Data Offloading and Energy Management in MEC Networks With Distributed Power Source: A DRL Approach

The rapid proliferation of Internet of Things (IoT) devices has led to an exponential growth in data generation, exacerbating the inherent limitations of conventional IoT systems, particularly in computational capacity and energy sustainability. To address these challenges, this paper proposes an integrated framework combining mobile edge computing (MEC), distributed energy resources, and smart grid technology. Within this framework, distributed power sources not only supply energy to IoT devices but also engage in energy trading with the power grid. We formulate a joint optimization problem to maximize both computational efficiency and weighted energy trading benefits through optimal task offloading decisions and energy allocation strategies. The resulting mixed-integer nonlinear programming (MINLP) problem is decomposed into a master problem and a subproblem for efficient solution. A deep reinforcement learning (DRL) approach is developed to determine optimal offloading decisions, while convex optimization are employed to solve the energy allocation subproblem. Extensive experimental evaluations demonstrate that our proposed method achieves superior convergence behavior and system performance compared to existing benchmark algorithms.