WDM-based optical transport networks provide huge bandwidth for client services. Typically, each wavelength channel can carry 2.5Gbs or up to 10 or 40Gps capacity. For such a huge amount of capacity, a single normal application service generally cannot fully utilize it. Only a group of fine application services are aggregated together, can the wavelength channel capacity be fully utilized. Under this circumstance, traffic grooming is a terminology or approach that is specially proposed on how to efficiently aggregate those low-rate traffic tributaries onto a large capacity trunk, so as to fully utilize it.

The traffic grooming concept was first defined for a ring-based optical transport network [1], where many low-rate SDH/SONET tributaries (such as OC-1, OC-3, etc.) are aggregated onto wavelength channel trunk, typically, OC-192 (10Gbs) or OC-768 (40Gbs). The objective of the grooming process is to minimize the number of required add/drop ports on OADM nodes given a certain set of demand requests between ring-node pairs. Traffic grooming in a ring-based network can save overall network design cost by minimizing the number of add/drop ports.

Lately, the concept of traffic grooming was extended to mesh-based transport networks [2]. Although conceptually the scheme is functional, there exists the technical challenges on how to design those OXC node that can efficiently support the functionality of traffic grooming. Although it can be an interesting research topic, which did get much attention from academy, it is still doubtful whether this type of network architecture is cheaper than a design that grooms traffic simply based on core IP routers connected to OXC nodes.