Modeling Demands Forecasts with Probability Distributions in DWDM Optical Networks

Optical Dense Wavelength Division Multiplexing (DWDM) networks undergo a constant evolution in response to ever increasing traffic. Therefore every DWDM network operator has to upgrade facilities in response to the ever growing traffic demand. For this purpose software design tools are needed that will allow for predicting the traffic demands and also optimise the required DWDM resources. This contribution presents a software design suite that performs statistical analysis of the data traffic in a DWDM network edges and calculates estimates of the future values for the demand matrix elements with parametric/nonparametric probability distribution. Once the demand matrix elements are known an optimisation algorithm is used to predict the required DWDM equipment needed to meet the traffic demand. As an example a 3 node section of a representative DWDM network with 12 nodes was considered. Demand matrix is constructed by collating together traffic demands between each pair of nodes expressed in Gb/s.