A fast localization algorithm for large-scale wireless sensor networks

Most of the existing localization algorithms for large scale wireless sensor networks (WSN) have high complexities in time. To solve this problem, a fast visualization mapping from pairwise proximity matrix between nodes to corresponding coordinates was realized in WSN. Specifically, fast mapping algorithm procedure was introduced at first, and the reference nodes served as the pivot points. And choosing the longest diagonal was to be used for pivot lines. Thus, the transformation from relative coordinates to absolute ones was avoided. In order to improve the localization accuracy, the output of FastMap algorithm was sent to initialize MDS (multi-dimensional scaling). Area location of 600 m×600 m was set in MATLAB. Using the wireless signal attenuation model was to produce the virtual measuring points, and carried out the simulation experiment of the wireless sensor network which respectively contained 3600, 1600, 900, 576 and 400 nodes. The results show that the proposed algorithm not only keeps high positional accuracy, but also reduces the time complexity. This algorithm was used in smart supermarket guiding system. The average location errors of 21 shopping carts were 0.1585 m.