Near Optimal Indoor Localization With Coherent Array Reconciliation Tomography

Koski, Antti

Etd

Near Optimal Indoor Localization With Coherent Array Reconciliation Tomography

Public

<p>Our increased reliance on localization devices such as GPS navigation has led to an increased demand for localization solutions in all environments, including indoors. Indoor localization has received considerable attention in the last several years for a number of application areas including first responder localization to targeted advertising and social networking. The difficult multipath encountered indoors degrades the performance of RF based localization solutions and so far no optimal solution has been published.</p> <p>This dissertation presents an algorithm called Coherent Array Reconciliation Tomography (CART), which is a Direct Positioning Algorithm (DPA) that incorporates signal fusion to perform a simultaneous leading edge and position estimate for a superior localization solution in a high multipath environment. The CART algorithm produces position estimates that are near optimal in the sense that they achieve nearly the best theoretical accuracy possible using an Impulse Radio (IR) Ultra-Wideband (UWB) waveform. Several existing algorithms are compared to CART including a traditional two step Leading Edge Detection (LED) algorithm, Singular value Array Reconciliation Tomography (SART), and Transactional Array Reconciliation Tomography (TART) by simulation and experimentation. As shown under heavy simulated multipath conditions, where traditional LED produces a limited solution and the SART and TART algorithms fail, the CART algorithm produces a near statistically optimal solution. Finally, the CART algorithm was also successfully demonstrated experimentally in a laboratory environment by application to the fire fighter homing device that has been a part of the ongoing research at Worcester Polytechnic Institute (WPI).</p>

Creator