Average-consensus problem in multiagent systems

This paper is mainly devoted to the average-consensus problem among multiple agents in directed networks with fixed or switching balanced topology. To this end, both discrete-time and continuous-time consensus schemes are proposed. More specifically, a necessary and sufficient condition for reaching average-consensus is first given in networks with fixed topology, which can be reasonably interpreted by employing the theory of Markov process. Then, for the case of switching topology, it is shown that average-consensus can be asymptotically achieved if the union of the balanced directed interaction graphs is weakly connected across each finite time interval. In addition, for undirected networks with fixed topology and time-varying delays, average-consensus problem is discussed and an upper bound of delay tolerance is given to ensure the consensus. Finally, simulations illustrate the effectiveness of the obtained results.