O1: Essentials of Intelligent Optical Network in 2030

This workshop will discuss on different topics towards intelligent optical networks in 2030. Discussions will include requirements, optical networking technologies, gap analyses, possible research/standardization activities from Access to Cloud, and the application of Artificial Intelligence and Machine Learning (AI/ML). We look forward to hearing opinions from carriers, vendors and R&D institutes (including Standards bodies) during the workshop.

Subject/Issues to be covered:

High speed technologies:
Including 400G transmission systems as well as 100G-PON (and beyond) for x-haul, enterprise or residential access optical networks.

Reconfigurable networks:
Including all optical networks topology and/or architecture, SDN, synchronization and latency control, and distributed intelligence for real-time control. Additionally, it will include a view on the progress of the International Standards Group on network architectures evolution trend and standards

Open Access (Interconnect):
It includes approaches/views from ORAN, Open ROADM, Virtual/Cloud Network and vOLT.

AI/ML technologies and applications:
Aspects of intelligent networks include AI for network operations, smart ubiquitous network, digital twin, and network slicing.
Specific questions for the discussion on AI/ML include:

  • - What is the current status of scientific research of AI/ML-based approaches for optical network management and control?
  • - Where, what and how can we truly apply AI/ML methods in real optical networks?
  • - Can AI/ML -based solutions be applied for real-time optical communications/networks?
  • - What are the possible challenges?


Ching-Sheu Wang
Chunghwa Telecom
Albert Rafel
BT Applied Research
Luis Velasco
Universitat Politecnica de Catalunya
Sugang Xu


Hideki Nishizawa
Jochen Maes
Nokia Bell Labs
Jhih-Heng Yan
Chunghwa Telecom
Xiaoliang Chen
Sun Yat-Sen University
Frank Effenberger
Cheops Shen
China Telecom
Ming-Fang Huang
NEC Labs America

O3+O2+O4+O5: What Killer Technology will Provide Sustainable Growth in Optical Fiber Communications?

We will discuss the direction of sustainable capacity expansion medium (O3), active/passive device (O4 or O5), and signal processing/transmission technology (O2).


Toshihiko Hirooka
Tohoku University
Kunimasa Saitoh
Hokkaido University
Vittorio Curri
Politecnico di Torino
Haoshuo Chen
Nokia Bell Labs



O4+O5: Open-access PIC Providers, Fabless Companies, and Their Opportunities

Like a CMOS technology, photonic integrated circuits are becoming a generic technology not only for research purpose but also for commercial use. The ecosystem needs to be further developed by the players; providers, fabless companies and users.
This workshop brings a discussion and exchange opinions about the following issues.

  • - Silicon and related materials providers
  • - III-V and hybrid photonic integrated systems
  • - Fabless companies
  • - Gap between fabs technologies and fabless companies demands
  • - Valid business and technology model
  • - Trade-off of customization with higher investment costs vs. standardized processes with lower investment costs
  • - Toward development of PIC R&D and product


Martijn Heck
Eindhoven University of Technology
Tae Joon Seok
nEYE Systems
Guillermo Carpintero
Koji Yamada


David Harame
Arne Leinse
Twan Korthorst
Peter O'Brien
Tyndall National Institute
Kazuhiko Kurata
John Fini
Ayar Labs

P1+O4: 800G and Beyond in Intra and Inter Datacenters

Transmission capacity required by hyper-scale datacenters continues growing exponentially. Regardless of the technique of the transmission, including IM-DD or coherent, the reach inherently shortens as the capacity increases. While IM-DD technologies face limitations caused by the accumulated dispersion penalty, by low-pass characteristics of the components as well as Tx launch power and Rx sensitivity limitations, for coherent technologies the main challenges are to reduce cost and power consumption to be comparable with IM-DD. Key concerns are whether IM-DD can be used also for extended reach (10 km+) systems and if it is also a suitable solution for the >200 Gbit/s/lambda era. Future developments still need to prove that coherent can fill that gap with a reasonable cost.
This workshop aims at giving an overview of the latest expectations and requirements for 800G and beyond optical networks used for intra-/inter-datacenters from the perspectives of datacenter operators, standardization, and system/components vendors.
The topics include:

  • - When and where do we need 800G optics?
  • - Needs for hyperscale data center operators
  • - Benefits of higher rates per lane vs. wavelength multiplexing vs. spatial multiplexing
  • - DSP challenges, electrical/optical component requirements
  • - Direct-detection vs. coherent (or hybrid?): when will price levels of coherent match DD?


Stephan Pachnicke
Kiel University
Masahiro Nada
Son Thai Le
Nokia Bell Labs
Rang-Chen (Ryan) Yu


Session 1
Brad Booth
Dirk van den Borne
Nebosja Stojanovic
Pandelis Diamantopoulos
David Plant
McGill University
Session 2
Hong Liu
Chongjin Xie
Hiroshi Onaka
Katsumi Uesaka
Sumitomo Electric Industry
Radha Nagarajan

P2+O5: Photonic Hardware for Machine Learning and Computing

Recent developments in nanophotonics have revived the possibility that optical computers might outperform traditional electronic circuits, particularly in machine learning (ML) and optimization, where computation is bottlenecked by large amounts of data movement. Optics offers unique capabilities that are lacking in existing processors, including extreme bandwidth and connectivity, but weak nonlinearities and lack of large-scale integration have historically limited optical computing to the laboratory. Even if these issues are solved, it remains an open question whether the advantages in optics can overcome the orders-of-magnitude advantage that transistors hold in component density. Finally, optoelectronic integration, codesign, and algorithm choice will be critical to fully harnessing the advantages of photonic hardware.
This workshop brings together pioneers in photonic ML and optimization to evaluate the rapidly evolving state of the field, the strengths and weaknesses of photonics for computing, and prospects for future research and commercial development.


Ryan Hamerly
NTT Research
Yikai Su
Shanghai Jiao Tong University
Yuya Shoji
Tokyo Institute of Technology
Claudio Conti
University Sapienza