Optimization of Computation Offloading in Mobile-Edge Computing Networks with Deep Reinforcement Approach

Many diverse computation-intensive mobile applications such as augmented reality (AR), virtual augmented (VR) reality, and online gaming have evolved with the widespread use of mobile devices. Nevertheless, mobile devices have difficulties in completing the necessary computation in time because of their low computational power. However, wireless powered mobile-edge computing (MEC) networks offer a promising avenue for enhancing computational capabilities, particularly in resource-constrained environments. In this paper, an optimal approach leveraging Deep Reinforcement Learning (DRL) is proposed for online computation offloading decision dynamically within such networks. The proposed method dynamically optimizes task offloading decisions in response to evolving wireless channel conditions, thus maximizing computational efficiency. This paper demonstrates the efficacy of proposed approach, achieving significant improvements in computation rates compared to existing methods via extensive simulations. This work contributes to the advancement of efficient computation offloading strategies in wireless powered MEC networks, paving the way for enhanced performance in real-world applications like heterogeneous networks e.g., smart cities, health care, autonomous vehicle, etc.