A Thread Pool Scheduling Optimization Method of Real-time System

To deal with concurrent requests, thread pool technology is wildly used in multi-task real-time systems, which chief demand is timeout avoidance. In order to reduce request timeout ratio, least square support vector machine(LS-SVM) algorithm which parameters optimization based on hybrid quantum-behaved particle swarm optimization(HQPSO) was applied to estimate the execution time of requests. Also, a scheduling priority algorithm of thread pool was designed based on the estimation results. In performance test, a node of wireless sensor network is implemented to test the request timeout ratios of HQPSO LS-SVM based thread pool and other ones in different states. Test result indicates that HQPSO LS-SVM based thread pool has remarkable superiority of timeout avoidance while the thread amount of thread pool is set properly.