Abstract: Current super-peer search adopts flooding or random walk in routing method and their efficiencies are low.Meanwhile system does not combine the unstructured proximity search with structured data locating effectively. This paper proposed a new kind of super-peer topology based on hierarchical quadrant space callai Quad and introducai two unstructured super-peer search methods. The first one is backtracking and expanding search (BES) technique which attwins a tradeoff between flooding and random walk. BES considers both network traffic and search length. Second, bloom filter technique is adopted to improve the BES. Besides, Quad supports structured data locating. Simulations show that Quad super-peer search methods are better in search success ratio, search cost compared with current super-peer search techniques. Bloom Filter can effectively increase search efficiency and reduce search length. Meanwhile Quad is efficient in data locating.

Key words: P2P network, Super-peer, Search, Hierarchical quadrant space