Stochastic Consensus Of Multi-Agent Systems With Noisy And Delayed Communication

Stochastic consensus of multi-agent systems is a main and an important research subject in distributed cooperative control. This is because the information shared in a network is always prone to some uncertainties, which can destroy the performances and the behavior of the entire system.In this thesis, we will solve some distributed stochastic consensus problems of continuous-time single- and double-integrator multi-agents systems under two communication constraints. We will take into account the measurement/communication noises and the time-delays. That is, each agent in the team will measure or receive inaccurate information states from its neighbors corrupted by noises, and sometimes this information states are also delayed. The noises are considered non-state dependent or relative-state dependent white noises, and the time-delays are considered uniform.Throughout the thesis, seven consensus protocols have been proposed; Two consensus protocols for double-integrator MASs under non-state dependent measurement/communication noises. Two consensus protocols for single-integrator MASs under relative-state dependent measurement/communication noises. Two consensus protocols for single-integrator MASs under both non-state dependent measurement/communication noises and time-delays, and a consensus protocol for single-integrator MASs under both relative-state dependent measurement /communication noises and time-delays. For the consensus problems under non-state dependent measurement/communication noises, a time-varying consensus gain that satisfies the stochastic approximation type conditions is used to attenuate the noises. While, for the consensus problems under relative-state dependent measurement/communication noises, a constant consensus gain is sufficient. The Lyapunov based approach is adopted to study the stability and the convergence of our systems by employing some tools from algebraic graph theory, matrix theory and stochastic. In most of our consensus protocols, and to simplify the analysis of the consensus problem, the approach followed is mainly based on the decomposition of the original stochastic differential equations describing the closed-loop of the multi-agent systems into two decoupled parts by using appropriate state transformation matrices.The effectiveness of the consensus protocols has been proved throughout simulation examples.