Approaching the essence of the IOWN APN through access-side APN architecture, SDN, and drastic improvements in transmission performance
—Could you tell us about the research you are currently working on?
I’m continuing to work on the theme of research on new optical access networks that will accelerate the evolution of information and communication services.
In conventional networks, all traffic from the access network, to which user terminals are connected, are transferred to the core network through electrical processing in the telecommunications building closest to the users, so it consumes much power and causes delays. Communication equipment is dedicated equipment that varies depending on the service. In future networks, clear separation of optical and electrical processing will enable user terminals to access any necessary locations without electrical processing, which is executed only where needed. This will achieve drastic power savings and low latency. Electrical processing will be executed on general-purpose servers as much as possible (Fig. 1).
Fig. 1. Overview of research on new optical access networks.
To achieve the future All-Photonics Network (APN) shown in the lower part of Fig. 1, my research is progressing along three axes: research on the access-side architecture of APN, improving flexibility by softwarizing transmission functions, and drastically improving transmission performance of optical access networks.
Regarding the first axis of my research (research on the access-side architecture of APN), I explained in my previous interview (February 2023 issue) how we prototyped and demonstrated the Photonic Gateway (Ph-GW), namely, the gateway to the APN (see Fig. 1), by assembling components. My team is currently speeding up our research on Ph-GW for the future APN, which will cover the access side, and we are making various proposals and demonstrations including those for achieving plug-and-play functionality for APN terminals for supporting optical signals for various services.
The above achievements resulted in the reception of the Best Paper Award at the International Conference on Emerging Technologies for Communications (ICETC) 2022, and consistent acceptance of our papers at prestigious conferences in the optical communications industry (Optical Fiber Communication Conference and Exposition (OFC) 2023, European Conference on Optical Communication (ECOC) 2023, OFC 2024, and OFC 2025).
The IOWN Global Forum (IOWN GF) is continuing discussions on the overall APN architecture, which covers both the access and core sides of the APN. In the Open APN Architecture Task Force (OAA-TF), of which I serve as a co-taskforce leader, we have expanded the Open APN Functional Architecture to enable network operators to deploy Open APNs more easily and effectively. We have formulated and published architecture documents Version 2 (released October 2023) and Version 3 (released June 2025) in a manner that embodies the gradual evolution of the APN (Fig. 2).
Fig. 2. Key points of revisions of Open APN Functional Architecture.
Regarding the second axis of my research, “improving flexibility by softwarizing transmission functions,” in conventional communication equipment, functions required for each service are implemented using vendor-specific designs, so it is difficult to flexibly add or change functions in response to changing requirements. With the recent advancement in network-virtualization technology, some communication functions can now be implemented with software, so it is possible to flexibly add and change functions. We are expanding the functional domains achievable through software as we aim for a world in which all communication functions and services are provided via software on general-purpose server equipment used in datacenters (Fig. 3). It will thereby become possible to provide various services rapidly by simply replacing the optical module. To address the challenge of reducing latency, we have implemented physical-layer functions for a 10-Gbit/s passive optical network (PON) system, a typical optical access system, in software on a general-purpose server with a processing time of less than 1 ms without using dedicated large-scale integrated circuits (LSIs). We have also achieved signal processing for digital coherent systems, which are essential for long-distance transmission, in software on general-purpose servers with a transmission speed of 10 Gbit/s without using dedicated LSIs.
Fig. 3. Full softwarization of communication functions.
After achieving the above results, my team expanded its efforts to implement software-defined networking (SDN) on the basis of the softwarization of the transmission functions, in addition to pursuing transmission performance. Specifically, various processing functions, from signal-level processing (such as error correction) and communication-control processing (such as priority control) to applications (such as image analysis) are “containerized” (i.e., packaged as a combination of software and the environment required to run it), and “function chaining” is achieved by combining and executing the functions in sequence. Reports on these technologies were selected for presentation at OFC 2024 and publication in Journal of Optical Communications and Networking published by IEEE/Optica. To demonstrate scalability, we demonstrated simultaneous processing of data equivalent to traffic from over 40,000 user devices in a supercomputer environment, and our report on this demonstration was selected for presentation at ECOC 2025. Our aim is to implement these initiatives concerning SDN in actual networks around 2030.
—“Drastically improving transmission performance of optical access networks” is a theme that gets to the heart of implementing the access side of APN, right?
Regarding drastically improving transmission performance of optical access networks, conventional optical access networks are deployed within areas where the distance from each user to the nearest telecommunications building is 20 km or less, and PON systems in which bandwidth from 1 to 10 Gbit/s is shared among 32 to 64 users are used. Regarding the future APN, we are attempting to dramatically improve transmission performance by using digital-coherent access technology as a basic technology to enable optical access to wherever necessary. This technology enables larger capacity transmission through the use of phase of light in addition to intensity to transmit information and makes it possible to receive signals with ultra-high sensitivity through the use of coherent detection on the receiver side. Using digital-coherent access technology has the potential to achieve significant improvements in performance, such as achieving transmission speeds of over 100 Gbit/s, accommodating over 256 users, and extending access distance to over 80 km (Fig. 4), so we are focusing on achieving these goals. In our previous real-time verification experiments, we demonstrated a transmission distance of 40 km (twice the previous verified distance) and transmission and reception budget of 50 dB (100 times the previous verified value) in a PON system with 10-Gbit/s upstream and downstream.
Fig. 4. Drastic improvement in transmission performance through digital-coherent access technology.
By continuing to study ways to further increase transmission speed, my team proposed and demonstrated a simpler reception method for 100-Gbit/s access, which was presented at ECOC 2024. While this demonstration is still at the laboratory level, especially if we consider the application of digital-coherent access technology to the APN, the key point is not only increasing speed but also how to achieve it with an economical architecture. Since optical network units (ONUs) are installed individually for each user, a sharing scheme is not applicable to reduce their cost; therefore, a major challenge is how to achieve low-cost ONUs. Various methods for meeting this challenge have been proposed, and in face of that competition, my team is also currently making various proposals.
In accordance with Amara’s Law, long-term predictions often turn out to be wrong, but it is important to enjoy making predictions about the future
—What do you keep in mind as a researcher?
Being conscious of this for a long time, I believe it is important to think about major long-term changes—without being overly swayed by short-term trends—and to have fun predicting what the future holds. The underlying theme of this consciousness, as I have mentioned in the previous interview, is Amara’s Law, which states that we tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run. Short-term topics are easy to predict and understand, so we tend to have excessive expectations. On the contrary, long-term topics tend to be more volatile and often turn out to be wrong, so people tend to be more reserved. With those outcomes in mind, I try to make predictions while appreciating the fact that they may turn out wrong.
In the field of access networks, in which I worked, around the year 2000, it was the norm to access the Internet via a telephone line. At that time, technology called asymmetric digital subscriber line (ADSL) made it possible to significantly increase communication speed, so some people thought it was uncertain whether fiber to the home (FTTH) would become widespread; however, by 2016, FTTH had spread to more than twice as many homes as ADSL did at its peak. Not only has the Internet become faster but also all data are now processed on the cloud. Thus, the relationship between users like ourselves and the network has changed dramatically. In the 2010s, mobile communications became faster and smartphones became common, and most of the tasks that were previously only possible on personal computers (PCs) can now be done on smartphones. Since 2020, generative artificial intelligence (AI) has been advancing rapidly. When I think about what generative AI will do for us in 10 or 20 years, I think many aspects of our life will be beyond our imagination. In terms of networks, advances in generative AI are already increasing the need for faster, lower-latency connections between datacenters in various locations, and that need could lead to major changes in the nature of optical networks. I want to continue to imagine such major changes in a positive light as I continue my research.
I have been involved in international standardization activities as an associate rapporteur for International Telecommunication Union Telecommunication Standardization Sector (ITU-T) Study Group 15 Question 2 (dealing with issues related to optical systems for fiber access networks), chairman of Full Service Access Network (FSAN), and co-taskforce leader of IOWN GF OAA-TF. I believe that like two wheels of a cart, standardization and research/development should work in harmony, and I have been conscious of the balance between them since I was a young researcher. Pursuing both makes my work twice as fun. Standardization activities of ITU-T and FSAN have been handed over to the next generation, but I continue to hold discussions with them and make proposals. In particular, standardization and globalization will become increasingly important in the field of networks, so I am consciously trying to stay involved in those areas.
Taking on new challenges and doing things that interest you, even if they are not directly related to your research topic, can lead to new discoveries and research ideas
—What is your message to future researchers?
I have previously mentioned a famous speech by Steve Jobs, titled “Connecting the Dots,” which I still recall vividly. Steve Jobs’ interest in calligraphy, which he discovered by chance in a university library, led to the diverse range of fonts available on the Mac PC. The speech was about how we don’t know how certain dots will connect in the future. I took it as an indication that sometimes things that seem disconnected and unrelated at first may end up connecting by chance later on. As research progresses, when we make new realizations or discoveries, we tend to judge at that stage whether they will lead to the results we are seeking. However, even if things seem unrelated at the time, they may become connected in the future. To increase the chance of such connections and make the most of them, I encourage everyone to take on new challenges about things that interest you, even if they are outside of your own field of expertise.
I also hope you will have fun making long-term predictions while keeping Amara’s Law in mind. Such predictions are not limited to research; in fact, when thinking about the long term, you will be predicting what the world will be like 10 or 20 years from now. I think it will be even more fun to make predictions if you also think about what kind of work you will be doing 10 or 20 years from now, when you’ll presumably be 40 or 50 years old, for example. By making predictions from these different perspectives, you may make new discoveries, and those discoveries may lead to future connections, as Steve Jobs suggested in “Connecting the Dots.”
