DRWT: An Efficient Random Walk Algorithm for Unstructured P2P Networks

Yiming Zhang, Xicheng Lu, Dongsheng Li, Nong Xiao

Abstract: In this paper, we present a novel search algorithm called Dynamic Random Walk Team (DRWT), which achieves a better tradeoff between performance and overhead. The main difference between DRWT and the traditional Random Walk algorithms contains two aspects: (1) all nodes advertise their resource sharing information and maintain and broadcast the information with the EDBF-like manner, which discards the information dynamically when transmitted to neighbors; and (2) DRWT extends the concept of walker in traditional Random Walk to search team, which selects its search direction based on the resource location information at each intermediate node. Furthermore, each search team periodically contacts the requesting node and may be required to enhance its search intensity by sending out more walkers based on the distribution of the resource location information to accelerate the search process.