A bipartite consensus method for switching multi-agent systems under deception attacks

By designing a controller affected by time-varying time lag and random spoof attacks, combining the switching symbol diagram and error system, the binary inclusive consistency control problem of multi-agent systems under spoof attacks is solved, and the stability and security of the system are achieved , suitable for a variety of leader node situations.

CN118502236BActive Publication Date: 2024-11-19NANTONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410390794.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-02
Publication Date
2024-11-19
Estimated Expiration
2044-04-02

AI Technical Summary

Technical Problem

The existing technology has failed to effectively solve the problem of binary inclusive consistency control of multi-agent systems under spoofing attacks, especially in the presence of network topology switching and time lag, which leads to the loss of stability and security of the system being threatened.

Method used

A binary consistency method for switching multi-agent systems under spoof attacks is proposed. By establishing a switching multi-agent system model, a controller affected by time-varying time delay and random spoof attacks is designed, and a switching symbol diagram and error system is combined to realize The mean square asymptotic stability of the error system ensures that the leadership follows the binary consistency.

Benefits of technology

In the sense of mean square asymptotic stability of the error system, this method realizes the binary consistency control of multi-agent system under deception attacks. It is suitable for binary tracking consistency and binary inclusive consistency, and has higher universality and applicability. It can maintain the system stability under the combined influence of network inducing factors.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118502236B_ABST
    Figure CN118502236B_ABST
Patent Text Reader

Abstract

The present invention discloses a bipartite consistency method for switching a multi-agent system under a deception attack, which belongs to the technical field of consistency of a multi-agent system, solves the technical problem of bipartite consistency of a multi-agent system with a cooperative-competitive interactive relationship and network topology switching under a deception attack, and its technical solution includes the following steps: firstly, a switching multi-agent system model is established; then, a controller affected by time-varying time delay and random deception attack is designed; then, an error system between a leader node and a follower node in the multi-agent system is constructed; finally, a bipartite consistency condition for switching a multi-agent system under a deception attack is established. The beneficial effects of the present invention are: comprehensively considering the influence of network-induced factors such as switching symbolic networks, time-varying time delays and deception attacks, a controller is designed to simultaneously achieve bipartite tracking consistency and bipartite inclusion consistency of a switching multi-agent system under a deception attack in the sense of mean square asymptotic stability of the error system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of multi-agent system consistency, and in particular to a binary consistency method for switching a multi-agent system under a deception attack. Background Art

[0002] In the past two decades, the cooperative control of multi-agent systems has attracted widespread attention due to its increasingly broad application prospects in various fields such as drones, self-driving cars, spacecraft flight, smart grids, urban transportation, etc. Consensus, as the most common and most critical cooperative control, has become a research hotspot. In practical engineering applications, cooperation-competition interactions coexist in multi-agent systems, and bipartite consistency comes into being.

[0003] The bipartite consensus of multi-agent systems mainly focuses on the study of the two cases with and without a leader. The bipartite consensus problem with a leader is usually described as the bipartite tracking consistency of a single leader. However, in many engineering applications, multiple leaders often appear, such as position and navigation, environmental perception, and human-computer interaction, which is usually called the bipartite inclusion consistency problem. Therefore, according to the number of leaders, the leader-following bipartite consensus of multi-agent systems can be divided into the bipartite tracking consistency of a single leader and the bipartite inclusion consistency of multiple leaders.

[0004] The network topology of multi-agent systems is not fixed in practical engineering applications, so it is more practical to study the bipartite consistency with a switching topology. At the same time, the existence of time delay is inevitable, just like the existence of network topology switching, which will cause the system to lose stability and inevitably bring difficulties to the research. It can be seen that under the network topology switching, the bipartite consistency problem of multi-agent systems with node time delay and control input time delay is of great significance and challenging. At the same time, multi-agent systems often face the risk of security attacks during data transmission, which means that the designed controller is extremely vulnerable to malicious attacks, such as deception attacks and denial of service attacks. Few studies have focused on the problem of bipartite containment consensus control of multi-agent systems under network attacks. For example, by using a sampled data event-triggered control mechanism, the problem of secure bipartite tracking consensus of linear multiagent systems under DoS attacks was studied. Based on a control scheme based on historical information, the problem of bipartite containment of heterogeneous multi-agent systems under denial-ofservice attacks: A historical information-based control scheme was studied. There is currently no report on the problem of bipartite containment consensus control of multi-agent systems under deception attacks. In this case, the present invention proposes a bipartite consensus method for switching multi-agent systems under deception attacks. Summary of the invention

[0005] In order to solve the technical problem of bipartite consistency of a multi-agent system with cooperative-competitive interaction and network topology switching under deception attacks, the present invention proposes a bipartite consistency method for switching multi-agent systems under deception attacks.

[0006] The present invention is achieved by the following measures: A method for switching the bipartite consistency of a multi-agent system under a deception attack comprises the following steps:

[0007] Step 1: Establish a switching multi-agent system model

[0008] Consider a bipartite consensus problem in a multi-agent system with R leader nodes and NR follower nodes, where the leader node set Follower node set And R ≥ 1, the multi-agent system may have a single leader or multiple leaders.

[0009] The dynamic equation of the intelligent agent node i is

[0010]

[0011] in represents the state vector of the i-th node, u i (t) is the control input of the ith node, matrix and are constant matrices, f(·) and f d (·) are nonlinear functions without time delay and with time delay, respectively. The time-varying time delay τ(t) satisfies and is a constant,

[0012] The network topology of the multi-agent system is switched, as shown by the switching symbol diagram Describes the cooperation-competition relationship between intelligent nodes, where is the set of nodes, ε r(t) is the set of edges, is the weighted adjacency matrix, r(t) is the switching signal, Assuming that each leader node has no neighbor nodes, switch the symbol graph The adjacency matrix of The corresponding Laplace matrix is ​​expressed as The Laplace deformation matrix is ​​expressed as in

[0013] Step 2: Design the controller

[0014] In order to realize the bipartite consistency control of the multi-agent system, the network-induced factors such as transmission delay and security attack in the network environment are comprehensively considered, and a controller affected by time-varying delay and random deception attack is designed, which is expressed as follows:

[0015]

[0016] in and is the controller gain, is the deception attack function, β(t k ) is the probability function of a spoofing attack. It is used to describe the time series when a deception attack may occur, 0 = t0 < t1 < ... < t k <..., and t k -tk-1 =h>0,

[0017] According to the controller, the dynamic equation of follower node i is expressed as

[0018]

[0019] in

[0020] Step 3: Construct the error system between the leader node and the follower node in the switching multi-agent system

[0021] Using canonical transformation The formula (1) of the multi-agent system is transformed into the following form,

[0022]

[0023] When the agent node i belongs to the binary node subset When i =1, otherwise, φ i =-1,

[0024] make

[0025] Error signal

[0026] The error system in the form of a Kronecker product is obtained,

[0027]

[0028] in,

[0029] Step 4: Establish the bipartite consistency condition of the switching multi-agent system under deception attack. Under the meaning of mean square asymptotic stability of the error system, bipartite tracking consistency is achieved for the switching multi-agent system with a single leader, and bipartite inclusive consistency control is achieved for the switching multi-agent system with multiple leaders:

[0030] Given scalars α>ω>0, ε>1, if there exists a matrix P p >0, The positive scalars δ1, δ2 and r satisfy the following inequalities,

[0031]

[0032] in And η is the equation The only solution is

[0033] Then, under the action of the controller, the leader-follower bipartite consistency of the switching multi-agent system can be achieved.

[0034] Furthermore, the f(·) and f described in step 1 of the bipartite consensus method for switching multi-agent systems under deception attacks d (·) are nonlinear functions without time delay and with time delay, respectively, which means that the nonlinear functions f(·) and f d (·) is an odd function, and for any satisfy

[0035]

[0036]

[0037] Where σ1>0 and σ2>0 are known constants, ρ i ≥0,

[0038] Furthermore, the switching symbol graph described in step 1 of the bipartite consistency method for switching a multi-agent system under a deception attack Describe the cooperation-competition relationship between intelligent nodes, where the switching symbol diagram Refers to the switch symbol diagram Structurally balanced, with bipartite non-empty node subsets and Satisfies: When node i and node j belong to the same subset, then a ij ≥0, otherwise a ij ≤0, and for any switching mode symbol diagram Two node subsets and It remains the same that there is at least one leader node with a path to all follower nodes.

[0039] Furthermore, in a bipartite consensus method for switching a multi-agent system under a deception attack, the r(t) in step 1 is a switching signal, which means that the switching signal r(t) is a homogeneous Markov process of a continuous-time discrete state, and takes values ​​in a finite state set The transition probability of the switching signal r(t) from switching mode p to switching mode q is as follows,

[0040]

[0041] Where Δ>0 is the stay time, represents the transition probability, and The transition probability matrix is ​​κ = [μ pq ] S×S .

[0042] Further, the bipartite consistency method for switching multi-agent systems under deception attacks described in step 1 is the deception attack function, which means is a nonlinear odd function and satisfies the following inequality,

[0043]

[0044] Furthermore, the error signal described in step 1 of a bipartite consensus method for switching a multi-agent system under a deception attack It means that for any switching signal r(t), when R>1, are all non-negative matrices with row sums of 1, and the error signal in is the convex hull formed by the leader node state; when R = 1, is a single-column matrix with all entries equal to 1. The error signal

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) The present invention takes into account the comprehensive impact of network induction factors, such as switching symbol networks, time-varying delays, and deception attacks, and is more universal and applicable than existing methods that only consider a single network induction factor.

[0047] (2) In the sense of mean square asymptotic stability of the error system, this study proposes a bipartite consensus method for a switching multi-agent system under deception attacks, which is applicable to both bipartite tracking consistency and bipartite containment consistency control. It has greater research and application value than the existing separate bipartite tracking consistency research of switching multi-agent systems (“Delays-based distributed bipartite consensus control of nonlinear multiagent systems with switching signed topologies”) and bipartite containment consistency research of switching multi-agent systems (Bipartite containment control for multi-agent systems under fixed and markov switching topologies). BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a schematic diagram of a bipartite consistency method for switching a multi-agent system under a deception attack according to the present invention.

[0049] Figure 2is a symbolic diagram describing the topological structure of the multi-agent system in Example 1 of the present invention, wherein (a) is a symbolic diagram (b) is the symbol diagram (c) is the symbol diagram

[0050] Figure 3 1 is a diagram of a switching signal and a deception attack signal in Example 1 of the present invention, wherein (a) is a diagram of a switching signal; and (b) is a diagram of a deception attack signal.

[0051] Figure 4 is the state x of the multi-agent system in Embodiment 1 of the present invention i (t) trajectory diagram, where (a) is the state x i1 (t) trajectory diagram; (b) is state x i2 (t) trajectory diagram; (c) is state x i3 (t) Trajectory diagram, i=1,…,7.

[0052] Figure 5 is the state error of the multi-agent system in Example 1 of the present invention ||e(t)|| 2 Trajectory map.

[0053] Figure 6 1 and 2 are diagrams of switching signals and deception attack signals in Embodiment 2 of the present invention, wherein (a) is a diagram of switching signals and (b) is a diagram of deception attack signals.

[0054] Figure 7 is the state x of the multi-agent system in Embodiment 2 of the present invention i (t) trajectory diagram, where (a) is the state x i1 (t) trajectory diagram; (b) is state x i2 (t) Trajectory diagram. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Of course, the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0056] As attached Figure 1 As shown, consider the bipartite consensus problem of a multi-agent system with R leader nodes and NR follower nodes, where the leader node set Follower node set And R ≥ 1, the multi-agent system may have a single leader or multiple leaders.

[0057] The dynamic equation of the intelligent agent node i is

[0058]

[0059] in represents the state vector of the i-th node, u i (t) is the control input of the ith node, matrix and are constant matrices, f(·) and f d (·) are nonlinear functions without time delay and with time delay, respectively. The time-varying time delay τ(t) satisfies and is a constant.

[0060] Assumption 1: Nonlinear functions f(·) and f d (·) is an odd function, and for any satisfy

[0061]

[0062] Among them, σ1>0 and σ2>0 are known constants, ρ i ≥0,

[0063] The network topology of the multi-agent system is switched, consisting of S fixed-mode symbolic networks, and each fixed-mode symbolic network can be represented by a symbolic graph. to describe, where r(t) is the switching signal and the nodes in the symbolic network of each mode remain unchanged.

[0064] Let the switching signal r(t) be a continuous-time discrete-state homogeneous Markov process with values ​​in a finite state set The transition probability of the switching signal r(t) from switching mode p to switching mode q is as follows,

[0065]

[0066] Where Δ>0 is the stay time, represents the transition probability, and The transition probability matrix is ​​κ = [μ pq ] S×S .

[0067] Assuming that the network topology of the multi-agent system obeys Markov switching, then switching the multi-agent system can be done using the switching symbol diagram To describe the cooperation-competition relationship between nodes.

[0068] Assumption 2: For any r(t)∈{1,2,...,S}, the symbolic graph of each switching mode is The structure is balanced.

[0069] Assumption 3: For all switching modes, two node subsets and remains unchanged, and there exists at least one leader node with a path to all follower nodes.

[0070] Assume switching symbol diagram Each leader node has no neighbor nodes, so the adjacency matrix It can be expressed as The corresponding Laplacian matrix can be expressed as in According to Assumption 2, the diagonal matrix φ=diag{φ1,φ2,...,φ N} can be defined as: If Then φ i =1; otherwise, φ i = -1. have in

[0071]

[0072] Lemma 1: Under Assumptions 2 and 3, is the M matrix, and for each switching mode r(t), when R>1, are all non-negative matrices with a row sum of 1. When R = 1, is a single-column matrix with all entries equal to 1.

[0073] In order to realize the bipartite consistency control of multi-agent systems, a controller affected by time-varying delay and random deception attack is designed by comprehensively considering network-induced factors such as time delay and security attack in the network environment, which is expressed as follows:

[0074]

[0075] in and is the controller gain, It is used to describe the time series when a deception attack may occur, 0 = t0 < t1 < ... < t k <.... Assuming that the time series is equally spaced, t k -t k-1 =h, where h>0 is a known constant. Function β(t k ) follows a Bernoulli distribution, and the probability of its value being 0 or 1 can be expressed as prob{β(t k )=1}=θ and prob{β(t k )=0}=1-θ. is the deception attack function.

[0076] Assumption 4: Nonlinear Function is an odd function and satisfies the following inequality,

[0077]

[0078] According to the controller, the dynamic equation of follower node i can be expressed as

[0079]

[0080] in

[0081] Using canonical transformation Then the formula (14) of the multi-agent system can be converted into the following form,

[0082]

[0083] Notice that

[0084]

[0085] make

[0086]

[0087] Available,

[0088]

[0089] Re-order

[0090]

[0091]

[0092] Therefore, the error system can be expressed in the form of a Kronecker product as follows

[0093]

[0094] Definition 1: For any If satisfied

[0095]

[0096] in For multiple leaders, co{φ l x l (t)} is composed of multiple leader node trajectories φ l x l (t) forms a convex hull, and for a single leader, co{φ l x l(t)}=φ1x1(t) is the trajectory of a single leader node,

[0097] Multi-agent systems can achieve leader-follower bipartite consensus.

[0098] Note 1: Known error status From Lemma 1, we can see that when the multi-agent system has multiple leader nodes (R>1), Equivalent to the convex hull corresponding to each follower node When the multi-agent system has multiple leader nodes (R=1), is equivalent to the leader node's trajectory φ1x1(t), so, according to Definition 1, when When , the multi-agent system can achieve leader-follower bipartite consistency.

[0099] Theorem 1. Under Assumptions 1 to 4, given scalars α>ω>0, ε>1, if there exists a matrix P p >0, The positive scalars δ1, δ2 and r satisfy the following inequalities,

[0100]

[0101] in And η is the equation The only solution is

[0102] Then, under the action of the controller, the leader-follower bipartite consistency of the multi-agent system can be achieved.

[0103] Implementation examples.

[0104] Example 1. Consider a multi-agent system consisting of 2 leader nodes and 5 follower nodes. The system has a Markov switching inclusive controller. The symbolic graph corresponding to its 3 switching symbolic networks is shown in the attached figure. Figure 2 Symbol diagram and are all structurally balanced and have the same subset of bipartite nodes and

[0105] The system related parameters are given as follows:

[0106] C1=-2I3,C2=-0.2I3,

[0107] f d (x i (t)) = f(x i (t)), time lag function The transition probability matrix is The probability of a deception attack is θ = 0.3, and the deception attack function is By the attached Figure 2 Available,

[0108]

[0109] Then we can get Υ1=0.1 and Υ2=14.5145.

[0110] Let α = 4, ω = 2.5 and Then the equation The only solution is η = 0.3723. At the same time, let ε = 1.4 and h = 1, then Using MATLAB's CVX toolbox to solve linear matrix inequalities (24) to (27), we can obtain

[0111]

[0112] also and They are [492, 493, 488] and [5.5, 4.5, 4.5] respectively.

[0113] Attached Figure 3 The switching signal and deception attack signal, the system state trajectory and the error state ||e(t)|| are described respectively. 2 The trajectories are Figure 4 and Figure 5 As shown in the figure, the state of the follower node is always in the convex hull formed by the leader node, and the error converges asymptotically to 0. At the pulse moment of the deception attack and the network topology switching moment, even if the node state will be affected and jump, the state of the follower node is still in the convex hull, and the error state ||e(t)|| 2 The trajectory remains stable, indicating that the multi-agent system achieves bipartite inclusion consistency.

[0114] Example 2: Consider a multi-agent system consisting of 1 leader and 5 followers. The switching symbol network of this system is shown in the attached figure. Figure 2 , the leader node is node 1.

[0115] The system related parameters are given as follows:

[0116] The deception attack function is The remaining parameters are the same as in Example 1. By solving linear matrix inequalities (24) to (27), we can obtain

[0117]

[0118] also, and They are [192, 199, 190] and [3.5, 3.8, 3.0] respectively.

[0119] Attached Figure 6 The switching signal and the deception attack signal are described, and the state trajectory is shown in the attached Figure 7 shown.

[0120] By the attached Figure 7 It can be seen that the state trajectory of the follower node is asymptotically synchronized with the state trajectory of the leader, or the state trajectory with the same modulus but opposite sign as the leader. At the pulse moment of the deception attack and the symbol network switching moment, even if the node state will be affected accordingly and jump, it will still remain synchronized afterwards, indicating that the multi-agent system with a single leader node achieves bipartite tracking consistency.

[0121] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A bipartite consistency method for switching multi-agent systems under deception attacks, characterized in that: The steps include: Step 1: Establish a switching multi-agent system model Consider a bipartite consensus problem in a multi-agent system with R leader nodes and NR follower nodes, where the leader node set Follower node set And R ≥ 1, the multi-agent system may have a single leader or multiple leaders. The dynamic equation of the intelligent agent node i is: in represents the state vector of the i-th node, u i (t) is the control input of the ith node, matrix and are constant matrices, f(·) and f d (·) are nonlinear functions without time delay and with time delay, respectively. The time-varying time delay τ(t) satisfies and is a constant, The network topology of the multi-agent system is switched, as shown by the switching symbol diagram Describes the cooperation-competition relationship between intelligent nodes, where is the set of nodes, ε r(t) is the set of edges, is the weighted adjacency matrix, r(t) is the switching signal, Assuming that each leader node has no neighbor nodes, switch the symbol graph The adjacency matrix of The corresponding Laplace matrix is ​​expressed as The Laplace deformation matrix is ​​expressed as in Step 2: Design the controller In order to realize the bipartite consistency control of the multi-agent system, the network-induced factors such as transmission delay and security attack in the network environment are comprehensively considered, and a controller affected by time-varying delay and random deception attack is designed, which is expressed as follows: in and is the controller gain, is the deception attack function, β(t k ) is the probability function of a spoofing attack. It is used to describe the time series when a deception attack may occur, 0 = t0 < t1 < ... < t k <..., and t k -t k-1 =h>0, According to the controller, the dynamic equation of follower node i is expressed as in Step 3: Construct the error system between the leader node and the follower node in the switching multi-agent system Using canonical transformation The formula (1) of the multi-agent system is transformed into the following form, When the agent node i belongs to the binary node subset When i =1, otherwise, φ i =-1, make Error signal The error system in the form of a Kronecker product is obtained, in, Step 4: Establish the bipartite consistency condition of the switching multi-agent system under deception attack. Under the meaning of mean square asymptotic stability of the error system, bipartite tracking consistency is achieved for the switching multi-agent system with a single leader, and bipartite inclusive consistency control is achieved for the switching multi-agent system with multiple leaders: Given scalars α>ω>0, ε>1, if there exists a matrix P p >0, The positive scalars δ1, δ2 and r satisfy the following inequalities, in And η is the equation The only solution is Then, under the action of the controller, the leader-follower bipartite consistency of the switching multi-agent system can be achieved.

2. According to claim 1, a bipartite consistency method for switching a multi-agent system under a deception attack is characterized by: In step 1, the f(·) and f d (·) are nonlinear functions without time delay and with time delay, respectively, which means that the nonlinear functions f(·) and f d (·) is an odd function, and for any satisfy Where σ1>0 and σ2>0 are known constants, ρ i ≥0, 3. According to claim 1, a bipartite consistency method for switching a multi-agent system under a deception attack is characterized by: In step 1, the switching symbol diagram Describe the cooperation-competition relationship between intelligent nodes, where the switching symbol diagram Refers to the switch symbol diagram Structurally balanced, with bipartite non-empty node subsets and Satisfies: When node i and node j belong to the same subset, then a ij ≥0, otherwise a ij ≤0, and for any switching mode symbol diagram Two node subsets and It remains the same that there is at least one leader node with a path to all follower nodes.

4. According to claim 1, a bipartite consistency method for switching a multi-agent system under a deception attack is characterized by: In step 1, the r(t) is a switching signal, which means that the switching signal r(t) is a homogeneous Markov process of continuous-time discrete state, and its value is in the finite state set The transition probability of the switching signal r(t) from switching mode p to switching mode q is as follows, Where Δ>0 is the stay time, represents the transition probability, and The transition probability matrix is ​​κ = [μ pq ] S×S .

5. According to claim 1, a bipartite consistency method for switching a multi-agent system under a deception attack is characterized by: In step 1, is the deception attack function, which means is a nonlinear odd function and satisfies the following inequality, 6. The method for switching bisection consistency of a multi-agent system under a deception attack according to claim 1, characterized in that: In step 1, the error signal It means that for any switching signal r(t), when R>1, are all non-negative matrices with row sums of 1, and the error signal in is the convex hull formed by the leader node state; When R=1, is a single-column matrix with all entries equal to 1. The error signal