Research And Application Of Trusted Link Based Resilient Consensus Algorithms For Multi-Agent Systems

The consensus problem of multi-agent systems has become an academic hotspot,mainly owing to its wide applications such as distributed computing,load balancing,and formation control.However,few existing studies on resilient consensus consider the situation that the communication link is under attack and the network topology requirements for most resilient consensus algorithms are too harsh,thus how to achieve the resilient consensus under the attack is an urgent problem to be solved.Motivated by the research of previous works,this thesis further studies on the distributed resilient consensus algorithms under the false data injection attacks,and applies the algorithm to its application.Firstly,the concept of the trusted link is introduced to improve the fault-tolerance capability of the topology and the corresponding selection method is given.Combined with the attack problem,a new property named network robustness which describes the connection structure is used to evaluate topology fault-tolerance capability.The theory analysis indicates that by setting a part of the links to be trusted of the connected network topology,the specified network robustness can be reached.Secondly,a resilient consensus algorithm for static systems is proposed.The core of the algorithm is the design of the fault tolerance mechanism.Specifically,after sorting the received value,the agent removes some extreme values with respect to its own while preserving the data transmitted on the trusted link and uses the remaining data to update its state.Then,the necessary topology conditions for resilient consensus are given.Compared with the traditional algorithm,the proposed algorithm reduces the requirements of the topology structure so that could be adapted to the sparse network.Thirdly,a resilient consensus algorithm for time-varying systems is proposed.By considering the data received over a period to ease the problem of asynchronous update of the agent and performing data filtering based on the range of the safety interval,the resilient consensus to reach a final state value is ensured.Then the necessary topological condition for the system to achieve resilient consensus is given and proved,that is,to achieve resilient consensus with the proposed algorithm under the F-local attack,the network robustness must satisfy(K,2F + 1)-trusted robustness.Simulation results show the effectiveness of the proposed algorithm.Finally,based on the algorithm proposed in the previous work,the time synchronization problem among sensor nodes is considered.With the help of the linear update model of the clock,a relative physical drift ratio of each node is calculated by using the recent two data received from the same neighbor,and then a consensus-based time synchronization algorithm is proposed,which realizes the calibration of the deviated parameters of nodes in the system under the attack and thus achieves the time synchronization.The theory analysis confirms the feasibility of the proposed algorithm and the experiments show that the proposed algorithm can achieve time synchronization between sensor nodes.