Abstract: In order to improve node localization accuracy, a node localization algorithm based on adaptive random-walk module was presented for wireless sensor networks. First, a novel metric for relative distance among node sensors was modeled by applying the idea of random walk to the connectivity of system topology. Then an adaptive approach was designed to increase the validity of the metric. At last, node positions were finally obtained by embedding the metric in the classical localization algorithm distance vector-hop (DV-Hop). Simulation and outdoor environment's results show that the design achieves better robustness and positioning performance, and localization errors of the proposed method reduce by about 20%~30% compared with that of DV-Hop algorithm.

Key words: node localization, wireless sensor network, random walk, a metric for relative distance, adaptive algorithm