Relaxed computation of non-bifurcation progressive disaster recovery problem

Abstract

We propose relaxed computation for the non-bifurcation progressive disaster recovery problem. When massive failure occurs, failed components are gradually repaired since repair resources are limited. Though there are studies on disaster recovery problem to maximize the amount of recovered traffic considering this assumption, they are based on the maximum flow approach, where traffic bifurcation on an arbitrary node is allowed. This condition is not practical in an actual environment. We first formulate non-bifurcation progressive disaster recovery problem as 0-1 integer linear programming. Because the problem is NP-hard, we present a problem-decomposition method and obtain an improvement of 13% over the benchmark method.