Protection against failure of machine-learning-based QoT prediction

Abstract

Machine learning (ML)-based methods are being widely explored to predict the quality of transmission (QoT) of a lightpath. They are expected to reduce the signal-to-noise ratio margin reserved for the lightpath, thus improving the spectrum efficiency of an optical network. However, many studies on this prediction are often based on synthetic datasets or datasets obtained from laboratories. As such, these datasets may not accurately represent the entire state space of a practical optical network, which is exposed in harsh environments. There are risks of failure when using these ML-based QoT prediction models. Thus, it is necessary to develop a mechanism that can guarantee the reliability of lightpath service even if the prediction model fails. For this scenario, we propose to employ the conventional network protection techniques that are usually implemented to protect against network node/link failures to also protect against the failure of QoT prediction. Based on the two representative protection techniques, i.e.,  1 + 1 dedicated path protection and shared backup path protection (SBPP), the performance of the proposed protection mechanism is evaluated by reserving different margins for the working and protection lightpaths. For 1 + 1 path protection, we find that the proposed mechanism can achieve a zero design margin (D-margin) for a working lightpath, thereby significantly improving network spectrum efficiency, while not sacrificing the availability of lightpath services. For SBPP, we find that an optimal D-margin should be identified to balance the spectrum efficiency and service availability; the proposed mechanism can save up to 0.5-dB D-margin for a working lightpath, while guaranteeing service availability.