关键词: 对等网络, 任务调度, 参数拟合, 克隆选择, 数学模型

Abstract: Due to dynamic uncertainty of Peer to Peer (P2P) network and the complexity of task scheduling, it was difficult to cope with the problems of searching peers and balancing load effectively. To deal with these problems, the statistic indeterminate reasoning was applied to seek the set of idle peers which satisfying the needs of scheduling time from the large amount of the statistic data of the idle time held by the nodes. Meanwhile, the dynamic performance parameters of idle peers were fitted by using mathematical model, the set of idle peers were ordered by performance. And the tasks were matched to the idle peers to complete the task scheduling process according to the improved Immune Clonal Selection Algorithm (ICSA). The experimental results indicated that the proposed nodes searching mechanism could select the peers which were satisfying the needs of task scheduling more exactly, and this mechanism could effectively save network overhead and shorten task completion time.

Key words: peer to peer, task scheduling, parameters fitting, clonal selection, mathematical models