An Iterative Solution to the Optimal Premium-Class Routing Problem

This paper studies the problem of finding optimal routes for the premium-class traffic such that (1) it works correctly in the context of hop-by-hop routing for DiffServ networks; and (2) it maximizes the residual bandwidth on those bottleneck links so that the negative influences imposed by the premium traffic onto the best-effort traffic are minimized. This problem is called the Optimal Premium-class Routing (OPR) problem which is NP-Complete. We develop a novel heuristic algorithm, called the iHOP (iterative Hop-by-hop rOuting algorithm for the Premiumclass traffic). The algorithm is highly flexible and extensible. Its correctness and complexity are addressed. We show that our algorithm outperforms the conventional shortest-path routing algorithms. We also conduct extensive simulations to study the performance as well as the convergence behavior of iHOP. The results show that the iHOP algorithm yields significant performance gain in term of the saturate bandwidth.