Indoor positioning system using IEEE 802.11 WLAN channel estimates as fingerprints
Indoor positioning systems (IPS) are emerging technologies due to an increasing popularity and demand in location based service (LBS) indoors because traditional positioning systems such as GPS are limited to outdoor applications. Several IPS are proposed in literature, having the Wi-Fi based positioning system (WPS) as the most promising due to its superior accuracy and wide infrastructure deployment, making it a cost-effective choice. Several WPS have been proposed in the past, from which the best results are shown by so-called fingerprint-based systems. This thesis proposes an indoor positioning system which integrates traditional WLAN fingerprinting by using the received signal strength indicator (RSSI) with channel estimates for improving the classification accuracy with a low number of Access Points (APs). The channel measurements such as impulse response or others, characterize complex indoor area with strong multipath which is quite unique for each indoor location, thus providing a unique signature for better location dependent radio-map pattern recognition. The thesis first proposes an instrumentation design for extracting channel estimates using a Software-Defined Radio (SDR) environment. The instrumentation is designed using an NI-USRP peripheral and LabVIEW software. The wireless measurements are captured in offline mode, surveying the radio-map of a known indoor area. Then indoor positioning using channel estimates is proposed for several scenarios with low number (one and two) access points (APs) when traditional fingerprinting technologies do not work well. Three surveying granularities in grid locations are studied: 4, 8 and 12 ft. A Support Vector Machine (SVM) is used as the algorithm for pattern recognition of different locations based on the samples taken from RSSI and channel estimation magnitudes.