NeTS-NBD: Characterizing Large-Scale, Dynamic Peer-to-Peer Networks: New Sampling and Modeling Approaches

尽管P2P通信范式的重要性和P2P应用在互联网上的深远影响，但有关大规模P2P应用动态的基本特性尚未得到很好的理解。导出这些特征的基本挑战是捕捉大规模P2P系统随时间演变的代表性（即无偏见）观点。该项目研究并开发了新的测量和建模方法，以理解和准确地表征大规模P2P系统的动态特性。例如，pi在估计对等连接和对等属性时识别出偏差的拓扑和时间来源。pi还研究了采样动态图的想法，作为在大型动态P2P系统中捕获代表性测量的新方法。利用经验和建模工作（以及仿真）的创新组合来准确地推导P2P系统的关键特征，确定其潜在原因，并研究其对未来P2P系统的设计和评估的影响。更广泛的影响：本项目贡献的资源对P2P系统的研究和开发具有深远的影响，包括(i)一套新的高保真测量方法和工具，（ii）一类解释现有P2P系统动态特性的新型进化模型，（iii）实际P2P系统的公开测量档案，以及（iv）一个新的模块化会话级P2P模拟器。由此产生的测量方法、模型和工具将被纳入俄勒冈大学本科和研究生阶段新开发的课程。

Despite the importance of the peer-to-peer (P2P) communication paradigm and the far reaching impact of P2P applications on the Internet, fundamental properties concerning the dynamics of large scale P2P applications are not well understood. The fundamental challenge in deriving these characteristics is to capture a representative (i.e. unbiased) view of large scale P2P systems as they evolve over time. This project investigates and develops new measurement and modeling methodologies to understand and accurately characterize properties of the dynamics of large scale P2P systems. For example, the PIs identify the topological and temporal sources of bias in estimating peer connectivity and peer properties. The PIs also examine the idea of sampling dynamic graphs as a novel approach to capture representative measurements in large and dynamic P2P systems. An innovative combination of empirical and modeling efforts (along with simulation) is leveraged to accurately derive key characteristics of P2P systems, identify their underlying causes, and study their implications for the design and evaluations of future P2P systems.Broader Impact:The resources contributed by this project have far-reaching implications on the research and development of P2P systems and include (i) a set of new high-fidelity measurement methodologies and tools, (ii) a novel class of evolutionary models that account for the dynamic properties of existing P2P systems, (iii) a publicly available archive of measurements of actual P2P systems, and (iv) a new modular session-level P2P simulator. The resulting measurement methodologies, models and tools will be incorporated into newly developed courses at the undergraduate and graduate levels at the University of Oregon.