Abstract
This paper investigates the scenario of the Internet of things (IoT) data collection via multiple unmanned aerial vehicles (UAVs), where a novel collaborative multi-agent trajectory planning and data collection (CMA-TD) algorithm is introduced for online obtaining the trajectories of the multiple UAVs without any prior knowledge of the sensor locations. We first provide two integer linear programs (ILPs) for the considered system by taking the coverage and the total power usage as the optimization targets. As a complement to the ILPs and to avoid intractable computation, the proposed CMA-TD algorithm can effectively solve the formulated problem via a deep reinforcement learning (DRL) process on a double deep Q-learning network (DDQN). Extensive simulations are conducted to verify the performance of the proposed CMA-TD algorithm and compare it with a couple of state-of-the-art counterparts in terms of the amount of served IoT nodes, energy consumption, and utilization rates.
Original language | English |
---|---|
Pages (from-to) | 722-733 |
Number of pages | 12 |
Journal | Journal of Communications and Networks |
Volume | 25 |
Issue number | 6 |
DOIs | |
State | Published - Dec 2023 |
Keywords
- Collaborative UAVs
- data collection
- deep reinforcement learning
- energy efficiency
- IoT coverage
- trajectory planning