A Min-Max Optimization Problem on Traffic Grooming in WDM Optical Networks

In SONET/WDM networks, a wavelength channel is shared by multiplexed low-rate traffic demands. The multiplexing/de-multiplexing is known as traffic grooming and performed by SONET add-drop multiplexers (SADM). The grooming factor, denoted by k, is the maximum number of low-rate traffic demands that can be multiplexed in one wavelength. Since SADMs are expensive, a key optimization problem in traffic grooming is to minimize the total number of required SADMs to satisfy the full connectivity for a given set of traffic demands. In this paper, we study traffic grooming from a different point of view. We consider a Min-Max optimization problem to minimize the number of SADMs at the network node where the number of required SADMs is the maximum over all nodes. We focus on the unidirectional path-switched ring networks with arbitrary duplex traffic demands. We prove the NP-hardness of this min-max optimization problem, and propose a linear time (k+1/2 + 2)-approximation algorithm. We then show that the approximation algorithm achieves the worst case lower bound. We also study the all-to-all traffic pattern, and propose an algorithm achieving solutions only a constant factor away from the optimal ones. Extensive simulations are conducted as well to validate the performance of our algorithm.