Modeling and Scheduling for Sensor Network Quality-of-Experience Enhancement

The rapid advancement in low-cost sensor node design has led to the possibility of utilizing them in various monitoring applications such as environment monitoring (including inaccessible terrain), structural health monitoring and healthcare. Many of these safety-critical applications however, depend on the data obtained from these sensors to provide effective and timely response. Therefore, providing satisfactory quality of experience (QoE) to a user application is highly important. In sensor networks, QoE can be described as a measure of the network’s ability to consistently provide quality of information (QoI) while being functional for an expected lifetime. Thus, the two quantities characterizing QoE are: QoI and network lifetime. Quality of information, as the name suggests, defines the quality of the information received by an application, which is determined by a number of factors such as network dynamics, physical environment variations and varied application criteria. Expected lifetime is a term considered specifically for battery-powered sensor applications. We define it as the energy efficiency enhancement provided by a system to prolong the operational lifetime of a sensor network. This dissertation addresses the important and challenging problem of enhancing sensor network QoE. Quality-of-experience is characterized as, achievable maximum energy efficiency, while providing assured QoI. As a first step towards addressing this problem, a fundamental signal-to-noise ratio (SNR) based metric quantifying QoI is proposed. This metric addresses the different challenges such as environment unpredictability, network unreliability, varied application requirement and limited energy resource as aforementioned. Second, a comprehensive quality-energy model is proposed to exploit the interdependency existing between these two quantities. Third, utilizing these two metrics, QoE is defined. The proposed QoE metric is quantified as a function of energy saving ratio and quality satisfaction. Furthermore, adaptive sleep and packet scheduling algorithms are designed to demonstrate the usefulness of these metrics in enhancing QoE. Finally, analytical and simulation results presented validate the effectiveness of the different mechanisms we have proposed