A.K.M. Mahfuzur Rahman; Shimul Shil MEE10:87, pp. 63. COM/School of Computing, 2010.
Increasing popularity of extensively-organized and decentralized Peer-to-Peer (P2P) architecture emphasizes on the need to come across an overlay structure that can provide efficient content discovery mechanism, accommodate high churn rate and adapt to failures in the presence of heterogeneity among the peers. Since the 2-layer hierarchical models improves the P2P by offering some benefits to P2P paradigm and got popularity thereby, hierarchical models are brought into the account of current P2P researchers and still the process is under development. Many designs so far are introduced in hierarchical model like some hierarchical models has CAN or Kademlia algorithm inside them as DHT algorithm. A fundamental task of Peer-to-Peer applications is to find the peer efficiently that stores a desires data item. In our research work, we used 3-layers hierarchical model with distributed database architecture in different layer, each of which is connected through its root. The peers are divided into three categories according to their physical stability and strength. They are Ultra-superpeer, Superpeer and Ordinary Peer and we assign these peers to first, second and third level of hierarchy respectively. Peers in a group in lower layer have their own local database which hold as associated Superpeer in middle layer and access the database among the peers through user queries. In our 3-layer hierarchical model for DHT algorithms, we used an advanced Chord algorithm with optimized finger table which can remove the redundant entry in the finger table in upper layer that influences the system to reduce the lookup latency. Our research work finally resulted that our model really provides faster search since the network lookup latency is decreased by reducing the number of hops. The peers in such network then can contribute with improve functionality and can perform well in P2P networks.