Abstract: In this paper, we study the problem of coalition formation for task allocation in agent networks, in particular, we consider the situation where the profits of tasks are elapsed with time. A novel distributed algorithm is proposed, by designing a “quit-to-choose” mechanism, agents can make decisions with more rationality, so that more coalitions can be formed and the overall utility of the system is improved. Our algorithm is decentralized and it only requires local information about other agents. In addition, the complexity of the proposed algorithm is analyzed to show that it is easy to be implemented. We then conduct…a variety of experiments in four different networks to verify the performance of the proposed algorithm, the results demonstrate that compared with the greedy mechanism, our algorithm can highly improve the system profit.
Show more