An Efficient Data Extraction Mechanism for Mining Association Rules from Wireless Sensor Networks

With the advances of wireless sensor networks and their ability to generate a large amount of data, data mining techniques to extract useful knowledge regarding the underlying network have recently received a great deal of attention. However, the stream nature of the data, the limited resources, and the distributed nature of sensor networks bring new challenges for the mining techniques that need to be address. In this paper, we introduce a new formulation for the association rules, a well known data mining technique, that is able to generate the time relations between sensor devices in a particular sensor network. This new formulation will allow traditional data mining algorithms proposed to solve the classical association rules mining problem to be applied on sensor based class of applications that generate and use sensor data. The generated rules will give a clear picture about the correlations between sensors in the network and can be used to make decisions about the network performance, or it can be used to predict the sources of future events. In order to prepare for the data needed in the mining process and to maximize the network lifetime, a distributed extraction methodology is introduced, in this distributed methodology sensors perform optimization based on local computation to decide whether it will participate in sending data or not. Experimental results have shown that the distributed extraction solution is able to reduce the number of exchanged messages and the data size by 50% compared to a direct transmission of the data.