Data mining algorithms for wireless sensor network's data

Abstract

Classification of sensory data is a major research problem in wireless sensor networks and it can be widely used in reducing the data transmission in wireless sensor networks effectively and also in process monitoring. In order to examine the huge size of data set in stream model generated by sensor network, it will be analyzed different sensor's output signal, topology of sensors network, number of sensor parameters and number of acquisition data. In our wind energy monitoring, sensor node monitors six attributes: speed, direction, temperature, pressure, humidity, and battery voltage. Every attribute value is set as four measures: average, instantaneous, minimum, and maximum. This paper presents several data mining techniques applied on the wireless sensor network's data considered: Naïve Bayes, k-nearest neighbor, decision trees, IF-THEN rules, and neural networks. Before classification, the data was clustered in order to be labeled. A similarity based algorithm, k-means, was selected in the clustering process for its simplicity and efficiency. A conclusion that decision trees are a suitable method to classify the large amount of data considered is made finally according to the mining result and its reasonable explanation.