Geography Reference
In-Depth Information
8.5
Discussions and Conclusions
Evaluating the robustness of city road networks have been the focus of various
disciplines for a long time. Most of existing researches focus on the role of single
road segment or street and try to evaluate how the failure of single road may affect
the urban road network. Our results prove that the robustness evaluation varies in
different granularity network models due to the structural differences. This implies
that choose the appropriate granularity for network is essential for the urban road
network structure analysis.
In this article, we proposed a comprehensive robustness analysis procedure using
three types of urban road network model in different granularities in this paper.
Moreover, three successive attack strategies are carried out to evaluate urban road
network performances under pressure. This procedure can provide the concept
of structural robustness of the entire urban road network rather than single road
perspective. In addition, a series figures are used to describe the structural forms
of the urban road network under attacks. These figures can provide visuals on the
structural changes in the urban road network.
The comparisons between different models show that the three dual graphs
provide different network structural information and properties. According to the
comparison, the homogeneous degree distribution makes it difficult to identify the
important nodes in the segment based dual graph. The lack of sensitivity of segment
based dual graph under degree based attack suggests this granularity of view cannot
provide the network performance correctly under degree based attacks. The stroke
based dual graphs have a diverse degree and betweenness distribution, and sensitive
to the target attack. But stroke based dual graphs show no remarkable separation
occurring under intentional attacks that cannot truly reflect the reaction of urban
road networks. In addition, in the real world, a stroke especially the main roads
are always very long and cross over several areas. Under normal circumstances, the
incidents only cause several road segments and junctions blocked. Even the extreme
situation rarely leads to the malfunction of the whole stroke. Thus the stroke-based
network is not appropriate to evaluate network robustness.
The community based dual graph groups the topological correlated road seg-
ments into subsets, which takes the spread of traffic jam into consideration. The
attack to a node implies the possible response when the real urban street network
is attacked to a road segment or junction and affects the segments near it. The
experiments show sensitivity under target attacks, suggest a way to identify the
crucial community in urban road and imply a framework to compare the robustness
between cities. So we recommend the community based dual graph for urban road
networks to evaluate the network structure robustness.
Further work beyond this topic can be considered in different point of views:
Firstly, although we recommend the community based dual graph, the community
detection is an NP hard problem, how to choose the appropriate algorithms requires
more work. Secondly, different application requires different network model, how
to find the right model for a certain situation is still an open question. Thirdly, since
Search WWH ::




Custom Search