A popularity-based query scheme in P2P networks using adaptive gossip sampling

As P2P networks, such as many forms of social networking have been rapidly growing, numerous efforts have been made to improve the efficiency of the search operation especially in terms of response time and hit ratio. To this end, popularity-based schemes have recently attracted attention aimed at increasing search efficiency using content popularity ranking; however, these methods suffer from high cost and overhead, or inappropriate level of accuracy in specifying the popularity. In this paper, we propose an adaptive sampling scheme to make a tradeoff between cost and accuracy. This scheme relies on exchanging File Index Table (FIT) between peers in a local neighborhood using a Gossip Exchange Method (GEM). The proposed Hybrid Adaptive Search According to Gossip Exchange Method (HAS-A-GEM) is based on smart unstructured peer to peer overlays. We apply a hybrid overlay that efficiently combines topology-aware and interest-based links instead of random or DHT invoked connections. An analytical model as well as a simulation framework is developed to illustrate the performance of this scheme. The effectiveness of the proposed scheme is demonstrated under various conditions. Simulation results reveal that HAS-A-GEM performs well for large-scale networks, exploiting local content popularity when each local area contains enough number of peers.