Ken Sakamura has been talking pervasive computing since before the term was invented. The University of Tokyo professor, easily the most famous computer architect in Japan, first proposed the idea of networked, ubiquitous computing back in 1984, when he devised the open architecture known as The Real-time Operating System Nucleus (TRON). Since then, Sakamura has led the TRON project and several spin-offs in developing core technologies for an environment in which every object incorporates a computer and is linked to a network. The T-Engine Forum, for example, was established in 2002 to promote embedded development around TRON. A year later came the Ubiquitous ID Center, which aims to promulgate RFID technology around Sakamura's "ucode." Designed as the infrastructure for ubiquitous computing, ucode can tag anything from a can of peas on a supermarket shelf to the Empire State Building with its own unique code.
Sakamura recently showed EE Timeshis latest inventionthe Micro Ubiquitous Communicator, a personal identification device the size of a matchbox that works as a door key, electronic money and PC authentication.
EE Times: What led you to develop TRON in the 1980s?
Ken Sakamura: At that time, the United States was advanced in information systems and Japan's forte was consumer electronics. So I was interested in embedded computing for consumer products. I was quite sure that in the future, software would define all the functions of CE products. I thought it was mandatory to prepare an OS that would enable the development of high-performance software with high efficiency.
Since the OS would be the infrastructure for embedded systems, it should be open so that people could share it. It is not good for something that works as an infrastructure to belong to someone. Based on this conviction, I've been working on building an open platform for embedded systems for about 20 years.
As a research assistant at the graduate school of the University of Tokyo at that time, I wanted to work for the benefit of many people, not for a specific company. So TRON was not created to make money; its greatest strength is that it's open and free of charge.
The TRON effort now comprises ITRON for embedded systems, JTRON for Java implementations, BTRON for larger systems like PCs and PDAs, and CTRON for communications systems. Is this because of TRON's scalability?
Yes. It is scalable. Like Linux, TRON can be implemented in small and large systems. If an OS is an open architecture and free, its user community expands explosively and the system improves with feedback. That is impressive power.
TRON's base is hard real-time embedded systems such as in mobile phones, car engines and high-speed shutters in digital still cameras.
Has ITRON been widely accepted for embedded systems?
ITRON has become the actual infrastructure of Japan's electronics industry. More than 60 percent of embedded systems that use a microprocessor use the TRON OS. Some of the remaining systems do not have an OS, or they use a proprietary OS that is often based on ITRON. About 80 percent of OSes used in embedded systems in Japan are ITRON. Since ITRON is free, it does not yield a royalty. So I did not become a rich man through TRON.
T-Kernel is the latest version of TRON with enhanced networking capability. The base is the same as ITRON, but it was upgraded to cope with improvements in the networking environment. Eventually, all embedded microprocessors will have network accessibility.
In embedded-system development, co-development of hardware and software is another challenge. Software development for a PC is different because the hardware is fixed and reliable. In an embedded system, the hardware changes during the development process. It becomes difficult to determine on which side, hardware or software, each bug occurs.
To improve the situation, we built the new T-Engine platform, in which standardized hardware boards are available. Using the boards, hardware and software can be co-developed.
Another issue in embedded development is that the embedded microprocessor may change. In terms of cost and performance, the core of an SoC LSI may be replaced with some other core. Even if the core is changed, design engineers want to use the peripheral logic as is. But different microprocessors have different bus systems.
With the cooperation of Altera and Xilinx, both of which are members of the T-Engine Forum, the forum has opened logic on FPGAs that absorbs the bus differences of various MCUs. With this interface logic, even if the core chip is changed, software developed on the T-Engine board can be used with the new core. The interface logic supports most processors now and we are also going to add new processors. Consequently, software developed in the T-Engine development environment becomes core-chip-free after recompilation
Chipmakers are also proposing SoC development platforms. Does T-Engine compete with them?
There can be many choices. It is free for engineers to use or not use TRON. If they find a better development platform, they can use it. I believe that TRON will be the winner in the end. I want to bring the advantages of openness seen in PCs and the Internet to the embedded world. If we can pool the worlds intelligence, imagine what we could do. People without the open idea in the networking era will no longer be able to survive in the embedded-systems world. The new era has come.
What is the mission of the Ubiquitous ID Center and how does it differ from EPCglobal, which is involved in RFID standardization?
We believe there should be large varieties of RFID services, so the Ubiquitous ID Center will authorize various RFID specifications as standard ucode tags, insofar as the RFID format is open. It is not practical to set one standard format and push costs down by merit of scale, which is what EPCglobal seems to be aiming at.
Since this technology is still under development, we should collaborate where we can. If EPCglobal's RFID is made open when it is completed, we are going to authorize it under Ubiquitous ID. However, EPCglobal seems to focus on supply-chain management applications such as the model that Wal-Mart has adopted. Our target is not only supply-chain management, but also safety and security.
We also intend to embed chips in citiesi.e. to embed ucode tags in roads and buildings to establish a location information system. The tags provide location information to users of personal-navigation systems, or help handicapped people with the information in a ubiquitous environment. It seems difficult to realize these applications using only EPCglobal's RFID.
The T-Engine Forum and the Ubiquitous ID Center are proceeding overseas together. Branch offices have already been opened in Beijing and Shanghai in China, and in Korea, Singapore, Thailand and Vietnam. Our activity has focused on Asia, but now it is expanding to Europe. We are planning to expand activities to the United States next year.
Is the Ubiquitous Communicator a good demonstrator for the ucode?
Yes, it shows people what the ucode system can do. This unit charges automatically via a USB connection and consumes little power, so users would not feel that it is battery-operated. Even in such a small configuration, it is equipped with a display and an antenna for communication. It would be difficult to realize such a small unit with embedded Linux. TRON's forte is low-power consumption, resulting in long battery operation. It can realize mobile phones with 1,000hr standby times.
My Ubiquitous Communicator opens the door, but collaboration is the way. Companies such as IBM, Oracle, Microsoft and Sun Microsystems are among the roughly 500 members of the T-Engine Forum. If the advantageous technologies of U.S. companies and our own microtechnologies join hands, we can realize a true pervasive society in which everybody can access information any time, everywhere.
いまだに著者は、「TRONは政府のプロジェクトではなかった」などという建て前論を 持ち出しているが、 通産省の産業政策について研究したScott Callonの"Divided Sun: MITI and the Breakdown of Japanese High-Tech Industrial Policy, 1975-1993" (Stanford U.P.)は、法的な権限のないTRONプロジェクトに通産省がいかに強い影響力を 持ちえたかを分析している。NECでさえ、最後まで公式にはTRONに反対だとはいわず、 TRON協議会に入ったばかりか、CECにTRON仕様のパソコン(もちろん松下のOEM)を納入 までしているのだ。撤退の経緯については、おもしろいことが書いてある: