Graph drawing based WLAN indoor mapping and localization using signal correlation via edge detection

In indoor Wireless Local Area Network (WLAN) localization, the Received Signal Strength (RSS) fingerprinting involved in fingerprint based localization is always time-consuming and labor intensive. To solve this problem, we propose a novel indoor mapping and localization approach by using the spectral clustered time-stamped WLAN RSS measurements to conduct indoor mapping, as well as locate the target. Specifically, we rely on the off-the-shelf smartphones to sporadically collect a batch of WLAN RSS sequences in target environment, and then perform spectral clustering on the RSS sequences to construct cluster graphs. Furthermore, by using the orthogonal algorithm in graph drawing, we represent each cluster graph in a more readable manner. After that, the edge detection approach in image is adopted to form the unique logic graph. Finally, we conduct indoor mapping from the logic graph to ground-truth graph. The experimental results prove that our approach can not only effectively characterize the environment, but also provide satisfactory localization accuracy.