Abstract: Natural disasters often l

Abstract: Natural disasters often lead to regional failures that can
cause network nodes and links co-located in a large geographical
area to fail. Novel approaches are required to assess the network
vulnerability under such regional failures. In this paper, we investigate
the vulnerability of networks by considering the geometric
properties of regional failures and network nodes. To evaluate the
criticality of node locations and determine the critical areas in a
network, we propose the concept of α-critical-distance with a given
failure impact ratio α, and we formulate two optimization problems
based on the concept. By analyzing the geometric properties of
the problems, we show that although finding critical nodes or links
in a pure graph is a NP-complete problem, the problem of finding
critical areas has polynomial time complexity. We propose two
algorithms to deal with these problems and analyze their time complexities.
Using real city-level Internet topology data, we conducted
experiments to compute the α-critical-distances for different networks.
The computational results demonstrate the differences in
vulnerability of different networks. The results also indicate that
the critical area of a network can be estimated by limiting failure
centers on the locations of network nodes. Additionally, we find
that with the same impact ratio α, the topologies examined have
larger α-critical-distances when the network performance is measured
using the giant component size instead of the other two metrics.
Similar results are obtained when the network performance is
measured using the average two terminal reliability and the network
efficiency, although computation of the former entails less
time complexity than that of the latter.

ผลลัพธ์ (อังกฤษ) 3:

Abstract: Natural disasters often lead to regional failures that can cause network nodes and links
co-located in a large geographical area to
fail. Novel approaches are required to assess the Network vulnerability
under such regional failures. In this paper, we investigate the
vulnerability of networks by considering the geometric properties of
regional failures and network nodes.
To evaluate the criticality of node locations and determine the critical areas in a
network, we propose the concept of camera-critical-distance with a given
failure impact ratio camera, and we formulate two optimization problems based on the concept
.
By analyzing the geometric properties of the problems, we show that although finding critical nodes or links
in a graph is a pure NP-complete problem,The problem of finding
critical areas has polynomial time complexity. We
propose two algorithms to deal with these problems and analyze their time complexities.
Using REAL city-level Internet topology data, we conducted experiments to compute the camera
- critical-distances for different networks.
The computational results demonstrate the differences in
vulnerability of different networks.
The results also indicate that the critical area of a network can be estimated by limiting failure
centers on the locations of network nodes.Additionally,
we find that camera with the same impact ratio, the larger topologies examined have
α - critical - distances when the network performance is measured using the
giant component size instead of the other two metrics.
Similar results are obtained when the network performance is measured using the average
two terminal network reliability and the
, efficiency, although the former entails computation of time
less complexity than that of the latter.