In cloud computing system,it is a hot and hard issue to find the optimal task scheduling method that makes the processing cost and the running time minimum. In order to deal with the task assignment,a task interaction graph was used to analyze the task scheduling; a modeling for task assignment was formulated and a particle swarm optimization (PSO)algorithm embedded in the variable neighborhood search (VNS) to optimize the task scheduling was proposed. The experimental results show that the method is more effective than the PSO in processing cost,transferring cost, and running time. When the task is more complex,the effect is much better. So,the algorithm can resolve the task scheduling in cloud computing and it is feasible,valid,and efficient.
This paper deals with the discrete-time connected coverage problem with the constraint that only local information can be utilized for each robot. In such distributed framework, global connectivity characterized by the second smallest eigenvalue of topology Laplacian is estimated through introducing distributed minimal-time consensus algorithm and power iteration algorithm. A self-deployment algorithm is developed to disperse the robots with the precondition that the estimated second smallest eigenvalue is positive at each time-step. Since thus connectivity constraint does not impose to preserve some certain edges, the self-deployment strategy developed in this paper reserves a sufficient degree of freedom for the motion of robots. Theoretical analysis demonstrates that each pair of neighbor robots can finally reach the largest objective distance from each other while the group keeps connected all the time, which is also shown by simulations.
