Diffusive capture processes for information search

Abstract

We show how effectively the diffusive capture processes (DCP) on complex networks can be applied to information search in the networks. Numerical simulations show that our method generates only 2% of traffic compared with the most popular flooding-based query-packet-forwarding (FB) algorithm. We find that the average searching time, $<T>$, of the our model is more scalable than another well known $n$-random walker model and comparable to the FB algorithm both on real Gnutella network and scale-free networks with $\gamma =2.4$. We also discuss the possible relationship between $<T>$ and $<k^2>$, the second moment of the degree distribution of the networks.