| Wireless sensor networks (WSNs) for environment monitoring consist of a large number of small, low cost, battery-powered communication devices (so-called nodes), densely deployed throughout a remote or inaccessible physical space. These networks are widely used for systematic gathering of useful information related to the surrounding environment, such as temperature, humidity, seismic and acoustic data.; Energy/power conservation represents the key challenge in the design and operation of WSNs for environment monitoring, due to the expectation of unattended network operations over long periods of time, with rare or no possibility for replacement of node batteries. Node clustering is commonly considered as one of the most promising techniques for dealing with the given challenge. Namely, WSNs of cluster-based organization have the potential to reduce the volume of node-to-node and node-to-sink communication, thus reducing the overall energy consumption and extending the network lifetime.; In Chapter 3 of the thesis, we present a thorough theoretical analysis of clustered and non-clustered WSNs, and conclude that under a high degree of in-cluster data aggregation, clustered WSNs outperform their non-clustered counterparts. However, under a low level of in-cluster data aggregation, clustered WSN are certain to lose their performance supremacy.; In Chapter 4, we discuss the issue of optimal cluster size---how big the formed clusters should be. Our theoretical findings indicate that, energy-wise, a clustering algorithm that produces a smaller number of relatively large clusters is not necessarily more efficient than the one that results in a larger number of relatively small clusters. Considering the complexity of WSN implementation and the nature of data gathered in real-world environments, we conclude that clusters of 2-hop radius size provide near optimal performance in term of energy conservation.; In Chapter 5, we propose a new distributed WSN clustering algorithm: Local Negotiated Clustering Algorithm (LNCA), which employs the similarity of nodes' readings as the main criterion in cluster formation. In its general form, LNCA is capable of forming clusters of any arbitrary size, while requiring minimal exchange of gathered data---only one hop away (i.e., only among immediate neighbors). LNCA is shown to be highly effective in dealing with two main challenges faced by clustered WSNs for environment monitoring: (1) effective data aggregation; (2) minimization of the number of isolated nodes.; In Chapter 6, we provide a simulation-based performance comparison of multi-hop LNCA and two existing WSN clustering algorithms---Low-energy Adaptive Clustering Hierarchy (LEACH) and Weighted Clustering Algorithm (WCA). The obtained simulation results indicate that LNCA outperforms WCA and LEACH, for a wide range of cluster sizes and environment scenarios. At the same time, 2-hop LNCA is shown to be the most energy-efficient among all the examined algorithms, thus confirming the correctness of our theoretical findings from Chapters 3 and 4. |
