Decentralized Pairwise-Workloads Balancing Algorithm for Networked Multi-agent Systems

This paper addresses the problem of allocating pairwise-workloads in networked multi-agent systems, where the cost of each agent increases linearly with the workload the agent takes on, aiming for balancing the costs of the agents over the network. To achieve this goal, we propose a pairwise balancing algorithm following a decentralized mechanism to locally coordinate the pairwise-workloads within each pair and to guarantee the coordination against infeasibility. Theoretical analyses reveal that the proposed algorithm converges and leads to a refined allocation with less difference of costs. Simulation examples are performed as well to demonstrate the effectiveness of the algorithm.