CONSEL: Connectivity-based segmentation in large-scale 2D/3D sensor networks

—A cardinal prerequisite for the system design of a sensor network, is to understand the geometric environment where sensor nodes are deployed. The global topology of a largescale sensor network is often complex and irregular, possibly containing obstacles/holes. A convex network partition, so-called segmentation, is to divide a network into convex regions, such that traditional algorithms designed for a simple geometric region can be applied. Existing segmentation algorithms highly depend on concave node detection on the boundary or sink extraction on the medial axis, thus leading to quite sensitive performance to the boundary noise. More severely, since they exploit the network’s 2D geometric properties, either explicitly or implicitly, so far there has been no general 3D segmentation solution. In this paper, we bring a new view to segmentation from a Morse function perspective, bridging the convex regions and the Reeb graph of a network. Accordingly, we propose a novel distribut...

• Type: Journal
• Year: 2012
• Where: INFOCOM
• Authors: Hongbo Jiang, Tianlong Yu, Chen Tian, Guang Tan, Chonggang Wang
• Keywords: 3d Segmentation, Communications, Convex Decomposition, INFOCOM 2012, Small Concavity