Implementation schemes of multihop lightwave networks

This work compares two implementation schemes for multihop virtual topologies in a wavelength-division-multiplexing (WDM) passive star network. The first scheme, denoted by link mapping, is the traditional one-to-one mapping of the virtual topology links to the system wavelengths. In the second scheme, denoted by node mapping, a unique wavelength is assigned to each node (transmitter) in the system. The assignment of wavelengths to the system receivers is determined by the virtual topology. The node mapping scheme is based on dynamic resource sharing. The two are compared in terms of the system hardware receivers and their cost, the system performance, and the capacity per unit cost. Under homogeneous traffic condition, it is shown that unless the transmitter cost dominates the system cost, the capacity per unit cost in the link mapping scheme is always better than the corresponding node mapping scheme. Under nonhomogeneous traffic condition, examples in which the transmitter cost is either dominant or is comparable with the receiver cost and the capacity per unit cost is better in the node mapping scheme are shown