An Effective Metaheuristic for Partial Offloading and Resource Allocation in Multi-Device Mobile Edge Computing
Weirong He, Suxiang Wu, Jin Yuan Sun · 2021
Mobile edge computing (MEC) can provide mobile devices with high-quality computing services by offloading computation-intensive tasks to MEC servers, which are close to mobile devices, instead of submitting them to the cloud datacenter. The offloading strategy has a significant impact on the quality-of-service provided by the MEC system. Oriented toward a multi-device MEC system, this work takes into account the processing delay of computation tasks and the total energy consumption of the MEC system, and studies the partial offloading and resource allocation problem for minimizing the delay under the energy constraint. To solve the above-mentioned optimization problem, which is formulated as an NP-complete integer program, we propose an effective scheduling metaheuristic based on the solution exploration mechanism of artificial bee colony (ABC) algorithm. With a task sequence representing the scheduling solution, the ABC-based metaheuristic uses a position-based mapping method to convert each individual bee in ABC into a valid solution. For each converted solution, we further use a greedy search strategy to determine the order of tasks to be offloaded and the number of computing resources allocated for processing each offloaded task. The resultant task processing delay is used to evaluate the quality of the nectar source found by the bee. Our proposed method iteratively employs the ABC's population updating rule to update the positions of nectar sources, and identifies the current best solution corresponding to the best nectar source in the population. Experimental results demonstrate the advantageous of the proposed algorithm over baseline algorithms in reducing the task processing delay.