Research on Path Planning Algorithm for UAV-Assisted Internet of Things Data Collection

Abstract

In recent years, unmanned aerial vehicles (UAVs) have gradually entered people's vision and made significant progress in various fields. Compared with traditional ground communication, UAV has the advantages of strong mobility, on-demand deployment, flexible configuration, etc. It is often used to assist wireless sensor networks in data collection and improve network performance. This paper focuses on the path planning of data collection in wireless sensor networks assisted by UAVs. In the first part of this article, an improved differential evolution algorithm based on genetic perturbations (Gaussian perturbations) is proposed to address the problem of slow convergence speed and susceptibility to local optima in path planning using the standard differential evolution algorithm. This algorithm aims to improve the optimization performance of the algorithm. Next, the analysis of a wireless sensor network's energy consumption leads to the derivation of the energy consumption expression for ground air data transfer between the sensor network and UAV. Under the constraint of data transmission rate, the energy consumption of the sensor network is minimized by jointly planning the drone flight trajectory and sensor node power. The optimization problem is expressed as a nonconvex form with constraints. The original problem is solved to obtain an estimated optimal solution using the block Coordinate descent and the modified differential evolution approach introduced in this study. The second part of this article focuses on the path planning goal of unmanned aerial vehicles (UAVs) that cannot achieve real-time obstacle avoidance while also considering data collection efficiency when there are unknown obstacles on the ground. Based on this situation, this article first improves the cost function and search strategy of the standard $\mathbf{A}^{\ast}$ algorithm to improve the algorithm's search efficiency and path security. Secondly, throughput guidance is introduced into the evaluation function of the standard DWA algorithm to improve the efficiency of drone data collection. Finally, the improved $\mathbf{A}^{\ast}$ and improved DWA algorithms are fused to plan a drone trajectory that can avoid obstacles in real time while also taking into account the efficiency of data collection for ground sensor networks.