基于带偏随机游动的复制与搜索

Biased random walk-based replication and search

现有的对等搜索策略能够有效解决流行数据项的定位,但是在稀有数据项搜索方面所需的成本相对较高。根据无结构对等网络中节点度的幂律分布特征,提出一种新的数据项复制与搜索策略,通过基于带偏随机游动的方式定向复制数据项到系统中小部分高度节点,资源搜索采用相同的方式进行。分析和模拟结果表明,所提出的资源搜索算法能够以低消息开销和低延迟获得高的搜索成功率。

Although existing schemes could find popular items efficiently, it was very costly for retrieving rare items. According to the power-law degree distribution of unstructured P2P networks, a novel strategy was proposed to replicate and search items. Data items were replicated to some high degree nodes by way of a biased random walk. Resource search was also performed by a biased random walk. Analysis and simulations showed that the proposed algorithms could achieve high search successful rate with low latency and low message overhead.