Reliable and Energy Efficient Cluster-Based Wireless Sensor Networks

Title: Reliable and Energy Efficient Cluster-Based Wireless Sensor Networks
Authors: Hashmi, Shafiq Ullah
Date: 2011
Abstract: This thesis revolves around the Cluster-Head (CH) operations to provide an energy efficient and data reliable cluster-based Wireless Sensor Network (WSN). The tasks of the CH are very energy intensive and its role is vital in order to keep the network alive and provide maximum network coverage. It requires energy efficient protocols to extend the lifetime of the network and provide reliability to the clusters. To have an energy efficient cluster-based sensor network we propose a new paradigm for MAC layer protocols and provide a Dynamic Slot Assignment (DSA) scheme for TDMA-based MAC protocols in low activity sensor networks. In this scheme, based on the network activity, the connection is established between the cluster-head node and those sensor nodes which have data to send, and a TDMA slot is assigned to each of them dynamically. Then we present Dynamic TDMA Slot Scheduling (DTSS) scheme on top of the DSA scheme to improve the channel utilization and connection blocking probability with the introduction of some packet dropping probability. DTSS achieves this by enabling multi-slot assignment to the sensor nodes and slot-stealing from the less active sensor nodes. A Markov chain based mathematical model has been presented for both DSA and DTSS schemes for correlated connection arrival and we prove the viability and performance of the presented schemes numerically. Usually in the cluster-based WSNs, the CHs are rotated after a round to distribute the CH responsibility. The cluster stability collapses when the CH dies due to any reason. To provide cluster stability, this thesis proposes to have a Backup Cluster-Head (BCH) for those CHs which may not be able to continue for the whole round due to energy constraint. The BCH takes over the CH responsibility before it depletes and becomes the new CH, hence keeps the cluster stable. The performance has been evaluated based on the simulations. A Markov chain model is also presented to analyze the cluster reliability. We use non-homogeneous Markov process, along with Forward Chapman-Kolmogorov equations to illustrate the cluster monitoring period in a finite three state space model. We test the accuracy of the model. The results show that the presented model matches the behaviour of the cluster state transition and validates the simulation results and analysis.
CollectionTh├Ęses, 1910 - 2010 // Theses, 1910 - 2010
NR79717.PDF6.17 MBAdobe PDFOpen