Optimal clustering in wireless sensor networks employing different propagation models and data aggregation techniques

Wireless sensor networks have recently emerged as a cost-effective embedded sensing and signal processing combination for many applications. The convergence of developments in micro-electromechanical systems (MEMS), increased processing capacity of microelectronics, and advancements in wireless communication electronics has made wireless sensor networks viable. Applications typically demand that wireless sensor networks be left unattended. Efficient sensor battery usage is therefore an important objective. Clustering is a technique that can minimize energy consumption in a network, as well as provide scalability. In clustering techniques, a subset of nodes are clusterhead nodes that function as processing centers, receiving data from groups of nodes and thus partitioning the network into clusters. The clusterheads aggregate the data and transmit them to a base station. This thesis develops a clustered wireless sensor network energy model and determines the optimal number of clusterheads for minimizing the energy consumed in the network. Results from stochastic geometry are applied to the analysis of both a single level network and a multi-level network. This approach is used on a basic wireless sensor network model that assumes free space propagation loss for intra-cluster transmissions and multi-path propagation loss for transmissions from clusterheads to the base station. The energy model is then applied to networks that employ linear and nonlinear aggregation, and include error control. The optimal number of clusterheads is determined, given the node electronic energy expended, the type of aggregation employed, the propagation loss, and the network geometry. The effect of these parameters on the optimal number 'of clusterheads is analysed. A general model for wireless sensor network energy is developed using the log-distance path loss model, which agrees well with experimental results. The general model includes linear and nonlinear aggregation, and error control. The effect on the optimal number of clusterheads of the crossover distance, the number of nodes in the network, the type of data compression, and network geometry, is analysed. The analytical model is verified by simulation of the basic wireless sensor network model, network models that employ linear and nonlinear aggregation, and error control, and a general model that includes log-distance path loss.