Data-similarity-based IoT node selection for UAV trajectory optimization

Abstract

Data collected from densely deployed IoT sensor nodes via UAVs may incur remarkable redundancy, unnecessarily wasting the UAV power. This paper introduces a framework of UAV trajectory planning for optimizing energy efficiency via data-similarity-based node selection while maintaining sufficient information integrity out of the refined data. The proposed framework consists of three phases, namely data similarity determination, redundant nodes removal, and UAV trajectory planning. In particular, we propose a sliding-window dynamic time warping (SDTW) algorithm to quantify the data similarity between nodes. Then a hybrid genetic ant colony algorithm (HGACA) is introduced for the redundant node removal, where data similarity and UAV energy consumption are jointly considered. Finally, we formulate the trajectory planning problem as a three-stage integer linear programming (ILP) model, which clusters the nodes with minimal overlap and finds the shortest UAV cruising path that traverses each cluster head once and only once. The simulation result demonstrates that the proposed framework outperforms all the considered counterparts under various threshold values of data similarity in terms of execution time and power consumption while maintaining information integrity.