Performance analysis of efficient data distribution in P2P environment using hybrid clustering techniques

Abstract

In this paper, K-means algorithm has been applied for distributed large data using hybrid clustering techniques. K-means is a simple and scalable algorithm which can be applied on large datasets. It is one of the well-known unsupervised clustering algorithms that fail in providing structured to unstructured data to enable extraction of valuable information. Peer-to-peer (P2P) technologies divide the data or resources between the peers for managing the network bandwidth, network participants and processing powers. During the data distribution process in the P2P environments, accuracy, computation complexity and distributed clustering accuracy are the important issues as they reduce the entire system performance. So, the author in this paper considered the system for the distribution of data in P2P environment using mining techniques. The data have been distributed using the hybrid map reducing method which analyzes the large volume of data by performing filtering and sorting. The cluster approach analyzes and manages the neighboring relationship about the peer nodes that helps in the management of the cluster distribution in the dynamic environment. Determination of the efficiency of the cluster formed is done with the help of the hybrid clustering algorithm, and the related system architecture is proposed. The clustering efficiency has been enhanced in the P2P environment using the distributed data network. The efficiency of the formed cluster was evaluated in terms of Jaccard index, F-measures, mutual information and rand measure. The performance of the system was analyzed using the experimental results and discussions, namely, error rate, accuracy and time. The multi-objective system helps in easing the difficulties in the implementation of P2P environment sensitive to initial solutions.