دانلود رایگان مقاله آشفتگی نمودار و تغییرات طیفی مربوطه در توپولوژی اینترنت

عنوان فارسی
آشفتگی نمودار و تغییرات طیفی مربوطه در توپولوژی اینترنت
عنوان انگلیسی
Graph perturbations and corresponding spectral changes in Internet topologies
صفحات مقاله فارسی
0
صفحات مقاله انگلیسی
10
سال انتشار
2016
نشریه
الزویر - Elsevier
فرمت مقاله انگلیسی
PDF
کد محصول
E728
رشته های مرتبط با این مقاله
مهندسی کامپیوتر و مهندسی فناوری اطلاعات
گرایش های مرتبط با این مقاله
شبکه های کامپیوتری
مجله
ارتباطات کامپیوتر - Computer Communications
دانشگاه
لئویانگ تجهیزات الکترونیکی مرکز آزمون، چین
کلمات کلیدی
توزیع طیفی وزن، توپولوژی در حال تحول، توپولوژی اینترنت، طیف لاپلاس نرمال
چکیده

abstract


The normalized Laplacian spectrum (NLS) is a powerful tool for comparing graphs with different sizes. Recently, we showed that two NLS features, namely the weighted spectral distribution (WSD) and the multiplicity of the eigenvalue 1 (ME1), are particularly relevant to the Internet topology at the inter-domain level. In this paper, we examine the physical meaning of the two metrics for the Internet. We show that the WSD reflects the transformation from single-homed nodes to multi-homed nodes for better fault-tolerance and that the ME1 quantifies the initial star-based structure associated with node classification, both of which are critical to the robustness of the Internet structure. We then investigate the relation between the metrics and graph perturbations (i.e., small changes in a graph). We show that these two NLS metrics can be a good choice for study on the Internet optimization. Our work reveals novel insights into the Internet structure and provides useful knowledge for statistical analysis on complex networks.

نتیجه گیری

9. Conclusion


This paper focuses on the relation between graph perturbations and corresponding changes in the WSD and the ME1, and analyzes the physical meaning of the two spectral metrics embedded in Internet AS graphs, e.g., node classification, initial star-based structure, multihomed transformation and core-periphery decomposition. The two metrics reflect the NLS features with eigenvalues not only toward 0 (and 2) but also restricted to 1; i.e., they can be considered as the cost function representing the NLS. Additionally, they are independent of the network size of evolving AS graphs. Therefore, our contributions are useful for understanding the Internet structure and leading future applications of the NLS in AS graphs.


بدون دیدگاه