Abstract: |
This paper presents consensus control of a system of linear discrete time systems connected over a lossy communication network. Modeling the packet loss as a 2-state Markov chain, we describe the multiagent system as a dynamic system with Markovian jumping parameters. The control protocol is based on feedback of weighted subsystem state errors that achieves consensus in the nominal system without any packet loss. For the lossy network, we develop condition of consensusability that relates threshold of recovery rate with respect to packet loss rate. Consensus control of multiagent system under Bernoulli packet drop follows as a special case. The proposed method also shows the effects of connection weight of the network links on consensusability of the multiagent system. The results are illustrated by simulation of a multiagent system of multiple inverted pendulum connected over a four node communication network. |
|