Spectrum Allocation Policy Modeling for Elastic Optical Networks

Today, optical transmission technologies are able to support 400Gbps over a single optical channel. However, this capacity cannot fit in the current fixed frequency grid optical spectrum. On the other hand, high rate optical channels have to co-exist with different ranges of line rates in order to serve heterogeneous bandwidth requests from variety of internet applications. Today's fixed rate and rigid frequency grid optical transmission systems cause over provisioning, where usually more spectral resources are provided than necessary. Recently, the concept of elastic optical networks has been proposed in order to reduce this waste of resources. In networks with such feature enabled, modulation parameters and central frequencies are not fixed and the resources can be allocated with a fine granularity, in contrast to the traditional WDM networks. This flexibility makes it possible to adapt to the granularity of the requested bandwidth without over provisioning. However, this heterogeneous bandwidth allocation may on the other hand result in fragmentation of spectral resources under dynamic traffic. In this study we quantify the fragmentation in elastic optical networks and calculate the blocking probability (BP) together with fragmentation on an elastic optical channel. In this regard, an analytical model based on a Markov Chain is developed under dynamic and flexible bandwidth traffic scenario. By using this analytical model different spectrum allocation policies are compared. A novel spectrum allocation policy is proposed which has lower BP and fragmentation ratio compared to the existing strategies.