Localization Based on Adaptive Regulated Neighborhood Distance for Wireless Sensor Networks With a General Radio Propagation Model

Many range-free localization algorithms estimate the distance in between two nodes based on the hop-count information, which, however, is a coarse proximity measure. Recently, the regulated neighborhood distance (RND) has been proposed as a new proximity measure, and can greatly improve the localization accuracy, compared with those hop-count-based algorithms. However, the RND algorithm and most previous hop-based approaches are based on the disk communication model, and cannot be directly applied into the practical world with some general radio propagation model. In this paper, we revisit the RND-based localization with a general propagation model, where the received transmission power is modeled as a random variable and its expectation is a nonincreasing function of the distance between a transmitter and receiver pair. In particular, we define the neighborhood based on the packet reception ratio (PRR), and propose an optimal PRR threshold selection algorithm to adaptively adjust the RND-based distance estimation. We verify the effectiveness of the adaptive RND-based localization for the log-normal shadowing model and a polynomial fitting model from our field experiments. Simulation results show that the adaptive selection of the best PRR threshold helps to improve localization accuracy. Furthermore, the proposed adaptive RND-based localization can achieve higher localization accuracy, compared with the classical DV-hop localization with the same network configurations.