A New Load Balancing Approach for Parallel FP-Growth

Due to the exponential growth in worldwide information, companies have to deal with an ever growing amount of digital information. So the huge size of data and computation volume of new processing applications such as data mining, leads to new high performance parallel processing systems. One of the most important challenges of such application is quickly and correctly finding the relationship between available data. In this paper we target to reach the main goal of parallel processing which is high utilization of processing elements via load balancing and reducing the processing time. To reach an optimum load balancing algorithm, our proposed approach adopts a heuristic based on the weights of frequent item sets. Different scenarios of proposed method are considered and the processing time of those scenarios are obtained by GRIDSIM simulator. Finally, performance metrics of the proposed method such as computation time, Hit ratio of data etc... are extracted through simulation results. Keyword: load balancing, parallel computing, FP-Growth, frequent itemset mining