Localization and Mapping in Wireless Networks: Models and Algorithms

Abstract

In recent years, location-based services (LBS) have become a key component of wireless technology. A large variety of radio-based localization systems, such as the Global Positioning System (GPS), have been developed during last decades. However, satellite-based or cellular systems may have limited availability and reduced accuracy in indoor and dense urban environments. This is partly due to propagation effects such as multipath reflections and scattering of radio signals in such environments. Indoor localization has many important applications such as first responders, tracking and navigation in airports and shopping malls, surveillance, and robotics. The performance of wireless localization, especially in indoor environments, can be improved by cooperation among the network nodes, combining different sensing modalities, and exploiting multipath propagation of radio signals. In this thesis, the problem of cooperative network localization with different sensing approaches is studied. The employed sensing methods are pairwise distance estimation, hybrid distance and direction estimation, and multipath distance estimation. Novel problem formulations for cooperative localization are introduced and optimal algorithms are derived. Algorithms for high resolution distance and direction estimation, and network clock synchronization are also developed in order to support the proposed localization methods. High accuracy is facilitated by the broad bandwidth of wireless signals as well as the use of multiantenna transceivers in most current and emerging wireless systems. The proposed algorithms provide improved reliability and accuracy in comparison to the state-of-the art techniques. A geometrical model or map of the environment is necessary information for multipath-aided localization. Indoor maps are also vital to many other applications such as robot navigation and emergency response. Specular reflections of radio (or acoustic) signals form the walls and other objects in the environment contain rich geometric information, which can be exploited to create indoor maps. Two algorithms for indoor mapping with different sensing modalities are presented in this thesis. Multipath delay estimation, and hybrid multipath delay and direction estimation are employed. Typically mapping algorithms require exact information about sensor positions. Therefore, a radio-based cooperative simultaneous localization and mapping (SLAM) algorithm is also developed in this research work. The proposed algorithm exploits the multipath propagation of radio signals in order to jointly estimate the locations of the nodes and a map of the indoor environment. The proposed algorithms produce indoor maps with higher precision and enhanced geometric information compared to the state of the art.