A task segmentation and computing offload algorithm for mobile edge computing

In the existing mobile edge computing (MEC) environment, servers have limited power and cannot process data in real time. To solve this problem, paper proposes a task segmentation offloading algorithm for (MEC). Firstly, tasks of terminals are segmented under network framework cloud-side-device collaboration, optimization model is designed to minimize total delay devices. Secondly, resource allocation parameters action space discretized, based on Deep Q Network (DQN) proposed. The extracts feature from high-dimensional uses deep neural estimate Q-value actions maximize summation system delays. Finally, simulation experiments show that when cloud capacity 7 × 1011 central processing unit (CPU) cycle/s, proposed only 1.116 s, while delays comparison algorithms 1.251 s 1.193 respectively. Therefore, compared with algorithm, can better reduce delay.