Unstructured Peer-to-Peer networks consist of an infrastructure-less overlay on top of another network. Most of them use distributed algorithms for all operations, such as resource discovery or connectivity control. Research has shown that a considerable amount of the generated traffic is due to signaling messages. Furthermore, another challenge when implementing a Peer-to-Peer network is avoiding free riders, i.e. users trying to profit from the network without sharing their resources. In this paper a new approach to routing packets in such networks is presented using ant intelligence. Success messages are used as agents and the biological procedure of pheromone trails is used for forwarding new packets used in resource discovery. These agents carry an amount of pheromone which will be added to a pheromone table representing routes to other peers. This approach enables the network to adjust to the dynamic nature of Peer-to-Peer networks where new nodes connect and disconnect continuously. Peers that are free riding will be ultimately isolated from the rest of the network by limiting the number of messages directed to them. The authors have simulated an unstructured Peer-to-Peer network, such as Gnutella, that uses this method and the results are very promising. The amount of traffic used solely for resource discovery is greatly reduced enabling the users to use more bandwidth for transferring content.
- Ant intelligence
- Biologically inspired computing
- Peer-to-Peer networking
- Peer-to-Peer routing