Reliable and energy efficient cluster-based routing protocols for wireless sensor networks

Cluster-based routing protocols for wireless sensor networks have proved to be a very popular and effective innovation. They are inherently energy efficient and scalable owing to the distributed nature and hierarchical organization of sensor nodes, as well as the use of cluster heads in data reception, aggregation and transmission. However, their reliability is very limited because of the potential for sudden break down and the traffic congestion in a cluster head. A wireless communication link is also vulnerable to interference and noise. In addition, to form an optimal cluster is a NP hard problem. These problems make it very challenging to improve the reliability and energy efficiency simultaneously. To address these issues, this thesis proposes a number of clusterbased routing protocols that consider many challenging issues, such as the cluster number determination, the inter-cluster communication cost, the link quality and traffic congestion during the node clustering phase. This thesis contributes four innovative methods that improve both the reliability and energy efficiency of a wireless sensor network simultaneously. The first of these contributions is an optimum backup clustering technique, which reduces the re-clustering overhead of the network and safeguard a cluster head node from sudden break down. The second method, reliable and energy efficient inter-cluster communication, reduces the chance of a cluster head breakdown by developing routing paths that consider the optimal inter-cluster communication cost. This method also considers data loss due to poor link quality and congestion at the CH node. The third method, optimum cluster number determination technique for uniform wireless sensor network, integrates the wireless link quality factor analytically for estimating the optimal cluster number to be used in any suitable clustering protocol. Finally, joint optimization of number and allocation of clusters is introduced, which calculates the optimum cluster number at the time of node clustering. This is applicable in a wireless sensor network with both uniform and non-uniform node distributions. The performance of all the proposed methods is evaluated along with the computational complexity analysis and message overhead. To check whether the method promotes a sustainable environment, performance analysis of the backup clustering scheme has been presented for a certain portion of sensor nodes equipped with a solar cell. Statistical tests confirm that the new clustering methods exhibit significant improvements in terms of both reliability and energy efficiency over the most popular contemporary clustering protocols (e.g. HEED and only one existing backup clustering technique) with the comparable computational complexity and message overhead.