A Sampling-based Boundary Detection Method with Consideration of Boundary Shape in Dense Mobile Wireless Sensor Networks

In this paper, we propose a method which efficiently detects the geographical boundaries of sensor readings in mobile wireless sensor networks, in situations where sensor data must be gathered within a short period. Typically, it is enough for detecting the boundaries to gather sensor data only from nodes near the boundaries. However, when the number of such nodes is large, it becomes impossible to gather sensor data from all these nodes within a short period, due to the frequent occurrence of packet collisions. To solve this problem, our proposed method reduces the number of sensor data to be gathered, by taking samples of nodes. When taking samples, our proposed method considers the target boundary shape, in order to suppress the deterioration in the accuracy of the boundaries estimated from the gathered sensor data.