External consensus in multi-agent systems with large consecutive data loss under unreliable networks

This study discusses the external consensus problem for non-identical networked multi-agent systems with network data loss, considering both uniform and random consecutive data losses induced by long periods of transmission failure. A novel approach using a combination of prediction strategy and constant control gain based on gain error ratio (GER) is proposed for solving the problem. The main contribution of this study is a simple and practical way of selecting control gain using GER to help overcome prediction inaccuracies caused by the prediction process during a long period of consecutive data loss. The feasibility and performance of the proposed consensus protocol is demonstrated through simulation and practical experiment.