基于节点热度的改进Gnutella2搜索算法

An Improved Search Algorithm for Gnutella2 Based on Node's Popularity

Gnutella2的搜索算法在超级节点间转发消息采用泛洪机制,在大规模文件共享中产生较多冗余消息,降低了搜索效率。为提高搜索效率,提出一种基于节点热度的改进搜索算法。该算法根据P2P网络中的幂律特性,提出节点热度概念,根据节点热度选择转发超级节点,设计了基于节点热度的搜索算法;在节点热度的计算中,引入惩罚因子减少无效搜索;在搜索消息转发时,通过最热轮空策略来均衡高热度节点的负载。对实验结果的分析表明,该算法在保证搜索命中率的前提下,能有效减少平均搜索时间和平均跳数,提高了搜索性能。

The Gnutella2 search algorithm supernodes forward search messages by using flooding mecha- nism, which leading to redundant messages and inefficiency in large-scale file-sharing system. To improve the search efficiency, an improved search algorithm based on nodes＇ popularity for Gnutella2 is proposed. According to power-law property in the P2P network, the conception of nodes＇ popularity is proposed. A search algorithm based on node＇s popularity is designed; When the nodes＇ popularity is calculated, the penalty factor is used to reduce the invalid search; When forwarding the search messages, the hottest su- pernodes are rejected to balance the high popularity supernodes＇ load. The experimental results show that the NPS algorithm can reduce the average search time and hops while insuring the search hit rate and improving the search efficiency.