Shi Xiao

Tolerance of intentional attacks in complex communication networks

Motivated by recent developments in the theory of complex networks, we examine the tolerance of communication networks for intentional attacks that aim to crash the network by taking down network hubs. In addition to providing a brief survey of key existing results, we investigate two different effects that largely have been ignored in past studies. Many communication networks, such as the Internet, are too large for anyone to have global information of their topologies, which makes accurate, intentional attacks virtually impossible; most attacks in communication networks must propagate from nodes to adjacent nodes, utilizing local-network topology information only. We show that incomplete global information has a different impact on intentional attacks in different circumstances, and local information-based attacks can actually be highly efficient. Such insights will be helpful for the future development of efficient protection schemes against network attacks.

Robustness of scale-free networks under rewiring operations

Scale-free networks have strong tolerance against random failures yet are fragile under intentional attacks. Existing results show that the network robustness can also be affected by its correlation profile. Specifically, scale-free networks with larger assortativity coefficients generally tend to be more robust against intentional attack. In this letter, we reveal some interesting different observations. By proposing a simple rewiring method which does not change any nodal degree, we show that network robustness can be steadily enhanced at a slightly decreased assortativity coefficient. The tolerance against random failures meanwhile remains largely unaffected. Such observations demonstrate the more complicated relationship between network robustness and its assortativity level, as well as some new possibilities of network enhancement and protection.

Robustness of Complex Communication Networks under Rewiring Operations

Recent researches on complex networks show that some most important communications systems, including Internet and World-Wide Web (WWW), are scale-free networks with power-law nodal-degree distributions. It is claimed that such networks are robust under random failure yet fragile under intentional attack that takes down network nodes in a decreasing order of their degrees. In this paper, we show that different scale-free networks with exactly the same degree of every node (where one of them could be achieved by rewiring some links in the others) can have different tolerances to the intentional attack. This observation is important in two aspects: (i) It demonstrates that nodal-degree distribution does not define network robustness solely by itself. Additional measures are needed to take other factors into consideration; and (ii) It reveals new possibilities of developing protection strategies against the intentional attack

Robustness of complex communication networks under link attacks

Recent research results show that some most important complex communication systems, which usually can be modeled as scale-free networks with a power-law nodal degree distribution, may be fragile under intentional attacks that take down network hubs. We study the robustness of these networks under deliberate attacks which remove network links. Specifically, we evaluate the extreme case where an efficient graph-partitioning algorithm is applied, based on accurate network-topology information, to decide on the links to be removed. Simulation results show that even such type of calculated link-removal attack cannot easily split a complex communication network. Moreover, among the two split parts, the larger one generally remains as a scale-free network with a very small network diameter. We also consider the case where a specific set of nodes have to be split away from the major part of the network. Simulation results show that applying a graph-partitioning algorithm generally does not lead to a significantly more cost-effective solution than simply removing the given set of nodes together with the links connected to them.

A network flow approach for static and dynamic traffic grooming in WDM networks

In WDM networks, traffic grooming is a fundamental and important operation for aggregating multiple low-rate channels from the end users into a high-rate wavelength channel. In the literature, several traffic grooming problems were formulated where the common objective is to minimize the number of add-drop multiplexers required, and each of these problems was solved by specific algorithms. In this paper, we propose a network flow approach for solving a class of static and dynamic traffic grooming problems. In this approach, we transform a traffic grooming problem onto the minimum edge-cost flow problem through proper graph transformation and assignment of edge weights, and then solve this resulting network flow problem by an efficient heuristic algorithm. The network flow approach has two main advantages: (1) It is generally applicable to many static and dynamic traffic grooming problems. It can solve some existing traffic grooming problems as well as new problems that cannot be solved by the existing algorithms. (2) It performs at least as good as the best existing algorithm for static traffic grooming and significantly outperforms the existing algorithms for dynamic traffic grooming.

Tolerance of local information-based intentional attacks in complex networks

In this paper, we study the tolerance of complex networks against an intentional attack which takes down network nodes in a decreasing order of their degrees. Specifically, we evaluate an effect which has been largely ignored in the existing studies: in many real-life systems such as communication networks, attacks typically propagate throughout the networks from compromised nodes to their adjacent nodes, utilizing at most local network-topology information. By theoretical analysis and numerical simulations, we show that though different local information-based attacks have different performances, generally speaking, they can be highly efficient. Such insight shall be helpful for the future developments of efficient network attack and protection schemes.

On local link repairing in complex communication networks under intentional attack

Recent researches reveal that some most important communication systems, including the Internet and the World- Wide Web (WWW), may be modeled as scale-free networks with power-law nodal-degree distributions. It is found that such networks are robust against random node/link failures yet fragile under intentional attack that takes down network nodes in a decreasing order of their degrees. Various strategies have been proposed to improve network robustness against such kind of attack. In this paper, we propose and evaluate a simple local link repairing strategy. The main idea is to restore some of the links that have been cut when a network node is crashed by rewiring each of them to another node. To simplify the implementations, we restrict the restoration operation to be strictly decentralized with only local information being available to each node. Simulation results show that the simple strategy can significantly enhance network robustness. More interestingly, among the several methods that have been tested, the simplest one with the minimum local information exchanges and limited backup resources appears to achieve the best performance. Such observations provide us some insights useful for the future developments of efficient repairing schemes under the intentional attack.

NISp1-06: On Intentional Attacks and Protections in Complex Communication Networks

Being motivated by recent developments in the theory of complex networks, we examine the robustness of communication networks under intentional attack that takes down network nodes in a decreasing order of their nodal degrees. In this paper, we study two different effects that have been largely missed in the existing results: (i) some communication networks, like Internet, are too large for anyone to have global information of their topologies, which makes the accurate intentional attack practically impossible; and (ii) most attacks in communication networks are propagated from one node to its neighborhood node(s), utilizing local network-topology information only. We show that incomplete global information has different impacts to the intentional attack in different circumstances, while local information-based attacks can be actually highly efficient. Such insights would be helpful for the future developments of efficient network attack/protection schemes.