Impact of communication constraints on consensus finding in multi-agent systems
Niklas Goddemeier, Daniel L. Behnke, Christian Wietfeld · 2013
In this work we examine the influence of harsh communication conditions on cooperative consensus finding in multi-agent systems. We investigate the impact of network topology on the convergence behavior of cooperative consensus finding. Analysis of the effect of packet error rates on convergence speed and convergence error are carried out in simulations. We propose a model to determine the best number of communication links an agent should maintain in order to find the best tradeoff between spatial coverage and convergence speed. We use these results in our communication-aware mobility algorithms for unmanned aerial systems to actively influence the vehicles' network topology. Experimental implementations show that already small amounts of packet errors disable the agents to find the correct average consensus using well-known consensus algorithms. Consequently, we propose a novel consensus protocol which reliably converges to the average value of the initial agent states. Our results have been obtained using simulations and a hardware testbed.