| Peer-to-peer (P2P) networks are becoming increasingly popular because they offer the significant advantages of decentralized, fault-tolerant and better load balancing systems as well as simplicity and ease of use. Despite the multidimensional advantages of P2P networks, they face some serious challenges. This thesis addresses two major deficiencies of P2P network, which are scalability and efficient search mechanism, and proposes an agent based hierarchical model for P2P systems. The model is based on a semi-centralized hierarchical tree structure and the peers are grouped together based on their network closeness. Each group is joined to a service provider called agent and the agents can send request and response messages on behalf of the peers. This model also proposes an advanced search mechanism based on fastest link calculation to improve the efficiency of the system. Experimental results show that this model provides better network performance in terms of bandwidth and scalability. |
