Cascading failures in spatial complex networks

Yongxiang Xia,Cong Wang,Hui-Liang Shen,Hainan Song

DOI: https://doi.org/10.1016/j.physa.2020.125071

2020-12-01

Abstract:<p>In this paper, cascading failures in spatial complex networks are studied. The model we use, the spatial random geometric network, turns out to be robust to random attacks but very fragile to intentional attacks. Then, the relationship between the topology characteristics and the robustness of spatial networks is studied. By comparing with the Erdős–Rényi (ER) random network, the Barabási–Albert (BA) scale-free network and the Watts–Strogatz (WS) small-world network, the "robust yet fragile" property of spatial random geometric network is related to its heterogeneous betweenness distribution. Finally, the influence of parameter settings on the network robustness is studied. The results show that both the increase of the size and the expansion of the communication range of nodes will increase the average degree of the network, thus enhancing its robustness under intentional attacks.</p>

What problem does this paper attempt to address?

This paper aims to study the cascading failure phenomenon in spatial complex networks and explore the robustness of these networks under different attack patterns. Specifically, the author uses the spatial random geometric network model to study this problem and finds that this model is highly robust against random attacks but very vulnerable to intentional attacks. In addition, the paper also compares the differences in robustness between the spatial random geometric network and the Erdős–Rényi (ER) random network, the Barabási–Albert (BA) scale - free network, and the Watts–Strogatz (WS) small - world network, and analyzes the relationship between network topological characteristics and robustness.

### Main research questions:
1. **Mechanism of cascading failures**: Study how the attack on a node in the network can lead to a series of overload failures and may eventually cause a large number of nodes to fail.
2. **Robustness comparison of different network models**: By comparing the spatial random geometric network with the ER random network, the BA scale - free network, and the WS small - world network, explore their differences in robustness when facing random attacks and intentional attacks.
3. **Influence of network topological characteristics**: Analyze how the degree distribution and betweenness distribution of the network affect its robustness, especially the reasons for the "robust yet fragile" property.
4. **Influence of parameter settings**: Study how changes in the number of nodes and the communication range affect the robustness of the network.

### Key conclusions:
- The spatial random geometric network is highly robust against random attacks but very vulnerable to intentional attacks.
- The uneven betweenness distribution of the network is the main reason for the "robust yet fragile" property.
- Increasing the number of nodes or expanding the communication range can improve the robustness of the network because this will increase the average degree of the network.
- Protecting high - load nodes (i.e., nodes with the highest betweenness) is particularly important for improving the robustness of spatial networks.

### Formula summary:
- **Connection probability**:
\[
p = V(R)=\pi R^{2}
\]
- **Node degree distribution**:
\[
P(k;x;R)=\binom{N}{k}q_{R}(x)^{k}(1 - q_{R}(x))^{N - 1 - k}
\]
- **Node load**:
\[
L(i)=\sum_{s\neq i\neq t\in V}\frac{\sigma_{st}(i)}{\sigma_{st}}
\]
- **Node capacity**:
\[
C_{i}=(1+\alpha)\cdot L_{i}(0)
\]

Through these studies, the paper provides an important theoretical basis for the future design and optimization of spatial networks, especially in emerging technology fields such as the Internet of Things and drone swarms.