A Metaheuristic Algorithm for Mobility-Aware Task Offloading for Edge Computing Using Device-to-Device Cooperation

Task offloading is the fundamental mechanism in edge computing and offloading strategy is of great significance to the quality-of-service of an edge computing system. This paper proposes an effective metaheuristic algorithm for solving the mobility-aware task offloading problem in a edge computing environment that enables device-to-device cooperation. We first formulate the offloading problem as an integer programming-based optimization model that takes into account the moving trajectories as well as the deadline constraints upon tasks to minimize the energy consumption of the mobile device. By using a task sequence to represent an offloading solution, we propose an iterated local search algorithm (MAILS) to explore the most appropriate solution to the formulation problem. MAILS uses an efficient greedy strategy to allocate the tasks in the sequence such that the total energy for task offloading can be evaluated. To enhance the solution exploration capability, we incorporate an insertion-based local search strategy to iteratively identify more energy-efficient solutions. In case of being trapped into local optima, the iterative procedure employs a swap-based perturbation operator to perturb the task sequences to avoid local optimality. Extensive simulations on real-world human moving trajectories demonstrate that MAILS is capable of obtaining feasible offloading solutions under deadline constraints and outperforms baseline algorithms in terms of energy consumption.