Handling Large-Scale Data using Two-Tier Hierarchical Super-Peer P2P Network
Rasha Kashef, A Niranjan · 2017
In the ever-expanding world of IoT, data has not only increased in volume and velocity but has also moved from residing in centralized nodes to distributed nodes across multiple locations. Traditional data clustering technologies, based on centralized operations, cannot be scaled to efficiently manage Big Data, thus creating a need for clustering in distributed environments. To address the problem of modularity, flexibility, and scalability, a dynamic hierarchical two-tier architecture and model for cooperative clustering in distributed super-peer P2P network is presented in this paper. The proposed model is called Distributed Cooperative Clustering in super-peer P2P networks (DCCP2P). It involves a hierarchy of two layers of P2P neighborhoods. In the first layer, peers in each neighborhood are responsible for building local cooperative sub-clusters from the local data. Each node sends a summarized view of local data to its super-peer in a form of sub-cluster's centroids extracted from the local cooperative clustering, minimizing the exchange of information between nodes and their super-peers. In the next layer, sub-clusters are merged at each super-peer and at the root of the hierarchy, where one global clustering can be derived. The distributed cooperative approach finds globally optimized clusters and achieves significant improvement in global clustering solutions without the cost of centralized clustering.