A survey of AI-driven QoT analysis in lightpath management

Quality of Transmission (QoT) analysis is a foundational component in planning and operational management of modern optical networks. Traditional analytical models have limitations in accurately modeling complex physical-layer impairments in high-capacity systems. In contrast, recent advancements in Artificial Intelligence (AI) have enabled significant progress in QoT estimation. This survey systematically reviews AI-driven QoT analysis methodologies for Lightpath (LP) management. It examines key application scenarios, including unestablished LP feasibility prediction, intelligent spectrum allocation, and adaptive margin optimization. The existing approaches are categorized based on methodological frameworks and operational objectives. The review shows a predominant reliance on end-to-end path feature extraction, with occasional incorporation of per-channel metrics. A critical observation from this review is the prevalent neglect of inter-LP correlation effects in existing studies. This is notable despite their substantial impact on network-wide QoT dynamics. Finally, we identify unresolved challenges in current QoT analysis paradigms. We also propose future research directions to bridge gaps between data-driven predictions and physical-layer constraints in next-generation optical networks.