论文部分内容阅读
针对传统单处理器处理复杂多样雷达任务效果欠佳问题,提出在分布式异构系统中结合改进指派算法和匈牙利算法的优点并行调度雷达任务的算法.在原始指派问题每次给每个处理器指派一项任务基础上,改进算法引入处理器雷达任务堆积量实时检测技术,设定处理器任务堆积量阈值,每次对空闲的和任务堆积量小于阈值的节点分配任务,其它处理节点不分配.对不同任务堆积量阈值和不同处理性能差别的集群利用改进算法仿真并与原始算法仿真结果对比,改进算法明显缩短了任务处理时间,改进了系统环境负载平衡,且对处理性能差别较大的集群雷达任务并行调度改进效果更好.
In order to deal with the problem of traditional single-processor processing complex and diverse radar mission, the algorithm of parallel scheduling radar mission is proposed in the distributed heterogeneous system by combining the advantages of improved assignment algorithm and Hungarian algorithm.At the same time, Assigning a task, the improved algorithm introduces the real-time detection technology of processor radar stack volume, sets the threshold of processor stack volume, allocates tasks to idle nodes and nodes whose task stack volume is less than the threshold, the other processing nodes are not allocated Compared with the original algorithm, the improved algorithm can shorten the task processing time significantly, improve the system environment load balance and improve the performance of the cluster with different processing thresholds and different processing performance. Cluster radar task parallel scheduling to improve better.