Abstract: Aiming at the characteristics of randomness, dynamics, diversity and uncertainty of arrival time in the cloud fusion mode, two triggered tasks of the new tasks randomly reaching the cloud platform and the non-executable sub-tasks are considered. To improve the sub-tasks correlation degree and the workload balance of each group, the coupling degree of the sub-tasks is decreased, and an event-triggered mechanism (ETM) based optimization method for dynamic task decomposition mode is proposed. On the basis of the unified tasks description, the optimal flow of dynamic task decomposition is designed to determine the sub-tasks information flow. A multi-objective optimal model of the dynamical task decomposition based on groups is established. An improved adaptive genetic algorithm is proposed to solve the problem. The simulation results show that the proposed method can achieve the balance of cloud-fusion task-resource allocation, improve the efficiency and balance of task assignment; the presented optimization algorithm also obtains better precision and convergence performance than traditional methods.

Key words: cloud fusion, event-triggered, dynamic task decomposition, task grouping, adaptive genetic algorithm