Computation offloading and resource allocation for Mobile edge computing in smart grid

The integration of mobile edge computing (MEC) and the smart grid enhances data processing efficiency in Internet of Things (IoT) networks and enables bidirectional energy flow between MEC systems and the smart grid. MEC servers can harvest energy from renewable sources via smart meters and can also sell surplus energy back to the power grid through the same interface to gain economic benefits. We design a utility function that accounts for the volume of processed data, energy consumption, and profits from energy trading. To maximize this utility, we jointly optimize offloading decisions, user and server CPU frequencies, and data transmission rates. The resulting problem is formulated as a mixed-integer nonlinear programming problem and is decomposed into two subproblems. We first propose a deep reinforcement learning (DRL)-based algorithm to determine near-optimal offloading decisions. Next, we employ a Lagrangian-based method to optimize CPU frequencies and data transmission rates. Extensive simulations are conducted to evaluate the performance of the proposed algorithm. The results show that our approach outperforms existing baseline methods and achieves nearly 95% of the maximum utility function value.