Efficient Algorithms for Traffic Grooming in SONET/WDM Networks

Abstract

In SONET/WDM optical networks, a wavelength channel is shared by multiple low-rate traffic demands. The multiplexing is known as traffic grooming and carried out by SONET add-drop multiplexers (SADM). A key optimization problem in traffic grooming is to minimize the number of SADMs. This optimization problem is challenging and NP-hard even for unidirectional SONET/WDM rings (UPSR) with symmetric unitary traffic demands. In this paper, we give a linear time heuristic algorithm for this NP-hard problem. Empirical results show that the algorithm outperforms previous algorithms. The algorithm uses the minimum number of wavelengths, which are also precious resources in optical networks. An important subclass of the symmetric unitary traffic pattern is the regular traffic pattern, where each network node appears in exactly r symmetric demands. The regular traffic pattern is a generalization of the well known all-to-all traffic pattern, in which r = n - 1 for a network of n nodes. We prove that the optimization problem remains NP-hard for the regular traffic pattern on the UPSR. We also propose an algorithm for this problem with a better upper bound on the number of used SADMs than previous algorithms. This algorithm always uses the minimum number of wavelengths as well