Control Design and Filtering for Wireless Networked Systems

This dissertation is concerned with estimation and control over wireless networked systems. Several problems are addressed, including estimator design over packet loss links, control and estimation over cognitive radio systems, modeling and prediction of wireless sensor networks (WSNs), and localization with the Theater Positioning System (TPS). The first problem addressed is the state estimation of a discrete-time system through a packet loss link modeled by a Bernoulli random variable. The optimal filter is derived by employing exact hybrid filtering. The performance of the optimal filter is illustrated by numerical simulations. Next, we consider the problem of estimation and control over cognitive radio (CR) systems. A two-switch model is first used to model this link. The linear optimal estimator and controller are derived over a single CR link. Also discussed here is estimation and control of the closed-loop system over two CR links. Furthermore, a more practical semi-Markov model for the CR system is proposed. Two cases are considered, where one assumes that acknowledgement of the information arrival is not available while the other assumes it is available. In the former, a suboptimal estimator is proposed and, in the latter, sufficient conditions are derived for the stability of a peak covariance process. Then, a controller design for the semi-Markov model is developed using linear matrix inequalities (LMIs). Additionally, the third problem addressed is modeling, identification, and prediction of the link quality of WSNs, such as the packet reception rate (PRR) and received signal strength indicator (RSSI). The state-space model is applied for this purpose. The prediction error minimization method (PEM) is employed for estimating parameters in the proposed model. The method employed is demonstrated through real measurements sampled by wireless motes. The last problem analyzed is localization using a new navigation system, TPS. In this study, we focus on users' position estimation with the TPS when a GPS signal is not available. Several models are proposed to model transmission delays utilizing previous GPS signals. Last, a navigation scheme is provided for the TPS to improve its localization accuracy when the GPS signal is unavailable.