Quantized Consensus Of Multi-agent Systems

In the past decade,inspired by great potential in various engineering applications,multi-agent systems have been widely studied.The consensus problem with the objective for all agents to reach a common goal is a fundamental problem in the study of multi-agent systems,and has received considerable attention.In most early works on the consensus problem,the control law of each agent was designed based on the exact state information of itself and its neighborhood.However,implementation of such controllers is almost impossible in practice,due to the limited capacity of the communication channel between each pair of agents.In fact,the communication between each pair of agents in practice is often realized by the following three steps:?)the state information of the sender is encoded into a binary sequence;?)the binary sequence is transmitted by a channel;and iii)the neighbor uses the encoded information to estimate the sender's state.By the above process,it is obvious that quantization cannot be neglected in communication of multi-agent systems.This thesis studies quantized consensus for general linear multi-agent systems.The main results of this thesis can be summarized as follows.Firstly the consensus problem of general linear multi-agent systems over undi-rected graph is considered,under the communication constraint of limited band-width.Both periodic and event-triggered sampling are considered.By constructing a novel dynamic quantizer,a distributed protocol via periodical sampled and quan-tized data is first designed to solve this problem.It is shown that the required number of the quantization levels of the new quantizer remains to be small even if the number of the agents in the multi-agent system is large.Then,a novel event-triggered distributed control protocol is proposed based on a new dynamic quantizer to solve problem.Compared with periodical sampling for quantized consensus,the proposed event-triggered control approach can significantly save communication and thus energy resource.It is shown that with the proposed distributed control proto-col quantized consensus of the concerned multi-agent system can be achieved,and the Zeno behavior and continuous monitoring of the neighbors states are avoided.Secondly the quantized consensus problem of general linear multi-agent systems over directed graph is considered.Both discrete-and continuous-time cases are con-sidered.Based on the idea of assigning a priority level for each agent of the concerned discrete-time multi-agent system,a novel distributed control law is proposed.The proposed control law has two distinctive advantages:(?)it is fully distributed in the sense that it does not rely on the eigenvalues of the Laplace matrix associated with the topology and(?)the required data rate is independent of the number of agents and remains small even if the number of the agents in multi-agent systems is large.A distributed protocol via sampled and quantized data is designed to solve this prob-lem for continuous-time multi-agent systems.Compared with the existing works on quantized control of continuous-time multi-agent systems,not only the semi-global assumption on initial conditions is not required,but also the communication graph is not required to be balanced or strongly connected.Finally the cooperative output regulation problem of multi-agent systems over switching graphs,under the communication constraint of limited bandwidth,is con-sidered.A distributed protocol via sampled and quantized data is designed to solve this problem.Compared with the existing works on quantized control of MASs,the semi-global assumption on initial conditions is not required,but also the communi-cation graph is assumed to be jointly connected.