Workload Allocation Mechanism On Power Iot Edge Computing-based

The electric power Internet of Things perceives status information and makes control decision by using the Internet of Things technology,which improves the intelligence level of the power grid and is a crucial part of promoting the future development of the power industry.The power IoT edge computing-based can perform task processing at the network edge near data sources,reducing the delay and energy consumption of ubiquitous business terminals.However,the computing capacity of a single edge node is limited,and terminal tasks need to be reasonably scheduled between multiple edge nodes to meet business demands.Therefore,it has a greate significance for improving the performance of the power IoT to study the workload allocation mechanism in edge computing.At present,the workload allocation mechanism is limited to the optimization of single factor that affects delay or energy consumption,and it cannot satisfy the differentiated demands of ubiquitous business.This thesis considers the resource allocation of edge nodes and task allocation of terminals in the workload allocation process,and conducts research on businesses with different demands,including workload allocation with minimum service delay and workload allocation with optimization balancing delay and energy consumption.Aiming at the problem of minimizing the service delay of operation and maintenance terminal in the intelligent operation and maintenance scenario,this thesis proposes a BRT(Balanced initialization,Resource allocation,and Task allocation)algorithm,which initially uses a balancing strategy to balance the load among edge nodes,and then optimizes the resource allocation within single edge node using pheromone improving particle swarm optimization algorithm.Finally,the semi-positive definite relaxation algorithm is used to allocate tasks between edge nodes.The simulation results show that compared to optimizing only one factor of the workload allocation process,the proposed solution can significantly reduce the terminal service delay.Aiming at the problem of minimizing the delay of energy-constrained terminals in the UAV inspection scenario,this thesis proposes a Multi-Objective Evolutionary Algorithm based on Decomposition-Improved(MOEA/D-IM)algorithm,which optimizes resource proportion in the offspring using the steepest descent algorithm when generating children,and improves the overall performance of the population using the variable neighborhood size and maximum incremental fitness strategy when the neighborhood is updated.The task allocation with a smaller aggregation function value is finally obtained.The simulation results show that this algorithm can better balance the relationship between two optimization objectives,and the delay under different energy consumption constrains is shorter compared with other multi-objectives algorithms.