Measuring Peer-to-Peer Network Topology through Geo-Location-Aware Distributed Crawlers

摘要

Network topology information of peer-to-peer networks is useful for improving their performance as well as for generating an accurate model of topologies for simulations. Unfortunately, dynamic behaviors of peers such as join and leave makes it difficult to obtain the accurate network topology information. In this paper, we propose a method to accurately measure the topology of a Winny Network even under heavy churns by observing search queries through a geographically distributed crawler systems. Our method achieves high completeness of measurement with various techniques for distributed crawling based on geographical locations of Winny clients. We also show that a tree-structure crawling architecture is more effective than a single server system using queueing model and simulation.%P2Pファイル共有システムにおける検索リンクの形状（トポロジ）情報はネットワークの性能向上や現実的なシミュレーションの実施に必要である・しかし，ピアの動的な参加・離脱のため正確なトポロジ情報を得ることが困難である．本稿では，Winnyネットワークを対象とする検索リンクのトポロジを，検索クエリを用いて測定する方式を提案し，得られたトポロジ情報の結果の完成度に対して評価する．地理情報を用いて、さらに測定結果の完成度が高くなることを示し、木構造アーキテクチャによる改善のキューイングモデルおよびシミュレーションによる評価も示す。