Efficient Navigation for Unmanned Agents in Sparse Wireless Sensor Networks

Abstract

Many works have reported on various sensor network position estimation methods based on the relative distance measurement that can be used when the global navigation satellite system is environmentally denied or degraded.1–3) Among others, trilateration algorithms are widely adopted because of their simple principle.4–6) However, the algorithms possibly fail if the sensors have a low range of communication or the environment includes obstacles.7) Typically, such distance-based localization algorithms are used to construct a globally rigid network.8,9) In other words, albeit each sensor, called node herein, has a limited transmission range, unmanned agents, like unmanned aerial vehicles (UAVs), should be inside the coverage space to receive the sensors’ information.10) Therefore, the typical algorithm requires a network that can adequately cover a certain area and must be capable of communicating with at least three sensors at any point in the area. However, such a network is not always guaranteed. This study proposes a strategy to maximize UAV’s navigation in a sparse wireless sensor network (SWSN) in the manner of the shortest distance travel. The overlapping (or localizable) area, which is calculated using the positions of three disks constructed by the sensor’s transmission range, is used to characterize the possibility of localizing UAVs through trilateration. To ensure that a UAV travels from a starting point to a destination point via the localizable area, it must pass the points that are defined by a sensor set, called vertices. The keys are to find such vertices to define a graph that is flexible to various network complexities that are determined by the combination of sensors and reduce the number of search nodes or the total distance. To determine the shortest path, the Dijkstra algorithm,11,12) one of the most widely used algorithms, is applied with proper modifications. The feasibility of the proposed method is verified through twodimensional (2D) and 3D examples.