Energy Efficient Routing with Effective Clustering Mechanisms in Wireless Sensor Networks

Abstract

Recent development in sensor technology and wireless communication has motivated the development of billions of inexpensive sensor nodes, which attracts various applications of Wireless Sensor Networks (WSNs). Because of large application areas, performance metrics for any WSN are strictly application specific. However, energy conservation can be determined as a common metric for any successful application of WSN due to energy constraint of nodes. Also, it is very difficult to replace or recharge batteries of sensor nodes that monitor hostile environments, in which contemporary monitoring schemes requiring human intervention are risky, inefficient and sometimes infeasible. This leads to network longevity as one of the challenging issues of the sensor network. In the sensor network, energy consumption of the nodes is dominated by two major tasks. First, communication happens between the nodes due to the transmission or reception of the packets; and second, during the time spent by the node to listen or detect any event. Therefore, there are two different approaches for energy conservation. Energy can be conserved either by reducing transmission or reception of packets through load balancing; or reducing idle listening time through proper duty-cycling during event detection. The first approach covers, load balancing during routing, while the second one deals with the duty-cycling technique. Both of these approaches can be managed with effective clustering using cross layer optimization schemes. In this thesis, initially load balancing through energy efficient clustered routing techniques for static homogeneous wireless sensor network have been studied and investigated. Later, cluster head election techniques for energy efficient routing for static homogeneous clustered wireless sensor network have been developed by considering various parameters that affect the network lifetime. A cluster head election technique for energy and delay constrained applications of wireless sensor networks has been proposed and analyzed using two different types of distances between the communicating nodes. A protocol is proposed for energy efficient routing through Cross-layer Design, wherein, Ant Colony Optimization (ACO) is used to elect the cluster heads. Later, an approach is proposed to minimize cluster formation overhead by relaxing maximum energy criteria to elect cluster heads along with a scheme that enforces a node to become cluster head, if it has not decided to become cluster head or member node within the stipulated time. An algorithm is proposed to minimize the standard deviation in the number of cluster heads per round. It also reduces the difference of the relative distances of elected cluster heads from the sink. Both of these help to achieve near optimal load balancing for the cluster heads during each round. Later, a Bollinger Band based cluster head election scheme is proposed. Also, a method is devised to compute optimal cluster (grid) size to prolong the network lifetime. A data forwarding scheme is also proposed that results in the multi-hop routing, in which, elected cluster heads work as the data forwarders. This helps to improve the network lifetime by reducing idle listening time of the nodes, including cluster heads. It also manages proper duty-cycling for all the nodes. A distributed approach is proposed to minimize the cluster formation overhead. The proposed approach gives a fair chance to each node to become cluster head. This approach allows scheduling of data messages within the cluster along with the proper management of the duty-cycling time of the nodes. At last, a method is proposed for a heterogeneous network that dynamically computes the number of cluster heads for each round by considering remaining network energy and number of alive nodes in the network. The proposed approaches have been evaluated using extensive simulations.