Realistic dynamic traffic generation for WDM Optical Networks

This Paper presents a realistic approach on tackling the dynamic traffic generation problem for Wavelength Division Multiplexing Optical Networks. Given the relatively novel aspect of these networks, most research in this area was mainly focusing on the problems of routing and wavelength assignment while using fairly simplistic memoryless models for traffic generation in the simulated networks. The Internet of the very early days is way simpler than today's Internet. Although that itself is controversial, what is at least quite evident in today's Internet is its repeating-trends, long term memory process. This alone makes the old Markov Chain, Poisson, and the rest of memoryless models that we see still being used in simulations, actually stale and out of context. This fact makes any results based on such simulations highly questionable in terms of how close they are to reality and thus how practical to use those results in any applications at all. This paper is to address this controversial topic and present the Self-similar Pareto Process as a realistic convenient traffic generation model to simulate today's Internet traffic on its Wavelength Division Multiplexing Optical Networks Backbone.