为解决目前Random Walk改进算法中过于依赖历史搜索记录而导致动态网络环境下搜索命中率低、网络开销过高和稀有资源的搜索成功率提高不明显等问题,通过分析随机漫步的基本性质和易转向高度数节点的搜索特性,提出了一种双向随机漫步搜索机制——BRWS(bidirectional random walk search),并证明了其能够提高包括稀有资源在内的搜索成功率,抗扰动性强.分别在静态和动态网络环境中,将Random Walk,APS(adaptive probabilisticsearch),PQR(path-traceable query routing),P2PBSN(peer-to-peer based on social network)和BRWS基于RandomGraph、Scale Free网络、Small World网络3种拓扑进行了对比实验.结果表明,BRWS可以以较少的网络搜索代价,极大地提高搜索成功率;并在动态网络环境中,对稀有资源的搜索成功率也有显著提高.所提出的方法可适用于P2P文件分发网络应用中.
This article proposes a cooperative relaying strategy to efficiently utilize the relaying resources of Internet service providers (ISPs),speedup distribution and save server bandwidth costs. ISPs cooperatively relay for each other,and peers assist in distributing and fetching the content as near as possible. Base on the fluid model,a constrained model is derived to get optimized global distribution performance in the channel-based system with limited relaying resources. The multi-objectives of the model are minimizing the average distribution time and the distribution time of the slowest channel. Genetic algorithm (GA) is designed to solve the optimization problem. The relaying strategy based on GA can be run periodically to update the allocation policy of ISPs. The distribution performance of the relaying strategy is analyzed in the experiments and results show that GA can provide proper solutions for various network topologies.
HE Qian ,MENG Xiang-wu,SHANG Yan-lei,CHEN Jun-liang State Key Laboratory of Networking and Switching Technology,Beijing University of Posts and Telecommunications,Beijing 100876,China
