PERFORMANCE ANALYSIS AND OPTIMIZATION OF QUERY-BASED WIRELESS SENSOR NETWORKS

摘要

This dissertation is concerned with the modeling, analysis, and optimization of large-scale, query-based wireless sensorudnetworks (WSNs). It addresses issues related to the timeudsensitivity of information retrieval and dissemination, network lifetime maximization, and optimal clustering of sensor nodes in mobile WSNs. First, a queueing-theoretic framework is proposed to evaluate the performance of such networks whose nodes detect and advertise significant events that are useful for only a limited time; queries generated by sensor nodes are also time-limited. The main performance parameter is the steady state proportion of generated queries that fail to be answered on time. A scalable approximation for this parameter is first derived assuming the transmission range of sensors is unlimited. Subsequently, the proportion of failed queries is approximated using a finite transmission range. The latterudapproximation is remarkably accurate, even when key modeludassumptions related to event and query lifetime distributions and network topology are violated.ududSecond, optimization models are proposed to maximize theudlifetime of a query-based WSN by selecting the transmissionudrange for all of the sensor nodes, the resource replicationudlevel (or time-to-live counter) and the active/sleep schedule of nodes, subject to connectivity and quality-of-service constraints. An improved lower bound is provided for the minimum transmission range needed to ensure no network nodes are isolated with high probability. The optimization models select the optimal operating parameters in each period of a finite planning horizon, and computational results indicate that the maximum lifetime can be significantly extended by adjusting the key operating parameters as sensors fail over time due to energy depletion.ududFinally, optimization models are proposed to maximize theuddemand coverage and minimize the costs of locating, andudrelocating, cluster heads in mobile WSNs. In these models, the locations of mobile sensor nodes evolve randomly so that each sensor must be optimally assigned to a cluster head during each period of a finite planning horizon. Additionally, these models prescribe the optimal times at which to update the sensor locations to improve coverage. Computational experiments illustrate the usefulness of dynamically updating cluster head locations and sensor location information over time.