Reflection 2001 - The Third International Conference on Metalevel Architectures and Separation of Crosscutting Concerns - Kyoto, Japan, September 25-28, 2001


9/27 (Thu) 10:30-11:30

Dynamic Adaptation: Lessons Learned from Microkernel to Smart Space Computing

Hideyuki Tokuda, Dean of Graduate School of Media and Governance and Professor, Faculty of Environmental Information, Keio University

In this talk, we will discuss dynamic adaptation of programmed systems based on our software development efforts ranging from the RT-Mach microkernel to the Smart Space Computing Environment.

We have been working on the development of RT-Mach microkernel for embedded real-time systems and distributed multimedia systems for over several years. One of the challenging issues was to create dynamic adaptation mechanisms for supporting various types of run-time environment and application software. We have classified these mechanisms in to three categories based on the micorkernel architecture, namely SS (Service by Server), SL (Service by Library), and SK (Service by Kernel Module). We first look back these mechanisms based on our practical examples such as a UNIX server, Environmental server, scheduling policy module, and real-time threads.

We then talk about new challenges in creating dynamic adaptation mechanisms for smart space computing environment. In our smart space computing environment, a user should be able to continue the execution of his/her task despite of environmental changes, such as different processor, display, network, location, time zone, and so on. System and application software should be adaptable to such user's environmental changes as well as user's needs. We support context-aware and adaptable software together with dynamic service finding, binding, and re-binding mechanisms. A few examples of adaptable software such as VNA(Virtual Networked Appliances) and WN(Wearable Networks) will be demonstrated.

In conclusion, we summarize the lessons learned from the development of dynamic adaptable software.

Hideyuki Tokuda, Ph.D. in Computer Science

Hideyuki Tokuda received his B.S. and M.S. degrees from Keio University, Japan in 1975 and 1977 and a Ph.D. degree in Computer Science from the University of Waterloo in 1983. Between 1983 and 1990, he was a Senior Research Computer Scientist at Computer Science Department, Carnegie Mellon Universty. In 1990, he became an Associate Professort, Faculty of Environmental Information, Keio Univ. in Japan. He is currently Dean of Graduate School of Media and Governance and a Professor in the Faculty of Environmental Information, Keio University, Japan.

His research interests include distributed real-time systems, multimedia systems, mobile systems, communication protocols, embedded systems, information appliances, and intelligent environmental computing. He has created many operating systems and software tools such as Real-Time Mach, the ARTS Kernel, Shoshin, Scheduler 1-2-3, and ARM (Advanced Real-Time Monitor).

He is a member of the IEEE, the ACM, the IPSJ, and the Japan Society for Software Science and Technology (JSSST). He is currently the board member of IPSJ(Information Processing Society of Japan), JDLA (Japan Distance Learning Association), and JUEC (Japan Universities Association for Computer Education). He is also chairing SIGIAC (Information Appliance Computing) Group in IPSJ.