Measure-Based Research Of Kad Network

At present, Peer to Peer (P2P) network is one of the research hot points in the new generation of internet technology. While DHT(Distributed Hash Table) is the trend of P2P network. Kademlia is the most widely applied DHT algorithms. There are many kinds of Kademlia-based network, such as Overnet, eMule, RevConnect and so on. They differ only in details of implement. This thesis adopts the active/passive measure method to study Kademlia algorithm adopted in Overnet network.The main work of this thesis includes: network scale of Kademlia-based Overnet network, validity of the receiving IPs, statistics of different types of files in the network, statistics of IPs with multi-IDs, statistics of IPs corresponding to multi-ports, time characteristics of IP information received from outside passively, the effect of fake ID, and so on.We modify the KadC program, and connect to Overnet network for measurement. After a great deal of data analysis, we get research results as follows:(1) We study network scale of Overnet network and the types of file;(2) During data analysis, we find that there are two especial node IP: one IP corresponding to multi-ports, and one IP corresponding to multi-IDs;(3) After studying Overnet network with fake node and data statistics, we find that fake node enables local node to receive more outside information, which mainly includes the query information from other nodes.