Two-Scale Stochastic Optimization of Mobile Edge Computing Systems Powered by Smart Grid

In this paper, we propose two-scale online resource allocation and energy management (TSRE) for smart grid-powered mobile edge computing (SG-MEC) systems, adapting dynamic system parameters. We aim to minimize the time-average cost of the ahead-of-time energy planning and real-time energy trading of the base stations (BSs), and the energy consumption of the mobile users. By generalizing the Lyapunov optimization and the stochastic subgradient method, the energy planning (sub)problem is solved using only historical data, while the real-time decisions are decoupled over time and made in a distributed manner. The feasibility and asymptotic optimality of the proposed TSRE are presented by theoretical analysis.