Deep Reinforcement Learning Based Freshness-Aware Path Planning for UAV-Assisted Edge Computing Networks with Device Mobility

Abstract

As unmanned aerial vehicles (UAVs) can provide flexible and efficient services concerning the sparse network distribution, we study a UAV-assisted mobile edge computing (MEC) network. To satisfy the freshness requirement of IoT applications, the age of information (AoI) is incorporated as an important performance metric. Then, the path planning problem is formulated to simultaneously minimize the AoIs of mobile devices and the energy consumption of the UAV, where the movement randomness of IoT devices are taken into account. Concerning the dimension explosion, the deep reinforcement learning (DRL) framework is exploited, and a double deep Q-learning network (DDQN) algorithm is proposed to realize the intelligent and freshness-aware path planning of the UAV. Extensive simulation results validate the effectiveness of the proposed freshness-aware path planning scheme and unveil the effects of the moving speed of devices and the UAV on the achieved AoI.