UAV-Enabled Mobile Radiation Source Tracking with Deep Reinforcement Learning

Employing unmanned aerial vehicle (UAV) in localization and tracking system can bring many attractive advantages due to its high mobility, on-demand deployment and low cost. In this paper, we utilize the UAV as a mobile sensor to close up and track a mobile radiation source only based on the received signal strengths. We aim to maximize the sum of received signal strengths at the UAV receiver during a certain time interval, while taking the UAV’s maximum speed and fly region constraints into account. However, it is very challenging since the positions of radiation source are unknown and dynamically changing. To address this issue, we propose a deep reinforcement learning (DRL) based framework. We first reformulate the original problem into a Markov decision process (MDP). Then, we apply the double deep Q-network (DDQN) with dueling network structure and accordingly develop a multi-step dueling-DDQN learning algorithm for radiation source tracking. The simulation results demonstrate the effectiveness of the proposed algorithm under various parameter settings.