K. Lukac (Switzerland), Z. Lukac, and M. Tkalic (Croatia)
WDM, RWA, wavelength conversion, learning automata
This study focuses on the routing and wavelength assignment (RWA) problem in wavelength-routed optical WDM networks. We investigate the effectiveness of adaptive algorithms based on the learning automata concept applied in relatively large networks as opposed to traditional fixed alternate routing. We also study the benefits of full wavelength conversion on the resulting connection blocking of the proposed routing algorithm as compared to routing without wavelength conversion. One of the main advantages of the proposed algorithm is that full knowledge of the network state is not necessary. By simulating a mesh network we report performance results under different traffic conditions. The study shows that both approaches have similar behaviour in case of a uniform capacity distribution with or without the wavelength continuity constraint. However, if link failures appear i.e. in case of a non-uniform capacity distribution which corresponds to practical network realisations the proposed algorithm outperforms fixed alternate routing. It is especially superior if in the previous case wavelength conversion is enabled thus resulting with large blocking reductions.
Important Links:
Go Back
