SUPSI Research on Explainable AI for Optical Networks Wins the Sir Charles Kao Award

• August 5th, 2026

Researchers Omran Ayoub, Davide Andreoletti and Prof. Silvia Giordano from the Institute of Information Systems and Networking (ISIN) have received the 2026 IEEE Communications Society Charles Kao Award for the Best Optical Communication and Networking Paper. The award-winning study demonstrates how Explainable Artificial Intelligence (XAI) can make the use of machine learning models in the automated management of optical networks more transparent and reliable.

Named after Sir Charles Kao, the 2009 Nobel Prize in Physics laureate and pioneer of fibre-optic communications, the award is presented annually to the best paper published in the previous three years in the IEEE/Optica Journal of Optical Communications and Networking (JOCN). Papers are evaluated based on scientific quality, originality, impact, clarity of presentation, and their potential to open new research directions.

The 2026 edition recognised the paper Towards Explainable Artificial Intelligence in Optical Networks: The Use Case of Lightpath QoT Estimation, published in January 2023 by Omran Ayoub, Davide Andreoletti and Silvia Giordano (SUPSI), Sebastian Troia and Massimo Tornatore (Politecnico di Milano), and Andrea Bianco and Cristina Rottondi (Politecnico di Torino).

"This is one of the first contributions worldwide to introduce and apply Explainable Artificial Intelligence (XAI) techniques to optical networks, opening a new research direction towards more reliable and transparent network management systems," explains Dr. Omran Ayoub, Co-Head of the Reliable and Secure Computer Networks research area.

Machine learning models are often described as "black boxes", as the reasoning behind their predictions remains difficult to interpret despite their high level of accuracy. This lack of transparency is particularly critical in the management of essential infrastructures such as the optical networks that support modern telecommunications.

The award-winning study focuses on machine learning models used to estimate Quality of Transmission (QoT), a key parameter for predicting signal quality in optical links. Using XAI techniques, the researchers demonstrated how the models' decisions can be interpreted and validated against the underlying physical behaviour of the network, improving trust in their predictions and enabling potential anomalies to be detected before operational use.

"This recognition goes beyond a single scientific paper. It acknowledges a research direction that continues to drive our work: making artificial intelligence and machine learning more trustworthy, interpretable and suitable for real-world deployment in communication networks. As networks become increasingly autonomous, explainability is no longer just a desirable feature, but a fundamental requirement," says Dr. Omran Ayoub.

Paper:
O. Ayoub, S. Troia, D. Andreoletti, A. Bianco, M. Tornatore, S. Giordano, C. Rottondi, Towards Explainable Artificial Intelligence in Optical Networks: The Use Case of Lightpath QoT Estimation, Journal of Optical Communications and Networking, vol. 15, n. 1, pp. A26-A38, gennaio 2023.