Energy Efficiency Aware Collaborative Multi-UAV Deployment for Intelligent Traffic Surveillance

Abstract

The existing city traffic surveillance systems are mainly based on passive monitoring by using fixed sensors, which cannot fully meet the highly dynamic monitoring requirements of intelligent traffic. To address this concern, this paper employs flexible Unmanned Aerial Vehicles (UAVs) to provide active monitoring services in a cooperative way. Firstly, we comprehensively consider the energy consumption and network connectivity constraints of the UAV system to establish a multi-UAV-based mobile monitoring model aiming at maximizing task energy efficiency. Then, we formulate the collaborative multi-UAV surveillance problem as a multi-agent Markov decision process to determine the optimal strategies for UAVs. Next, we propose a Collaborative Multi-UAV Deployment (CMUD) algorithm based on Multi-Agent Deep Deterministic Policy Gradient (MADDPG) by designing an effective reward function and improving the experience replay scheme. In addition, we introduce a policy integration scheme for the proposed CMUD algorithm to solve the problem that the UAV overly relies on the movement strategies of other UAVs in the non-stationary multi-UAV traffic monitoring environment. Simulation results show that the proposed CMUD algorithm can effectively accelerate the exploration of movement strategies with the guarantee of stable connectivity and efficiently improve task energy efficiency.