QoS-based cooperative computation offloading scheme and resource allocation in mobile edge computing

Abstract

This study is about the methods of partial computation offloading decision and computing resource allocation of edge servers which uses edge computing resource and secondary cloud computing resources cooperatively in the mobile edge computing environment.The edge computing environment, which consists of an edge network connected by terminals, has low latency but lacks computing performance than typical cloud resources, making it difficult for the edge server to handle many concurrent computation offloading requests from many terminals.Therefore, the edge server needs to determine a proper computing resource allocation, secondary cloud resource usage, and computing cost to satisfy multiple computation offloading requests simultaneously and smoothly.In this edge computing environment, terminals need to determine the number of partial computations to offload in consideration of their quality-of-service(QoS) satisfaction to maximize the energy and time gains of terminals.In the edge-terminal domain, the time and energy gain of the terminal is defined based on the law of marginal utility, and then it proposes a solution to allocate computing resources using the water-filling method for the edge server, and a solution to obtain the partial computation offloading strategy for the terminals.The strategic behaviors of terminals under the proposed computing resource allocation scheme are analyzed within a game-theoretic approach.In the edge-cloud domain, it provides task scheduling and computation resource allocation methods to reduce the cost of the edge server while satisfying the operation processing deadline guaranteed to the terminal through the branch-and-bound method and heuristic method.Through numerical analysis, we show that the terminal's partial computational offloading determination method can guarantee a minimum time gain even if the number of simultaneous requests is large.In addition, the server's resource allocation and secondary cloud resource utilization determination schemes show that the utility of terminals can be secured by more than 10\% compared to other existing techniques.