Coalition-structure’s generation methods are usually employed to solve team allocation optimization problems or cooperative computing scheduling problems in the case of multitasking concurrency. In edge computing environments, affected by such factors as a large number of edge nodes, weak computing power, multiple optimization objectives and multiple constraints, the traditional methods can hardly guarantee the optimization speed and the optimal solution’s quality when solving similar problems. Based on the advantages of cooperative game algorithms and heuristic algorithms, we propose a coalition-structure’s generation method suitable for edge computing environments in this paper. Firstly, we introduce the concept of bargaining set and remove the impossible coalition-structures by judging the no-bargain coalition to narrow the strategic space. Secondly, for increasing the optimization speed and the optimal solution’s quality, we improve the inertia weight computing method and the particle state determination method of the primary discrete particle swarm, propose M-ary discrete particle swarm optimization (MDPSO). Finally, we design a series of contrast experiments and verify that this method boasts obvious advantages in optimization speed, the optimal solution’s quality, stability, and other aspects.

在多任务并发的情况下，联盟结构的生成方法通常用于解决团队分配优化问题或协作计算调度问题。在边缘计算环境中，受边缘节点数量大，计算能力弱，优化目标多，约束多等因素的影响，传统方法在解决相似问题时难以保证优化速度和最优解决方案的质量。基于协同博弈算法和启发式算法的优点，提出了一种适用于边缘计算环境的联盟结构生成方法。首先，我们介绍了议价集的概念，并通过判断无议价联盟来缩小战略空间来消除不可能的联盟结构。其次，为提高优化速度和最优解的质量，我们对一次离散粒子群的惯性权重计算方法和粒子状态确定方法进行了改进，提出了M元离散粒子群优化算法（MDPSO）。最后，我们设计了一系列对比实验，验证了该方法在优化速度，最优解的质量，稳定性等方面具有明显的优势。