The WDM architectures that have been proposed, or are being implemented, arebased on transmitters and receivers that can be precisely tuned to predetermined, fixed wavelengths. Maintenance of such fixed wavelength channels when the transmitters and receivers are separated geographically under diverse ambient conditions is not a simple task. We are working on the implementation and evaluation of a novel WDM network that does not rely on fixed wavelength channels, and as such is superior in terms of robustness and cost effectiveness compared to existing WDM architectures. The lasers, amplifiers and detectors required are either commercially available, or will be fabricated as a part of this project. A comprehensive performance model for the network is also being developed. The features of the MAC protocol give new architectural flexibilityto WDM systems, in particular enabling the design of systems with greatly relaxed manufacturing and operating wavelength tolerances. This project is done in collaboration with Prof. H. Temkin of Colorado State University and Prof. Jon Sauer of the University of Colorado.