Flow Allocation with Path Protection in Next Generation Internet Networks

Abstract

In this work, we study the flow allocation problem in Next Generation Internet (NGI) networks. Given a capacitated network, flow allocation consists of a simultaneous routing and bandwidth allocation problem for flows realizing a set of source-destination demands. We consider the case where demands are elastic and the flow realizing one demand can be arbitrary split over multiple paths. Moreover, we ensure that the allocated flows are survivable to network failures using path protection. We investigate the possibility of sharing backup capacity, thus enabling more efficient resource consumption. Depending on the optimization objective, we provide two different approaches to the problem. The first approach maximizes the total allocated volume and the second one maximizes the minimum allocated volume. We build mathematical formulations based on generalizations of the maximum multicommodity flow and the maximum concurrent flow problems. Moreover, we provide new approximation algorithms based on a primal-dual approach. These algorithms compute a solution that is within a guaranteed factor of the optimal. Numerical results enable to get deeper insights on the quality and compared performance of the different approaches.