Reinforcement-Learning-Based Scheduling for Age of Information Optimization in MEC Multi-User Network

In this paper, we consider an optimization problem in a mobile edge computing (MEC) multi-user network for collecting fresh information under bandwidth and energy constraints. We use the popular metric Age of Information (AoI) to measure the data freshness. To solve the age optimization problem, we first relax the bandwidth constraints, and then decouple the problem, where each sub-problem is formulated into a constrained Markov decision process (CMDP) problem. Due to the fast mobility of the users, the distribution of channel state information and size of raw data is difficult to obtain for the base station. To overcome the unknown channel statistics challenge, we introduce a Reinforcement-Learning-based Q-learning algorithm to learn the channel information, which helps to guide the optimal scheduling decision. Simulation results demonstrate that the proposed algorithm can achieve the better AoI performance compared with other traditional algorithm while satisfying bandwidth and energy constraints.