Wavelength translation in WDM networks: Optimization models and solution procedures

Wavelength-division multiplexing (WDM) is the best technology currently available to handle the rapidly increasing demand for bandwidth in fiber-optical telecommunications networks. In a WDM wavelength-routed network, end users communicate with one another using all-optical channels called lightpaths. Such lightpaths are used to support point-to-point connections and may span multiple fiber links. In the absence of wavelength translators, a lightpath must use the same wavelengths from origin to destination along each fiber span. Translators located at nodes permit wavelength translation within a lightpath. This investigation presents an empirical study comparing solutions that forbid translation with those that permit translation. For our 20 test problems, the extreme cases of no translation and translation at every node are solvable using CPLEX. For the more difficult intermediate cases, a special tabu search heuristic was developed. For those difficult cases, the tabu search procedure ran approximately 25 times faster than CPLEX using the default settings and in 78% of the problem instances found solutions that were as good as or better than those found by CPLEX.