Performance Analysis of a Data-Driven Quality-of-Transmission Decision Approach on a Dynamic Multicast-Capable Metro Optical Network

Abstract

The performance of a data-driven quality-of-transmission (QoT) model is investigated on a dynamic metro optical network capable of supporting both unicast and multicast connections. The data-driven QoT technique analyzes data of previous connection requests and, through a training procedure that is performed on a neural network, returns a data-driven QoT model that near-accurately decides the QoT of the newly arriving requests. The advantages of the data-driven QoT approach over the existing Q-factor techniques are that it is self-adaptive, it is a function of data that are independent from the physical layer impairments (PLIs) eliminating the requirement of specific measurement equipment, and it does not assume the existence of a system with extensive processing and storage capabilities. Further, it is fast in processing new data and fast in finding a near-accurate QoT model provided that such a model exists. On the contrary, existing Q-factor models lack self-adaptiveness; they are a function of the PLIs, and their evaluation requires time-consuming simulations, lab experiments, specific measurement equipment, and considerable human effort. It is shown that the data-driven QoT model exhibits a high accuracy (close to 92%–95%) in determining, during the provisioning phase, whether a connection to be established has a sufficient (or insufficient) QoT, when compared with the QoT decisions performed by the Q-factor model. It is also shown that, when sufficient wavelength capacity is available in the network, the network performance is not significantly affected when the data-driven QoT model is used for the dynamic system instead of the Q-factor model, which is an indicator that the proposed approach can efficiently replace the existing Q-factor model.