A task segmentation and computing offload algorithm for mobile edge computing
Xingyan Shi · The Journal of Engineering · 2022
Abstract In the existing mobile edge computing (MEC) environment, edge servers have limited computing power and cannot process data in real time. To solve this problem, this paper proposes a task segmentation and computing offloading algorithm for mobile edge computing (MEC). Firstly, the computing tasks of mobile terminals are segmented under the network framework of cloud‐side‐device collaboration, and a task offloading optimization model is designed to minimize the total delay of mobile devices. Secondly, the computing resource allocation parameters and action space parameters are discretized, and a computing offloading algorithm based on Deep Q Network (DQN) is proposed. The algorithm extracts feature from high‐dimensional data and uses deep neural network to estimate Q‐value of actions to maximize the summation of system delays. Finally, simulation experiments show that when the cloud computing capacity is 7 × 10 11 central processing unit (CPU) cycle/s, the total delay of proposed algorithm is only 1.116 s, while the total delays of comparison algorithms are 1.251 s and 1.193 s, respectively. Therefore, compared with the comparison algorithm, the proposed algorithm can better reduce the total system delay.