Data Collection Optimization for UAV-Assisted Wireless Sensor Networks

Abstract

The last decade witnessed rapid advancement of digital technology, such as the Internet of Things (IoT), having enormous applications in various domains. Wireless Sensor Networks (WSNs) play a central role in the context of IoT for providing a massive amount of data captured by ubiquitous sensors. Thus, data collection becomes crucial to feed fresh data into IoT services while avoiding data overwritten due to limited storage capacities of IoT sensors. With high agility, mobility and flexibility, the Unmanned Aerial Vehicles (UAVs) have recently received considerable attention for data collection in WSNs.

In this thesis, we investigated novel problems for optimizing the data collection efficiency in UAV-assisted WSNs. We conducted studies to maximize data collection volume via UAV hovering duration allocation. Furthermore, we studied the maximization of the data collection utility by jointly considering the UAV hovering location recognition and the UAV hovering duration allocation. We also investigated the design for UAV data collection trajectory, with the consideration of the UAV energy consumption on both hovering and mechanical movement, where the variation of the data transmission rate is also considered.

We firstly address the UAV hovering duration allocation problem, which is a fundamental but crucial issue for data collection optimization. For WSNs with a special deployment, we develop an optimal algorithm which guarantees the maximized volume of collected data. For general cases, we devise an efficient algorithm with a provable approximation ratio, which significantly improves the data collection efficiency.

We secondly deal with the UAV hovering location identification problem. It is widely considered to be intractable to precisely determine the UAV's hovering locations from infinitely many potential ones jointly with the data collection. Most existing studies either neglect this critical issue or discretize the UAV serving area into small regions with a given size, which results in the inevitable utility loss of data collection. To tackle this problem, we propose a novel algorithm to precisely determine the UAV's potential hovering locations and devise an efficient approximation algorithm with guaranteed data collection performance.

We thirdly explore the UAV traveling trajectory designing problem, which focuses on both the hovering location identification of the UAV and the UAV traveling path designing for improving data collection efficiency in WSNs. We consider the energy consumption of the UAV on both hovering and mechanical movement together with the variation of sensors' data transmission rate, and develop a promising algorithm to find a closed UAV data collection tour for maximizing the data collection volume.

We finally evaluate the performance of all proposed algorithms in this thesis through experimental simulations. Simulation results show that the proposed algorithms significantly outperform existing algorithms. Show more