AGE OF INFORMATION-AWARE PROXIMAL POLICY OPTIMIZATION IN MEC-ENABLED HEALTHCARE INTERNET OF THINGS

Abstract

The Healthcare Internet of Things (H-IoT) enables continuous remote patient monitoring through wearable sensors and smart medical devices. However, the limited computational capability of H-IoT devices and the dynamic nature of wireless edge networks make efficient computation offloading a challenging task. Moreover, conventional offloading schemes primarily optimize latency and energy consumption while overlooking the freshness of medical information, which is critical for timely healthcare services. To address this issue, this paper proposes an Age of Information-aware Proximal Policy Optimization (AA-PPO) framework for computation offloading in Mobile Edge Computing (MEC)-enabled H-IoT networks. The proposed framework jointly minimizes the Average Age of Information (AoI), end-to-end latency, and energy consumption by learning adaptive offloading policies according to dynamic network conditions. A comprehensive system model is developed, incorporating healthcare task generation, wireless communication, computation, queueing, AoI, latency, and energy consumption. The computation offloading problem is formulated as a Markov Decision Process (MDP), and an actor–critic-based PPO algorithm is employed to obtain an efficient offloading policy. Simulation results demonstrate that the proposed AA-PPO consistently outperforms Local Execution, Random Offloading, Greedy Edge Offloading, DQN, and Double DQN. Compared with Double DQN, the proposed approach reduces the Average AoI by up to 20.5%, Peak AoI by 23.4%, end-to-end latency by 12.8%, and energy consumption by 13.3%. These results demonstrate the effectiveness of the proposed framework in maintaining fresh healthcare information while improving the overall efficiency of MEC-enabled H-IoT systems.