Indoor localization is one of the issues of the research community and many algorithms, solutions proposed and implemented. In this thesis we focus on scalable, decentralized and RSSI map based localization system for indoor environments. A mobile device in wireless sensor networks can determine its own position by field strength measurements of beacon node signals. Predefined selected trained point measurements are collected and saved in the database with the position and additional information. This method is called received signal strength fingerprinting. The location determination considers the current and saved measurements then estimates the location with minimal distance to the current measurements.
In this fingerprinting localization, complexity and scalability of signature (RSSI) database is very significant. In addition to this, signature size should be small in size. During the distance estimation process only those signatures which are exactly related to mobile device location has to be given by the system. For that objective, we proposed enhanced signature database scheme and spline interpolation based relevant signature retrieval method to achieve better performance and scalability on fingerprinting localization systems.
One of requirements of decentralized localization system is estimation algorithm which should be lightweight because whole system runs on low power based sensor nodes. We propose inverse weighting algorithm to calculate mobile node location using current and offline signature database. This scheme can get better accuracy comparing Motetrack’s centroid averaging approach. We evaluated and compared performance of the proposed algorithm and methods with previous Motetrack [1] through simulation. Results of simulations show that proposed methods have better performance and accuracy than Motetrack system.
