Parallel Evolutionary Association Rule Mining for Efficient Summarization of Wireless Sensor Network Data Pattern

Evolutionary computation is widely used to solve dynamic problems such as association rule mining(ARM) by adapting the solutions to changes of the data pattern. The summarization in the ARM for a wireless sensor network(WSN) is still an issue when its applied to big number of sensor input from multiple location. This paper proposes a parallel processing of ARM for efficient WSN processing using genetic network programming (GNP). The proposed method adopt the concept of "assumption" as rules from the training data and optimize the rules definition to be a "specific rules" for each sensors network location via parallel evolutionary processing. Then summarization is build by calculating hierarchy of "common rules" or similarity between different location. The simulation done using the set of weather forecast sensors. The results shows that the proposed method is capable to efficiently summarize the sensor input from multiple location with online processing and archived a close result to conventional method that performed without on-line processing.