Consensus And Distributed Optimization Of Multi-agent Systems With Constrained Communication

The information; communication in multi-agent systems is usually constrained by some factors, such as the data block length and communication channel width. The stagnation of information can occor in the process of the information communication, and it leads to the time-delay. The limited consensus of the multi-agent systems with communication delay is investigeted in the present work. The digital channel of wireless communication has limited bandwidth, which can only transmit limited data among individuals. Therefore, the problem of information quantification is a realistic problem that must be considered. Based on this, the main work of the paper is as follows:The first part is the study of the limitation of the multi-agent system with communication delay. Assume that each agent has a first-order continuous dynamic "behavior and its state is constrained to a convex set, and multi-agent systems composed of network topology is dynamically switching. Under the condition that the communication delay is fixed and bounded, the projection consensus protocol can not only-ensure that the state of each agent can reach consensus, but also prove that the final consensus value is located in the intersection of all limited sets of each agent.The Lyapunov-Krasovskii function is used to analyze the uniform convergence of the multi-agent system in this part.The second part is the distributed optimization of multi-agent system quantitative information communication. Due to the actual digital channel usually has a limited bandwidth, we investigate the distributed optimization problem based on the individual dynamics of continuous time first-order integrator form on switching network by using uniform quantizer. This part set the objective function of the whole system as a sum of all the objective functions of agents, each agent only knows its own objective function and exchanges quantitative information with its neighbors. It eventually allows all agents reach consensus and the network objective function to achieve optimum. By constructing the appropriate Lyapunov function, the convergence results are given and the limit of the error bounds is obtained when the quantization information communication and the collective optimization problem with the regularity constraints are presented. Our investigated results show that-when the communication interval and the fixed delay are all bounded, of the coherence protocol proposed in the first part makes each individual state to reach a final consensus, and final consensus value is located in the intersection of all limited sets of each agent. And under the effect of a uniform quantizer, the consensus algorithm which the second part gives is convergent and the delay does not change the convergence of the algorithm.