Date Added: Apr 2010
The authors study the relationship between failure localization and the properties of link restoration algorithms, employing a quantitative measure of a network's ability to recover from two-link failures. This model allows people to consider issues of failure localization that cannot be addressed through models that assume only single failures. Based on the relationship between algorithmic properties and restoration failures, they construct a failure classification hierarchy that provides insight as to the relative value of advances in algorithm design. Finally, they apply this classification scheme to three networks from the literature and discuss the results in terms of their importance for link restoration algorithms.