Despite of low battery power, low memory space and low computational speed wireless sensor network is best resources of data collection and gathering in surrounding areas and it is very useful in battle field, climate monitoring as well as medical field, agriculture and it also observed temperature, humidity, vibration and sound etc. Due to low battery and memory space as well as low computational speed sensor node does not applicable for long time; to overcome such type of problem, data reduction is needed. Various data collection approaches are proposed by researchers like Prediction, aggregation and compression but objects of all theory is to reduce data. in this paper we propose a data collection method which is based on prediction theory and apply least mean square algorithm with variable step size, LMS is best solution for data prediction and collection in WSN because it is not needed any Perrier knowledge of statistical properties of collected data and also reduced 93.5% of redundant data in real dataset when applying on star, tree and cluster networks with few variation.
Volume 11 | 09-Special Issue
Pages: 815-820
DOI: 10.5373/JARDCS/V11/20192637