Dynamic Versus Static Wavelength-Routed Optical Networks

TLDR: New analytical and algorithmic lower bounds for the wavelength requirements of dynamic networks are proposed to evaluate the optimality of the already proposed algorithms whose wavelength requirements are quantified - by means of simulation- with and without the wavelength continuity constraint.

Abstract: It is widely believed that dynamic operation of wavelength-routed optical networks will help overcome the inefficiencies of static allocation in terms of wavelength utilization. In this paper, this hypothesis is reviewed. The wavelength requirements of dynamic wavelength-routed optical networks are quantified and compared to static routing. To do so, new analytical and algorithmic lower bounds for the wavelength requirements of dynamic networks are proposed. They are used to evaluate the optimality of the already proposed algorithms whose wavelength requirements are quantified - by means of simulation- with and without the wavelength continuity constraint, as well as for uniform and nonuniform traffic demand. Results show that, without wavelength conversion capability, the benefit of dynamic wavelength-routed optical networks is not significant: lower wavelengths requirements than static networks are achieved only at low traffic loads (< 0.3 - 0.4). In wavelength convertible networks instead, dynamic operation leads to lower wavelength requirements than static operation over a wide traffic load range (< 0.7 - 0.8), making dynamic operation attractive in this type of network. Under nonuniform traffic it was found that dynamic operation achieves slightly higher wavelength savings than in the uniform traffic case. But the savings were not high enough as to impact on results of the uniform traffic case. These results can aid network operators in the design and optimization of optical WDM networks.