Gossip algorithm for multi-agent systems via random walk

This paper proposes an asynchronous gossip framework where agents move according to independent random walks over a location graph and interactions may occur only when two agents share the same location. Our goal is to investigate how average consensus may be achieved when agents’ motion occurs over a set of discrete locations with topological constraints. This could be used to model the spreading of information across moving crowds or the coordination of agents monitoring a discrete set of points of interest.