Research on Optimization Method of Resource Scheduling Strategy for Unbalanced Load Edge Computing
Currently,common cloud edge collaborative computing architectures are prone to problems such as dif-ficulty in meeting task latency requirements due to uneven load distribution,frequent execution of cloud platform re-source scheduling actions,and low utilization of network and computing resources.Aiming at the characteristics and demand scenarios of unbalanced feature computing load,a deep reinforcement learning cloud edge collaborative com-puting offloading method MA-LSTM-DQN is proposed,which applies trend moving average algorithm and integrates LSTM network.The simulation experiment of the simulation environment and the test results of the actual operator's edge computing load application scenario show that the proposed unloading decision method in the unbalance compu-ting load application scenario decrease the processing delay by 17%and scheduling frequency by about 40%,and im-proves the overall performance index to a certain extent.
Mobile edge computingResource schedulingReinforcement learningMDPComputing offloading
万方数据
维普
