Privacy-Preserving Cooperative-Tracking Control via Switching Heterogeneous Directed Graphs

Distributed cooperative tracking control has been attracting an increasing number of interests recently with wide application scenarios. It entails faster responses to its neighboring node and is more flexible than a centralized one. However, the leakage of information due to direct exchanges of sensitive data among individual agents may hinder the approach from being adopted in a broader scope of applications. In this talk, I will introduce my recent work: privacy-preserving cooperative tracking control in multi-agent systems, which aims for cooperative-tracking consensus without leaking the state information of the participating agents. We consider a general directed pairwise communication topology, where the weights among individual agents are time-variant either spontaneously or adjustably. We then derive the condition for the control to reach cooperative-tracking consensus and show the relationship between the asymptotic convergence factor and the choice of control gain. After that, we analyze the privacy-preserving performance against eavesdroppers. We prove that the deviation of each node’s state can be arbitrarily large while the received information by the adversaries remains the same, which means that the privacy of nodes’ states is preserved against adversaries. Numerical examples are shown to demonstrate the effectiveness of our proposed algorithm.