Introduction of clustering
Today wireless sensor networks are
gaining popularity in various fields such as habitat monitoring, robotic toys,
battlefield monitoring, and many more applications including biomedical sensing
and weather monitoring.
Wireless sensor networks are usually composed of hundreds to thousands of sensor nodes, which appear to be randomly spread all over the places. Each node is equipped with sensors to capture information and a communication module to communicate with other nodes. A wireless sensor network collects information and then relay to back-end server through a base station. The sensor networks are usually served by one or more base stations depending on its capability of coverage. The base station acts as a gateway for collecting information from the sensor nodes.
Wireless sensor networks have scarce energy resource and if excessively used will affect the life time of the sensor nodes. Thus energy source is the most critical element in wireless sensor network. Any algorithm developed must avoid unnecessarily utilization of the sensor nodes to save energy.
The algorithms that use centralized control and global properties of sensor network have inherent shortcomings in the approach. For example, each node has limited resource-energy, limited memory and computation power. To achieve the design goals of scalability and robustness of the network is very lame in a large-scale sensor networks. It causes large amount of data transfer which induces heavy traffic and prolonged time delay which is undesirable. An error in transmission (possibly developed in the process) or a failure of a node involved in communication could cause a severe system failure.
In this context, “Clustering” would play a vital role in the operation of wireless sensor networks. The purpose of clustering is to break down the entire network into clusters with which we aim to reduce the centralized control which was necessary. With the Cluster formation, cluster member sensor nodes only require
to communicate with the cluster head which is also known as cluster leader. This leads to energy conservation of member sensors and increases the lifetime of the network, reduces the routing tables and, reduces redundant messages which ultimately saves time.
The most and the only challenging task that the localized algorithm posses is the possibility of excessive utilization of cluster-head due to its leadership role, that may shorten the lifetime of the node. However, it is far more efficient, has longer lifetime and expected better performance in comparison to the centralized algorithm.