Distributed consensus of multi-agent systems over general directed networks with limited bandwidth communication

This paper investigates the problem of distributed average consensus over general directed digital networks, in which communication channels have limited data transmission rate. Based on an encoder–decoder scheme and a zoom-in technique, a kind of consensus algorithm is presented. Considering the quantization levels of each connected directed communication channel are time-varying at each time step, a rigorous theoretical analysis is performed and the average consensus can be achieved asymptotically. The upper bound of the quantization levels at each time step and the convergence rate are analytically characterized. Moreover, a theorem is also obtained for the case of fixed quantization level. Two numerical examples are given to illustrate the correctness of the theoretical analysis.