Research on Mobility-Aware Computation Offloading and Resource Allocation Strategy

BAN Yuqi,DUAN Liguo,WEN Haoyu,LI Aiping,ZHAO Jumin

DOI: https://doi.org/10.19678/j.issn.1000-3428.0066522

2023-01-01

Abstract:In mobile edge computing， user equipment offloads computationally intensive tasks to edge servers for execution to reduce execution delay and energy consumption. This requires 5G technology-based applications to support the high-speed movement of devices during computing. However， much of the current research on computational offload solutions is focused on static scenarios. To improve the quality of user experience， this study investigated a computational offloading scheme that took into account device movement trajectories in mobile edge computing and hence better suited multi-device and multi-MEC server scenario. Since this scheme considered multiple factors， such as device mobility， computing and communication resources， channel states and mission requirements， it could be described as a mixed-integer nonlinear programming problem. In order to reduce the difficulty of solving， this study decomposed the problem into the problem of offloading server selection and the problem of computing resource allocation and sub-channel selection under the fixed server selection scheme. The convex optimization technique and the improved Kuhn-Munkres algorithm are used to solve the subproblem. Meanwhile， this study also designed a heuristic offload server selection algorithm based on the solution of the subproblem， and obtained a suboptimal offload solution with polynomial time complexity. Then， simulations were conducted by the EdgeCloudSim tool， of which the results proved the effectiveness of the algorithm proposed in the present study， in comparison with five other commonly used offloading algorithms. The experimental results show that the average system utility gap between the algorithm and the exhaustive method can be controlled within 2.3% when it meets the real-time requirements of the task. The performance of the proposed offloading scheme under different task types and the applicability scenarios of the algorithm are analyzed in the subsequent experiments.