Wireless Positioning and Tracking for Internet of Things in GPS-denied Environments

Wireless positioning and tracking have long been a critical technology for various applications such as indoor/outdoor navigation, surveillance, tracking of assets and employees, and guided tours, among others. Proliferation of Internet of Things (IoT) devices, the evolution of smart cities, and vulnerabilities of traditional localization technologies to cyber-attacks such as jamming and spoofing of GPS necessitate development of novel radio frequency (RF) localization and tracking technologies that are accurate, energy-efficient, robust, scalable, non-invasive and secure. The main challenges that are considered in this research work are obtaining fundamental limits of localization accuracy using received signal strength (RSS) information with directional antennas, and use of burst and intermittent measurements for localization. In this dissertation, we consider various RSS-based techniques that rely on existing wireless infrastructures to obtain location information of corresponding IoT devices. In the first approach, we present a detailed study on localization accuracy of UHF RF IDentification (RFID) systems considering realistic radiation pattern of directional antennas. Radiation patterns of antennas and antenna arrays may significantly affect RSS in wireless networks. The sensitivity of tag antennas and receiver antennas play a crucial role. In this research, we obtain the fundamental limits of localization accuracy considering radiation patterns and sensitivity of the antennas by deriving Cramer-Rao Lower Bounds (CRLBs) using estimation theory techniques. In the second approach, we consider a millimeter Wave (mmWave) system with linear antenna array using beamforming radiation patterns to localize user equipment in an indoor environment. In the third approach, we introduce a tracking and occupancy monitoring system that uses ambient, bursty, and intermittent WiFi probe requests radiated from mobile devices. Burst and intermittent signals are prominent characteristics of IoT devices; using these features, we propose a tracking technique that uses interacting multiple models (IMM) with Kalman filtering. Finally, we tackle the problem of indoor UAV navigation to a wireless source using its Rayleigh fading RSS measurements. We propose a UAV navigation technique based on Q-learning that is a model-free reinforcement learning technique to tackle the variation in the RSS caused by Rayleigh fading