Gnutella网络中基于消息跳数的分段搜索策略

A Paragraph Searching Strategy Based on Hops in Gnutella Networks

洪泛机制在查询过程中会产生大量的冗余消息,这制约了网络的可扩展性.在洪泛机制中,低跳时消息的覆盖范围迅速增长而冗余消息的数量相对较少,而高跳时冗余消息的数量迅速增加而覆盖范围增长缓慢.对此,提出了基于消息跳数的分段搜索策略hpsearch.hpsearch策略在低跳时仍采用洪泛算法,以保证查询的高覆盖范围,而高跳时不再向所有的邻居转发消息,仅从邻居节点中随机选取部分作为转发对象.分析和实验结果表明,hpsearch能够在保持高覆盖范围的前提下,明显降低冗余消息的数量,提高搜索效率.

Flooding mechanism is very inefficient because it results in a great amount of redundant messages which greatly restricts the scalability of networks.To solve the problem that coverage grows quickly while the number of redundant message is few when hop is low and redundant message increases quickly while coverage grows slowly when hop is high,this paper proposes a new paragraph searching strategy based on hops named hpsearch.In hpsearch strategy it also use flooding algorithm to keep a high coverage when hop is low and it selects only part of neighbors randomly as the object to forward when hop is high.Analysis and simulation results show that the hpsearch can greatly minimize the number of redundant messages while retaining a high message propagating,and it greatly improves the searching efficiency.