ENHANCED FEATURE SPACE CLUSTERING VIA SPECTRAL PARAMETER WEIGHTING FOR ACOUSTIC FINGERPRINTING BASED INDOOR LOCALIZATION

Chatumal Perera, Sharada Amarakoon, Dulaj Weerakoon, Roshan Godaliyadda, and Parakrama Ekanayake

Keywords

Indoor positioning, location based fingerprinting, acoustic localization, principal component analysis, clustering

Abstract

Location based fingerprinting techniques have become a popular solution for indoor positioning due to their robust performance under non-line of sight and multipath conditions. This article introduces a novel method of constructing location based fingerprints using audible sound via principal component analysis for indoor positioning applications. A special parameter weighting technique is introduced here that warps the feature space to increase the inter-to intra-cluster distance while giving more priority to parameters with higher reliability. This article also extends this idea to generate aggregated clusters that group existing clusters into umbrella clusters. This enables two phases of clustering. First an unknown location is matched to a region containing several grid points, after which the exact location is found within the region.

Important Links:



Go Back