Toward Collinearity-Aware and Conflict-Friendly Localization for Wireless Sensor Networks
Kai Bu,Qingjun Xiao,Zhixin Sun,Bin Xiao
DOI: https://doi.org/10.1016/j.comcom.2012.04.023
IF: 5.047
2012-01-01
Computer Communications
Abstract:Localization aims at determining node positions and is essential for many applications in wireless sensor networks (WSNs). Most existing localization protocols adopt graph rigidity theory as the theoretical basis. The rigidity theory assumes that every three nodes are noncollinear in a two-dimensional graph; this assumption, however, may not always hold in WSNs. A lack of node collinearity verification places a limitation on localization accuracy. Furthermore, existing localization protocols explore only distance constraints for localization, giving rise to another limitation on localization percentage. Against these limitations, this paper presents two approaches toward collinearity-aware and conflict-friendly rigidity-based localization for WSNs. The proposed approaches are expected to increase both localization accuracy and percentage of traditional rigidity-based localization protocols. First, to achieve collinearity-awareness, we investigate node collinearity and propose a detection method to mitigate localization errors induced by probably collinear nodes. Second, to achieve conflict-friendliness, besides distance constraints, we explore distance conflicts to eliminate position ambiguities. Distance conflicts relax the sufficient condition of 3-connectivity for localizability to 2-connectivity; this relaxation can significantly improve localization percentage. For example, trilateration using distance conflicts yields a higher efficacy in both convex and non-convex WSNs and requires only a 25% lower average connectivity degree to locate 95% of sensors. The proposed approaches can be conveniently incorporated into existing localization protocols with small overhead. We validate their effectiveness of enhancing localization accuracy and percentage through both real and simulation experiments.