Consensus of Networked Multi-agent Distributed Detection

Cooperative target detection, including detection and data fusion stage, can improve detection performance by utilizing observation diversity. Consensus of distributed cooperative target detection under self-organizing ad hoc networks is the basis of multi-agent cooperation. Convergence rate of consensus-based data fusion is crucial for time-sensitive mission, which is affected by both detection parameters and communication network parameters. The second smallest eigenvalue of graph Laplacians called algebraic connectivity quantifies the convergence rate of consensus algorithms, which is a relative concept yet. The absolute quantity of convergence rate and the relationship with parameters are necessary for performance optimizations. In this paper, the sufficient condition for consensus convergence of multi-agent detection systems is derived. And contrast experiments show effectiveness of the sufficient condition. Then we derive and analyze convergence rate of the error dynamic system under given convergence accuracy, and give the relationship between parameters and performance by Monte Carlo simulations, which is an important theoretical support for parameter selections.