Acoustic fingerprint based smart mobiles indoor localization under dense NLOS environment

Aiming to tackle high accuracy indoor positioning for smart mobiles under dense NLOS (Non-line-of-sight) environment with low anchor deployment density, acoustic fingerprint method is studied in this paper. Considering the properties of the cheap commercial off-the-shelf acoustic components, a TOA based acoustic fingerprint indoor positioning architecture is proposed, and the positioning methods based on decision tree with bagging, BP neural network and weighted K nearest neighbor algorithm are analyzed and evaluated respectively, and called TOA-DTB, TOA-ANN and TOA-WKNN correspondingly. The experiment results show that the TOA-DTB method is superior to TOA-WKNN and TOA-ANN in complex indoor scenarios. By only using 2 Anchors in dense NLOS environment, the probability of positioning error less than 54 cm is 90%, probability of positioning error less than 38 cm is 80%, less than 30 cm is 64%. Thus, the proposed method can meet the needs of accurate indoor positioning for smart mobiles in dense NLOS environment with low anchor deployment density, which means a high values for real application and promotion.