A multi-user mobile edge computing task offloading and trajectory management based on proximal policy optimization

Mobile Edge Computing (MEC) has gained significant attention in recent years due to its capability of providing computation and storage resources closer to end-users. However, existing research has focused on addressing MEC resource optimization problems with explicit location information and has not considered the challenges posed by users who refuse to grant system access to Global Positioning System (GPS) information. To address this issue, this paper models the location-protected MEC problem as an Integer Non-Linear Problem (INLP) and proposes a novel mechanism to cater to various types of users while optimizing both computation time and service latency. This problem is proposed to be solved by decoupling it into a path recommendation problem and a task offloading problem. For different user types, a Proximal Policy Optimization (PPO) algorithm is proposed to recommend user paths. By following these paths, multi-user interference is minimized. Then, the Deep Q-Network (DQN) algorithm is employed to determine MEC offloading strategies to solve the task offloading problem. Experimental results demonstrate that the algorithm outperforms related approaches across multiple dimensions, including computation time, service latency, and stability. Additionally, the proposed mechanism effectively addresses the offloading challenge for privacy-conscious users in edge computing scenarios. The proposed mechanism has several potential applications, such as resource optimization in mobile cloud computing, the Internet of Things (IoT), and smart cities. It also addresses optimization problems for various types of users in mobile edge computing technology.