Spatio-Temporal Fingerprint Localization for Shipboard Wireless Sensor Networks

Abstract

Location-aware technology has numerous potential applications in safety, living, and services in the shipboard environment. Fingerprint-based methods that use the received signal strength indication (RSSI) to provide high positioning accuracy have been widely adopted for indoor localization. However, the RSSI can be severely disturbed by the steel structure and dynamics of the shipboard environment, which prevents it from being implemented in shipboard localization in practice. In this paper, a novel approach, called spatio-temporal fingerprint localization, is proposed. This approach can alleviate the impact of the dynamic shipboard environment and enhance the localization robustness. To adapt to the environmental dynamics, a radio time series is proposed to filter irrelevant noise from location fingerprints. An extraction method called radio spatial features is proposed to identify the highly location related features from the redundant RSSI information by using linear discriminant analysis and principal component analysis. Extensive experiments on the passenger ship Yangtze 2 demonstrate the effectiveness of the proposed algorithm in providing higher accuracy than previous fingerprint-based methods.