Signal Tracking and Localization Algorithms for Satellite and Mobile Networks

Abstract

This dissertation deals with the signal tracking and localization problems in satellite and mobile networks. The problems are addressed from both signal processing and networking perspectives. The main objective of this dissertation is to provide several models and approaches that can improve the accuracy and efficiency at the same time. Main results of the dissertation are specified below. The satellite signal tracking problem under interference is investigated for array antenna receivers from the Bayesian filtering perspective. A reduced state signal model is introduced where undesired signals are suppressed with the aid of beamforming. Reducing the state dimension and measurement size, this model significantly decreases the computational complexity of Bayesian filtering. Also, the signal model mismatch from lack of information of the undesired signals is avoidable. Following the proposed model, the problem is translated as the joint beamforming and signal tracking problem. In this regard, a signal tracking algorithm is developed based on the particle filter integrated with a robust hybrid beamformer. The radio map-based localization problem is addressed under the variation of characteristics of received signal strengths due to the changes of device types, device statuses, and environmental factors. From the fact that the signal from each access point experiences different channel characteristics, the received signal strengths of access points are modeled to have different variations and to vary over time by following a first-order autoregressive model. Based on the novel signal model, an extended Kalman filtering algorithm is proposed to track the location of a target device and the received signal strength of each access point at the same time. The proposed approach improves the compatibility of the radio map with heterogeneous devices without effort for updating the radio map. The multihop range-free localization problem is investigated for anisotropic networks with sparse anchors. In order to enhance the localization accuracy with a small number of anchors and local connectivity, methods for detecting whether the shortest path between nodes is detoured or not are developed. Furthermore, a heuristic model and a probabilistic model derived from the geometric understanding are introduced to approximate the shortest path between two nodes given the network connectivity information. Based on these models, two multihop range-free localization algorithms are presented that estimate the distance between nodes by taking into account the extent of detour of their shortest path.