Unmanned Aerial Vehicles (UAVs) have attracted widespread attention in the field of assisted data collection and localization due to their flexibility, mobility, and ease of deployment. In this paper, we put forth a framework for UAVs to leverage real-time localization data to assist data collection in a low-latency-sensitive, energy-efficient sensor network. This framework addresses the shortcomings of inadequate infrastructure coverage, the inherent unpredictability and ambiguity of the data collection process, and the challenges posed by sensor locations in the Internet of Things (IoT). The proposed framework markedly enhances system performance by leveraging real-time localization data to assist data collection. Specifically, we optimize the traditional algorithms in terms of model construction (e.g., the PER DQN algorithm, the Dueling DQN algorithm, etc.), data collection strategy, etc., to reduce the computational complexity and resource consumption of the UAV and assist the UAV in obtaining the optimal strategy for trajectory planning, in response to a series of problems, such as the lower processing efficiency and slower convergence speed of the traditional algorithms. Extensive simulation results show that our proposed solution improves the average data collection performance of the system by 17% and reduces the convergence time to 83.3% compared to traditional methods.

ipython==8.12.3 matplotlib==3.8.0 numpy==1.26.4 torch==2.1.0

Run the following command:

training：

python DQN.py

testing：

python watch_uav.py