Optical networks generate a vast amount of diagnostic, control, and performance monitoring data. When information is extracted from these data, reconfigurable network elements and reconfigurable transceivers allow the network to adapt not only to changes in the physical infrastructure but also to changing traffic conditions. Machine learning is emerging as a disruptive technology for extracting useful information from these raw data to enable enhanced planning, monitoring, and dynamic control. We provide a survey of the recent literature and highlight numerous promising avenues for machine learning applied to optical networks, including explainable machine learning, digital twins, and approaches in which we embed our knowledge into machine learning such as physics-informed machine learning for the physical layer and graph-based machine learning for the networking layer.

We propose a novel application of reinforcement learning (RL) with invalid action masking and a novel training methodology for routing and wavelength assignment (RWA) in fixed-grid optical networks and demonstrate the generalizability of the learned policy to a realistic traffic matrix unseen during training. Through the introduction of invalid action masking and a new training method, the applicability of RL to RWA in fixed-grid networks is extended from considering connection requests between nodes to servicing demands of a given bit rate, such that lightpaths can be used to service multiple demands subject to capacity constraints. We outline the additional challenges involved for this RWA problem, for which we found that standard RL had low performance compared to that of baseline heuristics, in comparison with the connection requests RWA problem considered in the literature. Thus, we propose invalid action masking and a novel training method to improve the efficacy of the RL agent. With invalid action masking, domain knowledge is embedded in the RL model to constrain the action space of the RL agent to lightpaths that can support the current request, reducing the size of the action space and thus increasing the efficacy of the agent. In the proposed training method, the RL model is trained on a simplified version of the problem and evaluated on the target RWA problem, increasing the efficacy of the agent compared with training directly on the target problem. RL with invalid action masking and this training method outperforms standard RL and three state-of-the-art heuristics, namely, k shortest path first fit, first-fit k shortest path, and k shortest path most utilized, consistently across uniform and nonuniform traffic in terms of the number of accepted transmission requests for two real-world core topologies, NSFNET and COST–239. The RWA runtime of the proposed RL model is comparable to that of these heuristic approaches, demonstrating the potential for real-world applicability. Moreover, we show that the RL agent trained on uniform traffic is able to generalize well to a realistic nonuniform traffic distribution not seen during training, thus outperforming the heuristics for this traffic. Visualization of the learned RWA policy reveals an RWA strategy that differs significantly from those of the heuristic baselines in terms of the distribution of services across channels and the distribution across links.

Maximized information rates of ultra-wideband (typically, beyond 100~nm modulated bandwidth) lumped-amplified fiber-optic communication systems have been thoroughly examined accounting for the wavelength dependencies of optical fiber parameters in conjunction with the impact of the inelastic inter-channel stimulated Raman scattering (SRS). Three strategies to maximize point-to-point link throughput were proposed: optimizations of non-uniformly and uniformly distributed launch power per channel and the optimization based on adjusting to the target 3 dB ratio between the power of linear amplified spontaneous emission and nonlinear interference noise. The results clearly emphasize the possibility to approach nearly optimal system performance by means of implementing pragmatic engineering sub-optimal optimization strategies.

Electrified railways are becoming a popular transport medium and these consume a large amount of electrical energy. Environmental concerns demand reduction in energy use and peak power demand of railway systems. Furthermore, high transmission losses in DC railway systems make local storage of energy an increasingly attractive option. An optimisation framework based on genetic algorithms is developed to optimise a DC electric rail network in terms of a comprehensive set of decision variables including storage size, charge/discharge power limits, timetable and train driving style/trajectory to maximise benefits of energy storage in reducing railway peak power and energy consumption. Experimental results for the considered real-world networks show a reduction of energy consumption in the range 15%–30% depending on the train driving style, and reduced power peaks.

No abstract