Networked control and efficient transmission in sensor networks

Enabling "intelligent environments" that are effortlessly automated is a key promise of sensor networks of the future. These networks have a wide range of domains in which they can be effectively deployed, including health-care, emergency response, manufacturing and surveillance. Unlike the majority of existing (and perhaps better-understood) network configurations, wireless-implemented sensor networks suffer from extremely stringent constraints in terms of scalability and end-goal of deployment. Thus, it is imperative that we determine solutions that are tailored to the constraints and goals of these systems, by bringing together ideas in the domains of control, computing and communications to a common analytical platform. In this dissertation, we build a theoretical framework that uses system theory, stochastic control, queuing theory and information theory to determine the following: 1. A characterization of the stability and optimal control policies with sensor querying (i.e. which set of sensors must be queried and when) using system theory and stochastic control; 2. A delay-optimal energy efficient transmission scheme for these networks (i.e. with what power level must they communicate) using heavy traffic limits and stochastic control; and 3. A cooperative transmission strategy for maximizing capacity of these networks (i.e. how they should encode their data to send the most through) using network information theory.