Maximizing quality of UAV-enabled surveillance over multiple restricted regions

In this work, we study the problem of maximizing the unmanned aerial vehicle (UAV) surveillance quality for multiple terrestrial targets and each target is located within a restricted region where the UAV cannot enter. We propose an angle-based surveillance quality metric to quantify the surveillance effectiveness of the UAV under the no-fly zone constraint. To maximize surveillance quality within the mission completion time constraint, our two-phase approach consists of: (1) determining the visiting order of targets through a minimum spanning tree-based traveling salesman problem (TSP) solution; (2) maximizing surveillance quality by developing a Maximum Quality In Time (MQIT) algorithm that iteratively optimizes the UAV trajectory. The MQIT algorithm incorporates a global-first-local-refinement heuristic to maximize surveillance quality, along with a designed Minimum Time with Guaranteed Quality (MTGQ) module. The MTGQ module employs dynamic programming (DP) to optimize the time allocation while satisfying predefined quality thresholds. More importantly, the MTGQ algorithm can operate as an independent module to solve time-critical mission problems with specified surveillance quality requirements. Extensive simulation results demonstrate that our method outperforms several baseline approaches across various scenarios, achieving a 6% improvement in surveillance quality.