Site: TRON Laboratory, Sakamura Laboratory
University of Tokyo
4-6-1 Komaba, Meguro-Ku
URL: http://tron.um.u-tokyo.ac.jp/TRON Date Visited: May 27, 1995
Report Author: J. Hollan
The TRON (The Real-time Operating System Nucleus) Project was originated by Dr. Ken Sakamura of the University of Tokyo 11 years ago. It is funded by a consortium of private companies and receives no government support. Having such a consortium based in a university is unusual. As of 1995, 69 companies were members of the TRON Association. The cost of membership includes a one-time entry fee that ranges from ¥900,000 (~$9,000) for an organization with board member status to ¥100,000 (~$1,000) for companies with more limited access. Annual dues range from ¥500,000 to ¥4,500,000 (~$5,000 to ~$45,000) for these two bounding membership categories. The original goal of the TRON project was to explore the use of large networks of microcomputers in civilian applications in which people are "surrounded by massive computing power." This led to a focus on new uses for computers and development of new architectures for distributed applications. One large project was the construction of a house of the future. This "TRON-concept intelligent house" was built in Tokyo in 1989. It had over a thousand computing elements (microcontrollers, sensors, and actuators). The house was dismantled in 1992, but there are now plans to build an intelligent office building.
The research and development philosophy of the TRON project is simply expressed by two phrases: computing everywhere and sticking to reality. The computing-everywhere view is similar to but predates the Xerox Parc idea of ubiquitous computing. The view forecasts an environment in which computers are spread everywhere and are part of most common objects. For example, an office would have computers embedded in the floor, walls, ceiling, chairs, desk, and common objects like pens. These computers would be interconnected into a network to support office work. The sticking-to-reality principle holds that future computing applications need to be tied closely to and integrated with real environments. This leads to a focus on augmenting reality rather than virtual reality.
The TRON project has historically attempted to address two parallel research fronts. One is exploration of application projects like the intelligent house, and the other is foundational projects concerned with developing basic technology. One of the fundamental projects has been the design of a series of VLSI TRON chips.
The project has proposed an open computing architecture and has developed a series of specifications. The goal is what is termed "loose standardization" that provides a set of rules for computer interfaces without presupposing particular hardware or software. The TRON-specification chip is designed to support the distributed applications envisioned. These chips are being manufactured by a number of companies (Toshiba, Mitsubishi, Hitachi, Fujitsu, and Matsushita). The GMicro/500 is the highest performance chip (32-bit microprocessor with 32-bit address bus and 64-bit data bus). In addition, a series of peripheral chips have been developed. The software development environment is C-based and a set of GNU software development tools (GCC and friends) have been ported to the TRON environment.
Other fundamental subprojects include: ITRON (Industrial TRON), an architecture for real-time operating systems for embedded computers; BTRON (Business TRON), an architecture for personal computers or workstations that focuses on providing a uniform human machine interface; and CTRON (Central/Communication TRON), a set of operating system interfaces for supporting switching and communication functions. Application projects include: intelligent house (1989-1992), intelligent building (final design stage), computer city, automobile traffic information system. Another TRON subproject focuses on the human-machine interface. Effective interfaces are seen as fundamental to creation of a highly functional distributed system. The project emphasizes four aspects of HCI: standardizing operation methods, taking into account the needs of the disabled, supporting multiculturalism, and assuring safety. The TRON Human Machine Interface Specifications were issued in 1993. Following this release, the group that designed them sponsored an HCI design competition for household and consumer products. The goal was to promote use of the TRON specifications, convey the importance of the human interface, and evaluate the proposed specifications.
This is a very ambitious project centered around a visionary project leader. The vision of computing everywhere is one that has had impact on conceptions of human-computer interaction and provides an interesting alternative to the more prevalent view of personal machines. This embedded computing view is shared by the ubiquitous computing efforts at Xerox Parc and by such recent projects as the intelligent room work in the MIT AI Laboratory.
It is hard to judge this project. Although there appears to be continuing support from companies and production of TRON architecture chips, there is little information available about the details of projects using the TRON architecture. Also, one wonders about the ability of small-volume chips to keep pace with larger efforts. In addition, the software efforts appear to be such that integration with other application programs will be difficult.
Specifics about the results of application projects such as the intelligent house have been difficult to obtain. This is perhaps a natural consequence of consortium funding and the need to provide details primarily to consortium members. Almost as interesting as the technology and novel vision for the future of computing would be more details about the sociology of establishing this very nontraditional company-university vehicle for research funding. That the consortium has existed for over a decade is in sharp contrast to the history of similar research consortia in the United States.
There have been a series of annual TRON Project Symposia. The 1987-1990 proceedings were published by Springer-Verlag in Tokyo. Since 1991 the proceedings have been available from the IEEE Computer Society Press. In addition, TRON Specifications (English Language) can be purchased from the TRON Association, U.S. Liaison Office, 6428 Clay Allison Pass, Austin, TX 78749 (FAX: 512-301-3508; Tel.: 512-301-3507).
[Note: for a more general description of the TRON Program, see JTEC Panel Report on Advanced Computing in Japan (Harrison et al., 1990, NTIS PB90-215765), 131-136.]