Privacy-preserving bipartite flocking of networked agent systems subject to dynamic competitive interactions and random link failures

Flocking refers to the self-organized coordination of agents’ movements, where ensuring privacy protection and managing random link failures are crucial for maintaining secure communication and robust functionality, especially in dynamic and decentralized systems. To satisfy the needs, this paper investigates the secure bipartite flocking control problem for networked agent systems over signed networks, considering eavesdropping attacks and random link failures. The privacy encryption constraint addresses the potential risk of external eavesdroppers gaining access to communication links, which could compromise the entire network, leading to significant information leakage and threatening the security of networked control. To mitigate these risks, a secure distributed control scheme is proposed, utilizing time-varying noise to ensure bipartite flocking behavior while safeguarding the positions and velocities of agents. Based on positional information augmented with noise, a weight function is introduced to dynamically adjust communication edge weights, reflecting variations in interaction strengths with fewer constraints. Furthermore, the algebraic relationship between the probability of random link failures and system stability is established, and the necessary topological conditions for achieving bipartite flocking control are obtained using the convergence method based on infinite products of sub-stochastic matrices. Finally, numerical simulations validate the effectiveness of the proposed secure bipartite flocking control scheme.