An Adaptive Multitasking Evolutionary Computation Offloading Algorithm for Mobile Edge Computing

In the mobile edge computing scenario, intelligent terminal devices can reduce the waiting delay by offloading the computing task to a server. The offloading scheme’s optimization has been proven an NP-hard problem. The heuristic algorithms, including evolutionary algorithms, are widely used to search for the optimal scheme. User experience is mainly limited by energy consumption and time delay. Most existing research results combine it linearly into a single objective or focus on the optimal solution in a specific area. Based on this, this paper proposes an adaptive multitasking evolutionary optimization algorithm, which considers multiple independent areas to be optimized. It abstracts the task offloading system model in each area as a multi-objective programming problem, aiming at minimizing the average energy consumption and delay of intelligent devices. By learning the user distribution and the similarity of tasks to be processed in different areas dynamically to adjust the degree of population communication, the convergence has been sped up. The performance of the proposed algorithm is verified by a set of instances.