Abstract: For decentralized peer-to-peer system, both distributed hash table (DHT) and flooding are deployed in resource search. Hybrid search model combines the advantages of the two techniques and is expected to be efficient for popular and rare contents. However, it must solve how to precisely identify the rare item in decentralized, self-organized P2P network, and current hybrid algorithms are still suffered from the problem of either the efficiency decrease or QoS decay in dynamic environment. And a prelimitary soluiton to this problem is also proposed.

Key words: Decentralized peer-to-peer, Distributed hash table(DHT), Flooding, Hybrid search, Content popularity, Search efficiency and quality

摘要： 对非集中式P2P系统，分布式哈希表和洪泛方法均可被用于资源搜索。结合了这两类方法的混合P2P搜索模式被期望有更好的性能，但需要找到准确判断资源知名度的有效方法。而在动态的P2P网络环境中，现有的混合搜索算法仍面临着搜索效率与质量有待优化的问题，该文在分析现有问题的基础上提出了一种解决思路。

关键词: 非集中式P2P, 分布式希表, 洪泛, 混合式搜索, 内容知名度, 搜索效率与质量