Abstract: Localization is one of the pivot technologies in wireless sensor networks. The traditional node localization schemes consider that the locations of anchors are reliable, which makes these schemes are invalid in some scenarios with unreliable anchors such as drifted anchors, fake anchors and malicious anchors. Aiming at solving this problem mentioned above, a distributed and lightweight node location verification framework (NLVF) is proposed. NLVF offers location verification service as an underlying technic for the traditional localization algorithms, including range-based localization algorithm and the range-free localization algorithm. NLVF can filter out these unreliable anchors by which the application area of traditional localization algorithms is enlarged. UNDA (unreliable node detection algorithm) is the key algorithm of NLVF. It constructs location reputation model based on mutual distance observation between neighbors in WSN. UNDA algorithm improves the localization reliability by filtering out these anchors with inferior location reputations. Extensive experiments are conducted to evaluate the performance of UNDA. Results show that NLVF is adapted to both of range-based and range-free localization schemes. It works better in the presence of three kinds of unreliable anchors. So, it yields general applicability. In addition, UNDA relatively has high accuracy, and the average success rate of detection is more than 95%, so NLVF yields significant practicability.

Key words: wireless sensor network (WSN), node localization, reliable localization, node location verification, distributed reputation model