Network nodes dimensioning assuming electrical traffic grooming in an hybrid OXE/EXC WDM network

In this paper, one considers the routing and wavelength allocation problem (RWA) under deterministic dynamic traffic demands. Optical virtual private networks (OVPN) are typical examples of such demands. In previous papers, we have introduced the concept of scheduled lightpath demands (SLD) defined as a tuple (s, d, alpha, beta, omega). The parameters s and d stand for the source and destination addresses, alpha and beta are the setup and teardown dates of the demand, and omega is the required capacity expressed in number of optical channels. In this paper, we extend this concept to the case where omega is not an integer but a real value. In this context we introduce the concept of scheduled electrical demand (SED). For instance, a 3 Gbps demand, assuming 2.5 Gbps optical channels, corresponds to one SLD and one SED with 0.5 Gbps capacity. We propose a routing and grooming strategy based on a simulated annealing (SA) global optimization technique. Our objective is to minimize the number of optical and electrical ports required to satisfy a set of SLDs and SEDs on a given infrastructure. An original and realistic traffic model already introduced in previous papers is used for our simulations. We investigate for multiple traffic scenarios the geographical distribution of the size of the OXC/EXC nodes