Many recent studies have underlined the critical influence of topology on network survivability. It has been shown [5-7] that different designs react differently to both random node failure and directed attack against key relays. In a closely related work, Pastor-Satorras and Vespigianni [8] demonstrate that the topology of the Internet actually favours the spreading of computer viruses. Finally, among other observations on the properties of complex networks, the review by Strogatz [9] emphasises the role of design on catastrophic cascade failures.
In order to determine to what extent adaptive security procedures can limit the amount of damage caused to a network by the propagation of a malicious entity, it is vital to establish a suitable benchmark. In particular, as one of the key issues addressed here revolves around the fact that topology is likely to become an increasingly dynamic real-time variable in distributed systems, the response of several possible designs should be carefully examined before robust solutions can be discussed. In this section, we investigate different scenarios, involving two unprotected network architectures. One is close to the scale-free topology, widely accepted as a realistic model of the Internet [10], the other is an original hybrid topology termed hypergrid , which we argue has desirable features in terms of robustness and scalability.
One of the most important things to understand when comparing topologies is that they dictate how traffic is distributed in the network. For example, in a scale-free architecture, the power-law distribution of node degree...
TABLE OF CONTENTS 
