and are vectors of random numbers between [0, 1], is a vector of linear values, and as the number of iterations increases, its value gradually decreases from 2 to 0. Its expression is:

[0014]

[0015] where, icurrent represents the current number of iterations, and imax is the maximum number of iterations.

[0016] Furthermore, the aforementioned step S3 includes the following sub-steps:

[0017] S3.1. According to the grey wolf algorithm, divide all UAVs in the UAV ad hoc network into four levels α, β, δ, ω. Compare the interference detection effects of all UAV individuals. Let the three levels α, β, δ represent the leaders of the UAV population, representing the optimal, sub-optimal, and third-optimal individuals respectively. The remaining UAVs are in the last level of the hierarchical classification system, that is, ω.

[0018] S3.2. Use grey wolves at each level to surround and hunt prey in the search space to find the best solution, that is, find the model weights that maximize the interference accuracy or minimize the loss.

[0019] Furthermore, the aforementioned step S3.2 includes the following sub-steps:

[0020] S3.2.1. The mathematical model of the surrounding behavior when surrounding prey is established by the following formula:

[0021]

[0022] where, itr represents the current iteration, is a vector representing the position of the gray wolf, is a vector representing the position of the prey. S3.2.2. When hunting prey, the alpha wolf is responsible for hunting, and the beta wolf and the omega wolf also participate in the hunting. In the optimization problem, based on the fact that the search space is usually unknown, and the optimal solution, that is, the position of the prey, is also unknown, the first three levels alpha, beta, and delta in the population are used to explore the position of the prey; the positions of the first three levels alpha, beta, and delta are used to update the positions of other members in the population, and this position includes the position of the omega wolf.

[0023] Furthermore, as described above, in step S3.2.2, the first three levels alpha, beta, and delta in the population are used to explore the position of the prey, as follows:

[0024]

[0025] where, respectively represent the distances between alpha, beta, and delta and other individuals, respectively represent the current positions of alpha, beta, and delta, The expression of is the same as that of in the initialization parameters, and is used to simulate the randomness of the gray wolf moving near the prey, is the current position of the gray wolf individual.

[0026] Furthermore, as described above, in step S3.2.2, after being affected by alpha, beta, and delta, the positions of the remaining gray wolf individuals are updated, as follows:

[0027]

[0028] where, respectively represent the position updates of the remaining gray wolf individuals after being affected by alpha, beta, and delta, and they determine the intensity and direction of the gray wolf moving towards the prey, Starting from 2, it gradually decreases to 0 during the iteration process. This decreasing process simulates the behavior of the gray wolf pack gradually approaching the prey during the hunting process. When their values are close to 0, the gray wolf pack is more inclined to search the surrounding area carefully to find the optimal solution. When their values are larger, the gray wolf pack may conduct a wider search in the solution space.

[0029] Furthermore, in step S3.2.2 as described above, after taking the average value of the individual position of the gray wolf in the next round of iteration is obtained, that is, the update direction of the model weight, as follows:

[0030]

[0031] On the other hand, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of any one of the methods of the present invention are implemented.

[0032] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods of the present invention are implemented.

[0033] Compared with the prior art, the beneficial technical effects of the present invention adopting the above technical solutions are as follows:

[0034] (1) Compared with the traditional interference detection method based on federated learning, this method can have a higher interference detection accuracy. For the ad hoc network of 20 unmanned aerial vehicles (UAVs), compared with the traditional federated averaging method, the interference detection accuracy can be increased by 3.43%.

[0035] (2) Compared with the traditional interference detection method based on federated learning, this method can effectively reduce the communication overhead. For the ad hoc network of 20 UAVs, compared with the traditional federated averaging method, the communication overhead can be reduced by 59%.

[0036] (3) This method can enhance the adaptability to unstable networks. When data transmitted during communication is lost, it can still maintain a relatively high interference detection accuracy. Description of the Drawings

[0037] Figure 1 is a flowchart of collaborative communication interference detection in an ad hoc network of UAVs;

[0038] Figure 2 is a schematic diagram of training an interference detection model in an ad hoc network of UAVs using the grey wolf algorithm;

[0039] Figure 3 is a schematic diagram of updating the global optimal model in an ad hoc network of UAVs using federated learning. Detailed Embodiments

[0040] In order to better understand the technical content of the present invention, specific embodiments are hereby given and described in conjunction with the accompanying drawings as follows.

[0041] In the present invention, aspects of the present invention are described with reference to the accompanying drawings, in which a number of illustrative embodiments are shown. The embodiments of the present invention are not limited to those described in the drawings. It should be understood that the present invention can be implemented by any one of the various concepts and embodiments introduced above, as well as the concepts and embodiments described in detail below, because the concepts and embodiments disclosed in the present invention are not limited to any embodiment. Additionally, some aspects disclosed in the present invention can be used alone or in any suitable combination with other aspects disclosed in the present invention.

[0042] Reference Figure 1 , the present invention provides a method for collaborative communication interference detection in an unmanned aerial vehicle (UAV) ad hoc network, including the following steps:

[0043] S1. For a number of UAVs within a region, the central server initializes the parameters of the Grey Wolf Optimization (GWO) algorithm and the interference detection model, and sends the initial interference detection model to the UAVs participating in this round of training;

[0044] S2. Use the FedGWO weights on the UAVs participating in the training to train the model until the preset number of iterations is reached to obtain the optimal model;

[0045] S3. Calculate the score based on the lowest loss value or the highest accuracy rate, and each UAV sends the optimal score value to the central server;

[0046] S4. On the central server, update the optimal model according to the UAV with the highest score, and send the optimal model of this round back to each UAV;

[0047] S5. Return to execute step S2 until the interference detection accuracy rate converges, and output the final optimal model for realizing interference detection.

[0048] As a preferred embodiment of the present invention, in step S1, the central server initializes the parameters of the Grey Wolf Optimization algorithm as follows: initialize the number of UAVs as the number of grey wolves n; initialize the search range, the maximum number of iterations w0, and the optimal solution α; initialize the vector The expression is:

[0049]

[0050] Where and are vectors of random numbers between [0, 1], is a vector of linear values, which decreases from 2 to 0 gradually as the number of iterations increases, and its expression is:

[0051]

[0052] Where icurrent represents the current number of iterations, and imax is the maximum number of iterations.

[0053] As a preferred embodiment of the present invention, referring to Figure 2 , step S3 includes the following sub-steps:

[0054] S3.1. According to the grey wolf algorithm, all unmanned aerial vehicles (UAVs) in the UAV ad hoc network are divided into four levels: α, β, δ, and ω. Compare the interference detection effects of all UAV individuals. Let the three levels of α, β, and δ represent the leaders of the UAV population, representing the optimal, sub-optimal, and third-optimal individuals respectively, and the remaining UAVs are in the last level of the hierarchical classification system, i.e., ω.

[0055] S3.2. Use grey wolves at each level to surround and hunt prey in the search space to find the best solution, that is, find the model weights that maximize the interference accuracy or minimize the loss.

[0056] As a preferred embodiment of the present invention, referring to Figure 3 , in step S3.2, when surrounding the prey, the mathematical model of the surrounding behavior is established by the following formula:

[0057]

[0058] where itr represents the current iteration, is the vector representing the position of the grey wolf, is the vector representing the position of the prey.

[0059] As a preferred embodiment of the present invention, when hunting prey, usually, the alpha wolf α is responsible for hunting. Sometimes, the beta wolf β and the omega wolf ω also participate in hunting. In an optimization problem, the search space is usually unknown, so the position of the optimal solution (prey) is also unknown. Based on this assumption, the first three levels in the population (i.e., α, β, and δ) should have a deeper understanding of the possible position of the prey. Therefore, the positions of the first three levels are used to update the positions of other members in the population (including the omega wolf ω). For this purpose, the following formula is proposed:

[0060]

[0061] where, respectively represent the distances between α, β, and δ and other individuals, respectively represent the current positions of α, β, and δ, are used to simulate the randomness of the grey wolf moving near the prey. This randomness simulates the possible uncertain and changing behaviors of the grey wolf pack when tracking the prey. Their values are usually randomly generated between 0 and 2, which helps to explore different regions of the solution space and avoid the algorithm falling into a local optimal solution prematurely. is the current position of the grey wolf individual.

[0062]

[0063] Among them, respectively represent the position updates of the remaining gray wolf individuals after being affected by α, β, and δ, which determine the intensity and direction of the gray wolves' movement towards the prey. The values of these coefficients affect the tightness of the gray wolves' surrounding the prey, thereby affecting the exploratory (exploring new areas) and exploitative (exploiting known areas) nature of the search behavior. It usually starts from 2 and gradually decreases to 0 during the iteration process. This decreasing process simulates the behavior of the gray wolf pack gradually approaching the prey during the hunting process. When their values are close to 0, the gray wolf pack is more inclined to search the surrounding area carefully to find the optimal solution. When their values are larger, the gray wolf pack may conduct a more extensive search in the solution space.

[0064] For After taking the average value, the individual position of the gray wolf in the next iteration is obtained, that is, the update direction of the model weight:

[0065]

[0066] On the other hand, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of any one of the methods in this example are implemented.

[0067] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in this embodiment are implemented.

[0068] Although the present invention has been described above with preferred embodiments, it is not intended to limit the present invention. Those with ordinary knowledge in the technical field to which the present invention pertains can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be determined by the scope defined in the claims.

Claims

1. A collaborative communication interference detection method in an unmanned aerial vehicle (UAV) ad-hoc network, characterized in that, It includes the following steps: S1. For several drones in the area, the central server initializes the parameters of the Grey Wolf Optimization algorithm and the interference detection model, and sends the initial interference detection model to the drones participating in this round of training; S2. Use the FedGWO weights on the drones participating in the training to train the model until the preset number of iterations is reached to obtain the optimal model; S3. Calculate the score according to the lowest loss value or the highest accuracy rate, and each drone sends the optimal score value to the central server; S4. On the central server, update the optimal model according to the drone with the highest score, and send the optimal model of this round back to each drone; S5. Return to execute step S2 until the interference detection accuracy converges, and output the final optimal model for realizing interference detection.

2. The collaborative communication interference detection method in an unmanned aerial vehicle ad hoc network according to claim 1, wherein In step S1, the central server initializes the parameters of the grey wolf algorithm as follows: initialize the number of drones as the number of grey wolves n; initialize the search range, the maximum number of iterations w0, and the optimal solution α; initialize the vector Its expression is: Among them, and are vectors of random numbers between [0, 1], is a vector of linear values, and as the number of iterations increases, its value gradually decreases from 2 to 0, and its expression is: Among them, \(i_{current}\) represents the current number of iterations, and \(i_{max}\) is the maximum number of iterations.

3. A method for collaborative communication interference detection in an unmanned aerial vehicle ad-hoc network according to claim 1, characterized in that, Step S3 includes the following sub-steps: S3.

1. According to the Grey Wolf Optimization algorithm, all drones in the drone ad-hoc network are divided into four levels: \(\alpha\), \(\beta\), \(\delta\), and \(\omega\). Compare the interference detection effects of all drone individuals. Let the three levels of \(\alpha\), \(\beta\), and \(\delta\) represent the leaders of the drone population, representing the optimal, sub-optimal, and third-optimal individuals respectively. The remaining drones are in the last level of the hierarchical classification system, that is, \(\omega\). S3.

2. Use the grey wolves at each level to surround and hunt the prey in the search space to find the best solution, that is, find the model weights that maximize the interference accuracy or minimize the loss.

4. A collaborative communication interference detection method in an unmanned aerial vehicle ad-hoc network according to claim 3, characterized in that, Step S3.2 includes the following sub-steps: S3.2.

1. The mathematical model of the surrounding behavior when surrounding the prey is established by the following formula: where itr represents the current iteration, is a vector representing the position of the grey wolf, is a vector representing the position of the prey. S3.2.

2. When hunting the prey, the lead wolf \(\alpha\) is responsible for hunting, and the second wolf \(\beta\) and the third wolf \(\omega\) also participate in the hunting. In the optimization problem, since the search space is usually unknown and the optimal solution, that is, the position of the prey, is also unknown, the first three levels \(\alpha\), \(\beta\), and \(\delta\) in the population are used to explore the position of the prey; the positions of the first three levels \(\alpha\), \(\beta\), and \(\delta\) are used to update the positions of other members in the population, and this position includes the position of the last wolf \(\omega\).

5. The collaborative communication interference detection method in an unmanned aerial vehicle ad hoc network according to claim 4, wherein In step S3.2.2, the first three levels \(\alpha\), \(\beta\), and \(\delta\) in the population are used to explore the position of the prey, as follows: Among them, respectively represent the distances between α, β, and δ and other individuals, respectively represent the current positions of α, β, and δ, The expression of is the same as that in the initialization parameters and is used to simulate the randomness of the movement of gray wolves near the prey, is the current position of the gray wolf individual.

6. A method for collaborative communication interference detection in an unmanned aerial vehicle ad-hoc network according to claim 4, characterized in that, In step S3.2.2, after being affected by \(\alpha\), \(\beta\), and \(\delta\), the positions of the remaining grey wolf individuals are updated, as follows: Among them, respectively represent the position updates of the remaining gray wolf individuals after being affected by α, β, and δ, which determine the intensity and direction of the gray wolves moving towards the prey. Starting from 2, it gradually decreases to 0 during the iteration process. This decreasing process simulates the behavior of the gray wolf pack gradually approaching the prey during the hunting process. When their values are close to 0, the gray wolf pack is more inclined to carefully search the surrounding area to find the optimal solution. When their values are larger, the gray wolf pack may conduct a more extensive search in the solution space.

7. A method for collaborative communication interference detection in an unmanned aerial vehicle ad hoc network according to claim 4, characterized in that, In step S3.2.2, for After taking the average value, the individual position of the gray wolf in the next iteration is obtained, which is the update direction of the model weight, as shown in the following formula:

8. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images