Since decentralized peer to peer networks like Gnutella do not require
any centralized directories or precise control over network topology or
data placement, they are extremely robust. However, flooding-based query
algorithms used by the networks produce enormous amounts of traffic and
substantially slow down the system. Recently flooding has been replaced by
more efficient k-random agents and different variants of such algorithms. In
the presentation, we present biology-inspired
multi-agent systems developed by us for searching peer to peer networks.
Agent behavior is regulated according to an immune-inspired mechanism of opportunistic proliferation and mutation. The autonomous agents multiply according to the availability of contents in the network. They also intelligently track the path followed by the neighboring agents so that collision and consequently repetitive visits of the same node can be avoided. This in turn helps in building up a cooperative and efficient search system. Through a series of experiments, we show agents undergoing opportunistic proliferation and mutation spread much faster in the network and consequently our multi-agent proliferation/mutation system produces better search output in p2p networks than random walk algorithms. |