Distributed Coordinate-free Boundary Nodes Identification in Wireless Sensor Networks

In a wireless sensor network (WSN), coverage holes may appear due to sensor random deployment or node failure. They may restrain the network from providing comprehensive and quality data for end users. In order to heal the coverage holes, the network first needs to localize them and the main idea is to find the boundary nodes on them. Although the problem of identifying the boundary nodes in WSNs has been intensively studied, most of them accomplished the work by using the coordinates of sensors. However, due to the comparatively expansive cost for equipping sensors with localization devices such as GPSs, coordinates are often unavailable in low-budget networks. Existing coordinate-free algorithms often perform their works in evenly distributed sensor networks with relatively high node density. Therefore, how to complete the task without assistance of coordinates still faces difficulties in general networks. In this paper, we present a distributed algorithm to distinguish the boundary nodes from inner ones in a network according to their 1-hop neighbors. In our approach, a sensor first finds out if there is a Hamiltonian circle in the neighborhood of it. Then, it analyzes characteristics of the circle to decide if the circle is located beside or around, based on techniques of neighborhood triangulation and point in triangle test without coordinates. Simulation results show the performance effectiveness of our method in identifying boundary nodes in networks with random and uneven sensor deployment, especially in the ones with small scattered coverage holes.