An Efficient Clustering-Based Method for Data Gathering and Compressing in Sensor Networks

Abstract

The purpose of this paper is to implement parallel test in the single processor auto test system and to improve the test efficiency with a lower test cost. The main factor that impacts the test efficiency of test system is the performance of the parallel task scheduling algorithm. This paper puts forward a heuristic parallel task scheduling algorithm: scheduling-Q which can meet the characteristics of the auto test system. Every test tasks uses some resources to put test the units under test. So, we can use the multi-threading technique to implement single processor parallel test. In test system some test tasks can be executed with different resource allocations. The task scheduling algorithm: scheduling-Q adapts well to this characteristic. It schedules the test tasks according to the task's earliest starting time and the test generalized resource loading. The generalized resource loading is embodied as task resources set loading based on resources allocation mode and task resources set loading based on task's starting time. The test resources with bigger loading have more opportunities to obtain task and are always in a busy state. Thus resources loadings can be balanced to a degree. So the parallel performance of test system can be improved with the algorithm. In addition, the algorithm adopts the strategy of heuristic local optimum search. The time complexity of the algorithm is decreased obviously.