April 30,2025, The Rump Session on the evening of the first day of the OFC exhibition every year is definitely the most anticipated part. I really admire that Americans can put aside their anxiety about the market and relax to discuss some interesting and challenging topics. This year's topic is quite innovative. Suppose the existing communication network architecture is completely destroyed and we have to start from scratch. How should we build a brand-new optical communication network?
One of the biggest advantages of Americans compared with Chinese is definitely their ability to have fun. In this session, the host divided the discussion guests into three teams, and they debated from aspects such as the infrastructure, underlying technologies, and applications to see whose future network solution is better. With the accompaniment of beer, it is a very interesting mental game. Unfortunately, the English spoken by Americans in such occasions is usually difficult for me to understand. I can't remember exactly how they debated, and I can only share with you some parts that impressed me deeply.
First of all, if we were to answer this question ourselves, how should we respond? Or rather, where does our greatest dissatisfaction with the current optical network? In my opinion, optical connectors are not as simple as electrical connectors. Optical devices always need to be coupled and packaged, and there are also issues such as chromatic dispersion, nonlinearity, and polarization at the optical level. There are too many factors that affect the transmission quality. Over the past 50 years, optical communication technology has gone through generations of evolution. If we could redesign the optical network, how would we choose these technologies?
Let's talk about optical fibers first. Except for the emergency balloon wireless communication solution, everyone believes that hollow-core optical fibers are definitely necessary. Although at this year's OFC, hollow-core optical fibers did not cause as much of a sensation as they did last year, presumably the current focus is more on promoting their applications. At least in this session, hollow-core optical fibers have received unprecedented recognition. Without chromatic dispersion and nonlinearity, many of the current problems in fiber optic communication are eliminated. So, when deploying optical fibers from scratch, without considering other commercial factors, this is a unanimous solution.
Looking at the deployment of optical fibers, it seems that the unanimous opinion is also to deploy as many optical fibers as possible. Use a cable with 1000 optical fibers and directly design the network capacity to be 100 times that of the current one, and plan for a 10-fold speed increase. I can't remember which team among them also proposed a cable deployment method combined with nuclear power, really pushing the capabilities of human technology to the limit.
How to solve the difficult problem of fiber optic coupling? One team proposed a THz solution. “The best optical network may not be 'optical'”. I still know very little about THz technology, but they said that as the carrier frequency increases, photon technology will become more and more expensive. By reducing the carrier frequency and widely applying multiplexing technology, optical communication with the help of THz technology can greatly reduce the problem of alignment and can reduce the energy consumption per bit. If this is really the case, it will be a whole new field.
Silicon photonics is the foundation. A large number of CPOs based on optical IOs are applied. Parallelization is carried out from various modes such as WDM, polarization, and spatial division. Also, standardize the DSP in advance and introduce IP technology throughout the network. This solution seems to make the most of the existing technologies.
How does the new optical network view the competition between IM-DD and coherent technologies at this OFC? Some people say that the one-step solution is to use IM-DD only within the data center, and leave everything else to coherent technologies. There are four little heroes, PAM4, exp(x), Aquadrature (orthogonal), and Fourier transform, which will be the superheroes of the new era.
A more sci-fi-themed solution is to build all future data centers underwater not far from the coastline. In this way, the problem of heat dissipation is greatly reduced, and there are already quite a few people living along the coastline, which can also improve the impact of latency.
Perhaps it can be said that this Rump Session is a showdown among the Nerds and Geeks of optical communication technology, because many issues such as the market and policies that affect technological development are no longer a problem. But isn't returning to the essence of technology exactly what makes OFC so attractive to us?
Appendix: Who are in the three teams?
Team A
Chris Doerr, Aloe Semiconductors, USA
Tad Hofmeister, Google, USA
Maxim Kushnerov, Huawei, Germany
Rebecca Schaevitz, Mixx Technologies, USA
Team B
Polina Bayvel, Univ. College London, UK
Vijay Vusirikala, Arista Networks, USA
Dave Welch, Attotude, USA
Peter Winzer, Nubis Communications, USA
Team C
Chris Cole, Coherent Corp., USA
Jorg-Peter Elbers, Adtran, Germany
Katherine Schmidtke, Eribel Systems, USA
Glenn Wellbrock, Verizon, USA
