Journals Proceedings

Abstract

Wireless sensor network is highly data centric. Data communication in wireless sensor network must be efficient one and must consume minimum power. Every sensor node consists of multiple sensors embedded in the same node. Thus every sensor node is a source of data. These raw data streams cannot be straightway communicated further to the neighboring node. These sensor data streams are first classified. A group of sensor nodes forms a cluster. Each node transfer data to a cluster head and then cluster head aggregates the data and sends to base station. Hence clustering and classification techniques are important and can give new dimension to the WSN paradigm. Basically, classification system is either supervised or unsupervised, depending on whether they assign new inputs to one of a infinite number of discrete supervised classes or unsupervised categories respectively. ART1 and Fuzzy ART are unsupervised neural network models which are used for classification of sensor data. ART1 model is used for classification of Binary valued data. While Fuzzy ART model can be used for analog data, wherein the input data is fuzzy valued.