Multi-task Multi-coalition Generation Problem: Model and Algorithm

Coalition formation has been a very active area of research in MAS. The existing work mainly researches how to generate an optimal coalition to one task, but few is about multi-task multi-coalition generation (MMG) problem. A model is established to MMG and the basic characteristics are analyzed. A parallel algorithm is suggested to solve this problem under the assumption of limited resources of the system, first finding the best achievable task set, then generating corresponding coalitions to each task. The efficiency and feasibleness of the proposed algorithm are demonstrated by simulation studies.