Simulation Study on the Convergence Time of a Discrete Consensus Algorithm for Distributed Task Assignment

This paper presents results on the convergence time of a previously proposed algorithm to solve the distributed task assignment problem on a network of agents. The considered algorithm consists in iterative local integer linear optimizations among the agents to cooperatively assign a set of tasks. Local optimizations are performed among a subset of randomly chosen neigh boring agents. This paper discusses a simulation study on the number of iterations and time required by the algorithm to find the best task assignment given the limited information available at each iteration. Simulations on large and small networks with local optimizations that involve a varying number of agents have been performed.