Online Dispatching And Scheduling Method For Tasks Based On Edge-cloud Computing Systems

Faced with the processing demands of massive data and long transmission delays in cloud computing systems,edge cloud computing systems are widely used to support various computing services.The dynamic task offloading(i.e.,task dispatching)and scheduling problem in the edge-cloud collaborative computing environment is studied in the paper.Facing the realsitic mobile edge-cloud computing scenario,considering the random arrival of independent tasks and the heterogeneity of computing resources,with the optimization objective of minimizing the total weighted response time of all tasks,a single-objective scheduling optimization problem model with limited resource constraints is constructed.For the problem model,an online offloading framework ICSOF(ICS Based Online Offloading Framework)based on the improved cuckoo search intelligent optimization algorithm is proposed in the paper.ICSOF periodically performs dynamic task offloading and scheduling calculations according to the required transmission delay and computing time.Evaluating the server computing resource status and task execution in real time.Under the ICSOF algorithm framework,the traditional cuckoo search algorithm is improved from the aspects of step scale factor and population encoding and decoding mode.The step adaptive adjustment mechanism is introduced.An improved cuckoo search intelligent optimization algorithm is proposed,which strives to search for the globally optimal task arrangement sequence,offloads the task to the best destination(edge server or cloud server),and improves the search ability of the algorithm.The offload process fully considers the delay sensitivity of the task.According to the availability of the server’s computing resources,the computing instance schedules the uncompleted tasks in a preemptive manner,and strives to achieve the efficiency of task scheduling.A method to manage the resource availability is designed.When the offloading and scheduling decisions are updated,manage and update the resource status synchronously and periodically.Based on the workload tracking of Google Cluster Trace,a simulation data set,simulation experiments are conducted.The parameters of the experimental platform are set and the parameters of the proposed algorithm are calibrated to determine the final algorithm experimental parameters.In order to verify the correctness,effectiveness and feasibility of the algorithm,the performance of the algorithm after parameter calibration is compared with the other four excellent unloading scheduling optimization algorithms in different parameters and scales.Simulation results show that the proposed algorithm has better performance and can effectively reduce the total weighted response time of all tasks.