Research On Task Scheduling Algorithm Based On Multi-processor Heterogeneous System In Internet Of Things

In the Internet of Things era,computing efficiency and energy consumption have become two critical issues as the Internet of everything continues to advance.Edge computing can greatly reduce the transmission distance between data source and traditional processing center(cloud server),thus reducing the data transmission time.In addition,multi-processor heterogeneous systems are widely used to process parallel workflow applications to improve system performance.However,while high performance brings high processing speed,it also inevitably brings high energy consumption.How to minimize energy consumption without affecting system performance has become a big challenge in the academic world.Based on this research background,this paper mainly explores how to minimize the energy consumption of parallel applications with priority constraints when they are processed in heterogeneous distributed computing systems with limited scheduling length.For this problem,most existing scheduling algorithms only focus on the optimal processor for each task.And in the task priority sorting stage,the optimal task priority sequence may be excluded.Based on this,a Predictive Energy Consumption Scheduling(PECS)algorithm is proposed to match the processor and frequency of each task.The main steps of the algorithm are divided into three parts:first,the concept of variable time space is introduced,and the deadline of the entire application is decomposed into each task;Secondly,the prediction energy consumption matrix is defined as the basis of allocation to predict the impact of current task allocation on subsequent task energy consumption.Thirdly,the processor with the lowest predicted energy consumption is matched to the corresponding task to form a final scheduling strategy to minimize power consumption.Next,this paper reexamines PECS algorithm and finds that it still has some defects in task priority establishment and variable time space allocation.To solve the above problems,this paper proposes two methods of priority fuzzy and backward search.After PECS has formed the allocation strategy,local adjustment can be made.A new Improved Predictive Energy Consumption Scheduling(IPECS)algorithm is developed.In this paper,Python is used to simulate the operating environment of heterogeneous distributed system,and two representative task flow models are used to simulate parallel applications in this environment,and simulation results are used to evaluate the performance of IPECS algorithm.Experimental results show that compared with the current mainstream scheduling algorithm,the results of IPECS algorithm will produce lower energy consumption under the premise of not exceeding the cut-off time specified by the program.