Energy Efficient Homogeneous Wireless Sensor Network Using Self- Organizing Map (SOM) Neural Networks

Today, Wireless Sensor Network (WSN) is becoming an interesting research area for wireless communication in very harsh or hostile environment. In WSN, limited battery power is considered as the main constraint; due to which the network lifetime is very low. To overcome this problem, many types of improvement have been carried out in both hardware and software levels. But still, there is a much more need to improve. Data clustering or classification using artificial neural networks (ANN), an emerging area of artificial intelligence is a step forward to enhance the network’s lifetime by means of optimizing some of its parameters like power battery backup, data traffic, end-to-end delay. Now-a-days, ANN has become one of the most popular techniques for solving real time optimization problems. In this paper, Kohonen’s Self-Organization Map (SOM) neural network algorithm has been efficiently used for data clustering; that learns to classify data without any supervision i.e. in unsupervised learning mode. We have analyzed and reduced the real data to make the network less bulky, communication gets faster as due to lager volume of data is get reduced, and end-to-end delay and power consumption of communication network also gets lowered.