A self-organizing cooperative k-coverage method for sensor networks based on game theory

Abstract

The present invention relates to a sensor network self-organization collaborative k coverage method based on game theory, comprising the following steps: a. Taking the effective coverage area of ​​the system as an indicator, establishing a collaborative k coverage global optimization mathematical model; b. Establishing a distributed collaborative decision-making model based on network potential game; c. Each sensor node determines the current position, coverage area and neighbor set, and initializes its action and memory vector; d. The sensor node calculates the individual payment value, the best response, the regret value and the non-inferior action set using the neighborhood information, and sends the result to the neighbor; e. The individual selects the restricted best response strategy using the mixed response rule, and updates the memory vector; f. The individual randomly selects the elements in the memory vector with the same probability, and updates the current action; g. Jump back to the step (d), iterates until the system is stable, and outputs the distributed optimization result. The method of the present invention can effectively realize autonomous collaborative decision-making and group intelligence emergence.

Description

Technical Field

[0001] The invention relates to a sensor network self-organizing collaborative k-coverage method based on game theory. Background Art

[0002] Wireless sensor network refers to a distributed network system composed of multiple sensor nodes with sensing, communication, computing and control capabilities. With the development of artificial intelligence technology and wireless communication technology, sensor network systems have gradually become an effective way to replace humans in performing tasks such as environmental monitoring and reconnaissance. In the military, wireless sensor networks composed of ground, air and space robots can be deployed to troops at all levels for reconnaissance and surveillance, providing troops with high-definition images, real-time battlefield information, environmental monitoring and combat casualty assessment information. In the civilian field, due to its excellent characteristics of self-organization and no need for human intervention, it can be used for monitoring in mountainous areas or more complex environments, especially in some places that pose safety risks to humans.

[0003] In order to obtain the most detailed detection information, it is usually necessary to configure as many sensor nodes as possible to cover the entire target area. However, in some extremely harsh battlefield conditions, the above configuration scheme cannot be implemented, and the sensor nodes can only be randomly scattered in the target area. In addition, the limitation of energy supply also determines that each sensor node has only a limited lifespan and usage time. Therefore, how to effectively balance the network service life and coverage performance has become a key issue in the configuration of sensor networks. For this reason, the engineering field has proposed a solution for redundant coverage of the target area, which is also called the sensor network k coverage problem. This solution divides the sensor nodes into k groups, and only one group of sensors is activated at each time, while the remaining nodes remain closed to save energy, and ultimately achieve the purpose of increasing the overall effective coverage area on the basis of ensuring the minimum usage time of the entire network.

[0004] Most of the k-coverage solution methods in the prior art are carried out under a centralized control framework, using centralized nodes and global information to optimize and guide the decision-making process. However, since all sensor nodes in a wireless sensor network can only interact with a limited number of neighborhood individuals, there is no central node that can obtain global information. It can be seen that this type of centralized communication structure in the prior art is not suitable for distributed network systems without central nodes. Therefore, how to maximize the effective coverage area of ​​the system by only using limited neighborhood communication and autonomous coordination has become an important research issue in the field of wireless sensors.

[0005] In summary, most of the existing solutions to the multi-sensor k-coverage problem focus on approximate solution algorithms under centralized structures. Only a few distributed solutions can achieve collaboration in a decentralized environment, but they still cannot refine and improve the global performance of the system. Summary of the invention

[0006] The purpose of the present invention is to provide a sensor network self-organizing collaborative k-coverage method based on game theory.

[0007] To achieve the above-mentioned object of the invention, the present invention provides a sensor network self-organizing collaborative k-coverage method based on game theory, comprising the following steps:

[0008] a. Taking the effective coverage area of ​​the system as an indicator, a global optimization mathematical model of collaborative k coverage is established;

[0009] b. Establish a distributed collaborative decision-making model based on network potential game;

[0010] c. Each sensor node determines its current location, coverage area, and neighbor set, and initializes its action and memory vectors;

[0011] d. The sensor node uses the neighborhood information to calculate the individual payment value, the best response, the regret value and the non-inferior action set, and sends the results to the neighbors;

[0012] e. The individual uses the mixed response rule to select the restricted best response strategy and updates the memory vector;

[0013] f. The individual randomly selects elements in the memory vector with the same probability and updates the current action;

[0014] g. Jump back to step (d), iterate until the system is stable, and output the distributed optimization results.

[0015] According to one aspect of the present invention, in step (a), the set of sensor nodes randomly distributed in a two-dimensional plane area is W = {s i} i∈N ;

[0016] Where N = {1, 2, ..., n}, s i represents the i-th sensor node, whose location coordinate is X i =(x i ,y i ), the coverage area is X i is the center of the circle, r i A circular area with radius Θ i ;

[0017] Each sensor node i All are equipped with wireless communication modules with a communication radius of c i >2max i∈N r i ;

[0018] If j In s i Within the communication range, sj For i Neighbors, denoted by node s i The neighbor set of i ={s j ∈W|||X i -X j ||≤c i},in,

[0019] According to one aspect of the present invention, the sensor network nodes are divided into k independent covering sets {S j} j∈K , where K = {1, 2, ..., k}, and the detection time is divided into k corresponding time slots {T j} j∈K ;

[0020] In each time period T j Only the cover set S is activated j The sensors in the node are kept silent.

[0021] Each node s i can choose to join any of the k covering sets, and its action set is A i =K, let its action be a i ∈A i , indicating s i Opt-in i A covering set;

[0022] Define a=(a1,a2,…,a n ) is a group, A=Π i∈N A i is a grouping space. For any group a∈A, the jth covering set is represented by S j ={s i ∈W|a i =j};

[0023] The goal of sensor cooperative k-coverage is to maximize the following global objective function by selecting appropriate grouping a∈A:

[0024]

[0025] in, Denotes the covering set S j According to one aspect of the present invention, in step (b), each sensor node is regarded as an intelligent individual with perception, communication and computing capabilities, and a potential game model G(P,{A i},{u i The steps of},φ) are as follows:

[0026] Establish the participant set P: Consider each sensor node in the system as an independent game participant, then the player participant set P = W;

[0027] Create action set A i :Every individual has the same set of actions A i =K;

[0028] Establish payment function u i :For any group or action combination a, the individual payoff function is defined as

[0029]

[0030] Among them, \ is a set subtraction operation, and the calculation of this value only depends on the neighborhood information of the game individual.

[0031] Establish potential function φ: For any grouping or action combination a, define the potential function as:

[0032]

[0033] According to one aspect of the present invention, in step (c), each sensor node s i Use positioning and communication functions to determine the current location X i =(x i ,y i ), coverage area Θ i And the neighbor set Ω i ;

[0034] Each sensor node randomly initializes its actions With memory vector Where l is the memory length and the superscript t represents the number of iterations.

[0035] According to one aspect of the present invention, in step (d), during the tth coordination process, individual s i The local calculation is performed using the following steps:

[0036] Use neighborhood communication to obtain the current action information of neighbor individuals

[0037] Calculate the payment function value u i , best response Regret Value With non-inferior action set

[0038]

[0039]

[0040]

[0041]

[0042] in, represents time t divided by s i The combined actions of all other individuals.

[0043] According to one aspect of the present invention, in step (e), using local communication, individual s i Get neighbor's regret value information And calculate the maximum regret value in the neighborhood and the corresponding neighbors

[0044] The best strategy for dealing with individual computing constraints For: If r i t =0, then If (i)r i t >r i t* or (ii)r i t =r i t* And i<j * ,but Otherwise, with equal probability, Choose any action as

[0045] use Update individual memory,

[0046] According to one aspect of the present invention, in step (f), individual s i exist Randomly select with the same probability and get the updated action

[0047] According to one aspect of the present invention, in step (g), the loop is iterated until all individual actions no longer change, then the coordination process ends and a near-optimal or optimal k-coverage cooperative grouping solution is obtained.

[0048] According to the concept of the present invention, from the perspective of game theory, the sensor nodes in the network are regarded as game participants with communication, perception, computing and decision-making capabilities. By defining the communication topology, strategy set and payment function, a distributed collaborative optimization model based on network potential game is constructed. The self-organizing coordination mechanism and algorithm are designed by using game learning, which can not only ensure the rapid convergence of the Nash equilibrium solution, but also provide an effective way to select and refine the Nash equilibrium through the adjustment of memory length. As a mathematical theory for studying the decision-making behavior of rational individuals that influence and depend on each other and the equilibrium results, game theory mainly includes the following three elements: participants, action sets and payment functions. Game theory is widely used in humanities and social sciences such as economics and politics, but it also shows broader application prospects in engineering fields such as distributed systems. Specifically, the main reasons and motivations for using game theory for autonomous collaborative k coverage of sensor networks include the following aspects:

[0049] First, in a distributed system lacking a centralized control node, sensor nodes can only interact and coordinate locally, which is the same as the local interaction and coordination of individuals in network games. Therefore, game theory provides a good autonomous decision-making framework for sensor network collaborative k-coverage;

[0050] Secondly, due to the characteristics of the cooperative k-coverage problem, the decisions of sensor nodes are mutually influential, that is, the benefits of a sensor node are also affected by the decisions of its neighbors. Therefore, game theory provides an effective way to analyze the conflicts and cooperation between sensor nodes.

[0051] Thirdly, due to the autonomous nature of sensor nodes, distributed solutions based only on local interaction and coordination may lead to performance confusion at the system level, making it difficult to form an effective cooperation solution. Potential game provides an effective way to solve the above difficulties. It is a special game problem that can ensure that the individual payment value is consistent with the global interest. In the potential game, the change in individual payment value caused by any individual unilaterally changing its actions is strictly consistent with the change in global interest. This model guarantees the existence of pure strategy Nash equilibrium and the inclusion of Nash equilibrium in the global optimal solution.

[0052] Based on the above reasons, the present invention establishes a distributed optimization model of self-organizing collaborative k coverage of sensor networks under the potential game framework, and designs a distributed algorithm based on synchronous random game learning, which can provide an effective solution for autonomous collaborative k coverage of sensors in distributed environments.

[0053] According to the solution of the present invention, the collaborative k-coverage of sensor networks is transformed into a Nash equilibrium solution and selection problem in potential games. A synchronous random algorithm based on game learning is designed using limited memory and mixed response strategies, which can effectively realize autonomous collaborative decision-making and the emergence of group intelligence.

[0054] According to a solution of the present invention, self-organized collaborative decision-making can be achieved without relying on global information and centralized control. Each sensor makes autonomous decisions based only on neighborhood information, which can realize the intelligent emergence of group collaborative decision-making, help improve the system's autonomy and robustness, and is particularly suitable for outdoor locations where manual control cannot work.

[0055] According to a solution of the present invention, the convergence and refinement of Nash equilibrium can be achieved. For any sensor network with undirected communication, the method can ensure the global convergence of Nash equilibrium and eliminate the conflict between individual interests and global performance in a distributed environment. That is, under any memory length, the method can ensure the finite time convergence of the Nash equilibrium solution.

[0056] According to a scheme of the present invention, the trade-off between calculation time and solution optimality is achieved. In order to meet the different requirements of different task scenarios, this method provides the freedom to balance the trade-off between negotiation time and collaborative efficiency. When the individual memory length l approaches 1, the system can quickly converge to the Nash equilibrium, giving priority to meeting the fast negotiation requirements. As the memory length increases, the global performance evaluation index corresponding to the obtained Nash equilibrium gradually increases, and the negotiation efficiency is more satisfied. In this way, by adjusting the memory length, it is possible to quickly converge to the Nash equilibrium, and also to use long-term negotiations to obtain a more efficient collaborative coverage solution.

[0057] According to a solution of the present invention, it can be transplanted to other distributed coordination control and optimization decision-making problems that can be described as potential games. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A flowchart schematically showing a self-organizing collaborative decision-making method based on game theory according to an embodiment of the present invention;

[0059] Figure 2 A schematic diagram schematically showing time and sensor grouping in one embodiment of the present invention;

[0060] Figure 3 A schematic diagram of a local payment function of a sensor node according to an embodiment of the present invention is schematically shown;

[0061] Figure 4 Schematically showing a multi-sensor cooperative k coverage example (n=15, k=3) of an embodiment of the present invention;

[0062] Figure 5 Schematically showing the convergence curve of the distributed cooperative k-covering algorithm of one embodiment of the present invention (m=1, 2, 3, 4, 5);

[0063] Figure 6 Schematically showing the convergence curve of the distributed cooperative k-cover algorithm of one embodiment of the present invention (m=6, 10, 15);

[0064] Figure 7 The convergence curve (m=20, 25, 30) of the distributed cooperative k-covering algorithm according to an embodiment of the present invention is schematically shown. DETAILED DESCRIPTION

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art,

[0066] The drawings required for use in the implementation are briefly introduced. Obviously, the drawings described below are only some implementations of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative work.

[0067] When describing the embodiments of the present invention, the orientation or positional relationship expressed by the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" are based on the orientation or positional relationship shown in the relevant drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the above terms should not be understood as limiting the present invention.

[0068] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments cannot be described one by one here, but the embodiments of the present invention are not therefore limited to the following embodiments.

[0069] In the following embodiments and drawings, T represents a complete time zone, T j represents the jth time slot of the partition, S j represents the jth covering set, s i represents the i-th sensor node, u i Represents sensor node s i The payment function is t, t represents the number of iterations, Y represents whether the condition is met (yes), N represents whether the condition is not met (no), n represents the number of sensors, and k represents the number of groups.

[0070] See also Figure 1 The sensor network self-organizing collaborative k-coverage method based on game theory and distributed collaborative decision-making belongs to the intersection of artificial intelligence, multi-agent theory and sensor network. The research on the distributed optimization theory and method of the earth observation satellite system based on game theory is proposed. This method first establishes a global mathematical model, then establishes a distributed game model, initializes the algorithm, performs local calculations, then performs memory updates and action updates, and finally jumps back to the local calculation steps for loop iteration until the final solution is obtained.

[0071] The method of the present invention is described in detail below with a specific implementation method. The experimental object is a distributed sensor network composed of n = 15 heterogeneous sensor nodes. In order to achieve a balance between network life and coverage performance, it needs to be divided into k = 3 independent coverage sets. Figure 4 As shown in Figure 1, the sensor nodes are randomly scattered in a 5km×5km two-dimensional plane. For the convenience of calculation, the target area is divided into 10×10 grids. When the center point of a grid is on the sensor s i When the grid is within the coverage area of ​​s i The experimental environment is MATLAB 2016a version, the operating system is Win10.0, 3.07Ghz, and 4G memory.

[0072] Taking the effective coverage area of ​​the system as an indicator, a global optimization mathematical model of cooperative k coverage is established. The set of sensor nodes randomly scattered in the two-dimensional plane area is W = {s i} i∈N Where N = {1, 2, ..., n}, s i represents the i-th sensor node, whose location coordinate is X i =(x i ,y i ), the coverage area is X i is the center of the circle, r i A circular area with radius Θ i Each sensor node s i All are equipped with wireless communication modules with a communication radius of c i >2max i∈N r i If j In s i Within the communication range, s j For i Neighbors, node s i The neighbor set of i ={s j ∈W|||X i -X j ||≤c i},in,

[0073] In the present invention, in order to achieve a compromise between network service life and coverage performance, the sensor network nodes are divided into k independent coverage sets {S j} j∈K , K = {1, 2, ..., k}. At the same time, if Figure 2 As shown, the detection time is divided into k corresponding time slots {T j} j∈KIn each time period T j Only the cover set S is activated j The sensors in each node are kept silent to save energy. i can choose to join any of the k covering sets, that is, its action set is A i =K, let its action be a i ∈A i , indicating s i Opt-in i A covering set. Define a=(a1,a2,…,a n ) is a group, A=Π i∈N A i is the grouping space. For any group a∈A, the jth covering set is represented by S j ={s i ∈W|a i =j}. The goal of sensor cooperative k coverage is to maximize the following global objective function by selecting appropriate grouping a∈A:

[0074]

[0075] in, Denotes the covering set S j The union of the coverage areas of all nodes in .

[0076] In this embodiment, for the collaborative k-coverage problem with n=15 and k=3, the following global objective function is established:

[0077] F(a)=f1(a)+f2(a)+f3(a).

[0078] Then, a distributed collaborative decision-making model based on network potential game (also called distributed game model) can be established. Each sensor node is regarded as an intelligent individual with perception, communication and computing capabilities, and a potential game model G(P,{A i},{u i},φ) as follows: establish a participant set P, regard each sensor node in the system as an independent game participant, then the game participant set is P = W; establish an action set A i , each individual has the same action set A i =K; establish payment function u i , for any group or action combination a, the individual payoff function is defined as:

[0079]

[0080] Among them, \ is a set subtraction operation, and the calculation of this value depends only on the neighborhood information of the game individual, that is, the information of the individual action and the individual neighbor's action; establish the potential function φ, for any group or action combination a, define the potential function as:

[0081]

[0082] In this embodiment, the set of game players P = {s1, s2, ..., s 15}, each individual has the same set of actions A i ={1,2,3}, the individual payment function is The potential function is φ(a)=f1(a)+f2(a)+f3(a).

[0083] After that, the algorithm is initialized, that is, each sensor node s i Determine the current location, coverage area, and neighbor set, and initialize its action and memory vectors. Specifically, each sensor node s i Use positioning and communication functions to determine the current location X i =(x i ,y i ), coverage area Θ i And the neighbor set Ω i Each sensor node randomly initializes its actions With memory vector Where l is the memory length and the superscript t represents the number of iterations.

[0084] Then, local calculation is performed, that is, the sensor node uses the neighborhood information to calculate the individual payment value, the best response, the regret value and the non-inferior action set, and sends the result to the neighbors. i Use the following steps to perform local calculations to obtain the current action information of neighbor individuals:

[0085] Use neighborhood communication to obtain the current action information of neighbor individuals

[0086] Calculate the payment function value u i (See Figure 3 ), best response Regret Value With non-inferior action set

[0087]

[0088]

[0089]

[0090]

[0091] in, represents time t divided by s i The combined actions of all other individuals.

[0092] In the memory update step, individuals use the mixed response rule to select the best response strategy and update the memory vector. Specifically, using local communication, individual s i Get neighbor's regret value information And calculate the maximum regret value in the neighborhood and the corresponding neighbors Then the individual calculation is limited to the best response strategy For: If r i t =0, then If (i)r i t >r i t* or (ii)r i t =r i t* And i<j * ,but Otherwise, with equal probability, Choose any action as Then, using Update individual memory and get a new memory vector

[0093] Then the action is updated, that is, the individual randomly selects elements in the memory vector with the same probability and updates the current action. Specifically, individual s i exist Randomly select with the same probability and get the updated action

[0094] Finally, jump back to the above local calculation steps, repeat the operation until the system is stable, and output the distributed optimization results. Specifically, the loop iteration is carried out until all individual actions no longer change, then the coordination process ends and a near-optimal or optimal k-coverage collaborative grouping solution is obtained.

[0095] Figure 5 , Figure 6 and Figure 7 This is the distributed optimization convergence curve obtained by the present invention using different memory lengths. Figure 5 The memory lengths are 1, 2, 3, 4 and 5 respectively. Figure 6 The individual memory lengths are 6, 10 and 15 respectively. Figure 7The values ​​are 20, 25 and 30. Table 1 below shows the statistical results of 500 Monte Carlo simulations under different memory lengths:

[0096] Memory length Global evaluation function Convergence time (s) 1 28.5900 0.0045 2 28.6010 0.0105 3 28.6665 0.0111 4 28.6970 0.0222 5 28.7895 0.0267 6 28.8140 0.0443 10 28.9140 0.1115 15 28.9730 0.3243 20 29.0020 0.3953 25 29.0180 0.4812 30 29.0415 0.7333

[0097] Table 1 (Statistical table of distributed collaborative k coverage optimization results under different memory lengths)

[0098] Therefore, as the memory length increases, the algorithm convergence time becomes longer, and the global evaluation function at the obtained equilibrium point gradually increases. The above experimental results show that the distributed sensor cooperative k-coverage method proposed in the present invention can not only ensure the global convergence of Nash equilibrium, but also balance and compromise the optimality of the solution and the convergence time by adjusting the memory length, which means that the experiment is successful.

[0099] In summary, the present invention regards each node in the sensor network as a game participant with autonomous communication, calculation, decision-making and control capabilities. Through the construction of strategy sets and individual payment functions, the sensor network collaborative k coverage is transformed into a Nash equilibrium solution and selection problem in a potential game. By using limited memory and mixed response strategies, a synchronous random learning algorithm is designed, which can effectively guarantee the convergence and refinement of the Nash equilibrium solution. Therefore, the present invention does not rely on the distributed solution of global information and central nodes, eliminates the conflict between individual interests and global performance, and improves the service life of the system without losing network coverage performance. In addition, in order to meet the different requirements of different task scenarios for coordination time and collaborative efficiency, this method also provides a way to effectively balance and compromise the two. In addition, the present invention is suitable for autonomous solution of distributed wireless sensor network k coverage problems, and its distributed collaborative decision-making algorithm can also be transplanted to other distributed coordination control and optimization decision-making problems that can be described as potential games, including distributed satellite collaborative mission planning, resource allocation, formation control, etc.

[0100] The above is only one embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A sensor network self-organizing collaborative k-coverage method based on game theory, comprising the following steps: a. Taking the effective coverage area of ​​the system as an indicator, a global optimization mathematical model of collaborative k coverage is established; b. Establish a distributed collaborative decision-making model based on network potential game; c. Each sensor node determines its current location, coverage area, and neighbor set, and initializes its action and memory vectors; d. The sensor node uses the neighborhood information to calculate the individual payment value, the best response, the regret value and the non-inferior action set, and sends the results to the neighbors; In step d, during the tth coordination process, individual s i The local calculation is performed using the following steps: Use neighborhood communication to obtain the current action information of neighbor individuals Calculate the payment function value u i , best response Regret Value With non-inferior action set Among them, s i represents the i-th sensor node, t represents the number of iterations, Θ i Represents sensor node s i Coverage area; j For i Neighbor nodes, Ω i For i The neighbor set of A; \ is the set reduction operation; A i Represents sensor node s i The set of actions; represents time t divided by s i the combined actions of all other individuals; e. The individual uses the mixed response rule to select the restricted best response strategy and updates the memory vector; In step e, using local communication, individual s i Get neighbor's regret value information And calculate the maximum regret value in the neighborhood and the corresponding neighbors The best strategy for dealing with individual computing constraints For: If but If (i) or (ii) And i<j * ,but Otherwise, with equal probability, Choose any action as use Update individual memory, f. The individual randomly selects elements in the memory vector with the same probability and updates the current action; g. Jump back to step d, iterate until the system is stable, and output the distributed optimization results.

2. The method according to claim 1, characterized in that In step a, the set of sensor nodes randomly scattered in the two-dimensional plane area is W = {s i } i∈N ; Where N = {1, 2, ..., n}, s i represents the i-th sensor node, whose location coordinate is X i =(x i ,y i ), the coverage area is X i is the center of the circle, r i A circular area with radius Θ i ; Each sensor node i All are equipped with wireless communication modules with a communication radius of c i >2max i∈N r i ; If j In s i Within the communication range, s j For i Neighbors, denoted by node s i The neighbor set of i ={s j ∈W|||X i -X j ||≤c i },in, 3. The method according to claim 2, characterized in that The sensor network nodes are divided into k independent covering sets {S j } j∈K , where K = {1, 2, ..., k}, and the detection time is divided into k corresponding time slots {T j } j∈K ; In each time period T j Only the cover set S is activated j The sensors in the node are kept silent. Each node s i can choose to join any of the k covering sets, and its action set is A i =K, and its action is a i ∈A i , indicating s i Opt-in i A covering set; Define a=(a1,a2,...,a n ) is a group, A=Π i∈N A i is a grouping space. For any group a∈A, the jth covering set is represented by S j ={s i ∈W|a i =j}; The goal of sensor cooperative k-coverage is to maximize the following global objective function by selecting appropriate grouping a∈A: in, Denotes the covering set S j The union of the coverage areas of all nodes in .

4. The method according to claim 3, characterized in that In step b, each sensor node is regarded as an intelligent individual with perception, communication and computing capabilities, and a potential game model G(P,{A i },{u i The steps of},φ) are as follows: Establish the participant set P: Consider each sensor node in the system as an independent game participant, then the player participant set P = W; Create action set A i :Every individual has the same set of actions A i =K; Establish payment function u i :For any group or action combination a, the individual payoff function is defined as The calculation of this value depends only on the neighborhood information of the game individual; Establish potential function φ: For any grouping or action combination a, define the potential function as:

5. The method according to claim 1, characterized in that In step c, each sensor node s i Use positioning and communication functions to determine the current location X i =(x i ,y i ), coverage area Θ i And the neighbor set Ω i ; Each sensor node randomly initializes its actions With memory vector Where l is the memory length and the superscript t represents the number of iterations.

6. The method according to claim 1, characterized in that In step f, individual s i exist Randomly select with the same probability and get the updated action 7. The method according to claim 1, characterized in that In step g, the loop is iterated until all individual actions no longer change, then the coordination process ends and a near-optimal or optimal k-coverage cooperative grouping solution is obtained.

Images