QoS-Aware Joint Offloading and Power Control Using Deep Reinforcement Learning in MEC

The mobile edge computing (MEC) relieves resource-constrained mobile devices from computation intensive tasks. However, it is difficult to design a joint offloading and power control method that minimizes the delay and the power consumption (including the local execution power and the transmission power). In this paper, we propose a two-step method solve the above problem. In the first step, we propose a QoS driven offloading strategy to minimize the queueing delay. In the second step, we apply a deep deterministic policy gradient (DDPG) method for power control. Simulation results show that our proposed framework achieves lower overall delay and energy consumption than existing methods.