TwoHop: A Searching Protocol Of P2P Network

Nowadays the P2P network is very popular, and there are plenty of P2P protocols most of which are based on the structure of Distributed Hash Table(DHT). We can see some applications using P2P protocols such as OceanStore based on Tapestry[19][20], and OverNet based on Kademlia[17]. Currently P2P systems have high lookup latency and are therefore only well-suited for applications that do not mind high-latency store and retrieve operations (e.g., backups) or that store and retrieve massive amounts of data (e.g., a source tree distribution). Although OneHop[12] and Kelips[18] guarantee that file lookups are resoved with O(1 ) time and complexity (i.e., independent of system size), the former isn't fitted for the large scalable network because it will consume more bandwith than being tolerated, and the latter needs more time to stabilize the network when some events happened.This dissertation begins with the current research situation and the developing trend of P2P network, and then introduces the current P2P protocols.In this dissertation, we describe a new scalable P2P protocol, TwoHop, partly originally from OneHop[12], using 3 layers in the network. Each node keeps two parts routing information like Kelips[18], one for local lookups, and the other for remote lookups. TwoHop can ensure the low lookup latency and the acceptable bandwith consuming even in the large scale network (e.g., 108 nodes).In the simulation, we can see differences between OneHop and TwoHop. TwoHop is suitable for the large scale network and in the relatively stable environment.There are no applications using TwoHop at present, and the C++ codes in simulations partly accomplish functions and goals in TwoHop protocol.