Architectures of Real-time Distributed TMO Kernels
and Middlewares for Portability and Interoperability

Jung Guk Kim: School of Computer Engineering & Information Communications, San 89-1, Mohyun, Yongin, Kyunggi, 449-791, Korea (TEL) +82-31-330-4367 (FAX) +82-31-339-3443 (E-Mail) jgkim@hufs.ac.kr (URL) http://rtdcs.hufs.ac.kr

Moon Hae Kim: Division of Computer Science and Engineering, 1, Hwayang-dong, Gwangjin-gu, Seoul 143-701, Korea (TEL) +82-2-450-3859 (FAX) +82-2-455-2589 (E-Mail) mhkim@konkuk.ac.kr (URL) http://www.rtselab.org

Abstract

The TMO (Time-triggered Message-triggered Object) model formalized earlier by Kim and Kopetz is a well-known real-time object model for distributed and timeliness-guaranteed computing. In a TMO instance, two special types of member threads run with their timing constraints. Those are periodic time-triggered threads with deadlines and message-triggered threads activated by distributed IPC messages. Since the invention of the TMO model, there have been much successful progresses in developing TMO execution engines on various kernel platforms by several co-operating research groups. Some engines were built as middlewares on Windows and Linux platforms and the others were developed as kernels by embedding the key TMO modules into existing kernels such as Linux and eCos. Up to now, the middleware engines have been mainly used in developing multimedia services, real-time simulations, and so on. On the other hand, the kernel engines such as TMO-Linux and TMO-eCos have been used in developing embedded real-time applications for controlling and monitoring of machines such as robots. As the middleware and kernel engines become to have quite high level of stabilities, the next step is to establish an integrated TMO environment for developing large real-time networked control systems with various servers, monitoring stations, and embedded equipments. The key issues to accomplish this unified TMO collaboration environment on heterogeneous machines are portability for extending variety of kernel platforms and interoperability among TMO-engines. For higher portability with which we can easily build a TMO engine on a new platform, a number of key components of TMO kernels and middlewares must be identified as separable modules. Those are the deadline scheduler, the intra-node and inter-node TMO IPC subsystem, API modules for time-triggered and message triggered methods, and the clock synchronization subsystem. For interoperability, we need a distributed IPC subsystem with common communication channel APIs and message semantics over various protocols.

In this paper, we present not only the general architectures of some TMO kernels and middlewares but also methods for increasing portability and interoperability, that is, how the key components of a TMO engine such as the deadline scheduler and the IPC subsystem can be modularized and where functional modules of a middleware or an existing kernel can be inserted. Finally, an example TMO configuration of a networked robot control and monitoring system with heterogeneous platforms is described.

Short Biography

Jung Guk Kim: Jung Guk Kim received his B.S. degree in Computer Science at Seoul National University in 1977, M.S. degree in Computer Science at Korea Advanced Institute of Science & Technology in 1989, and Ph.D. degree in Computer Science at Korea Advanced Institute of Science & Technology in 1986. He was a visiting scholar of University of California, Irvine from 1994 to 1995. In 1983, he joined Hankuk University of Foreign Studies, Seoul, Korea, where he is now a professor in the School of Computer Engineering & Information Communications. Since 1999, he has been a researcher of Software Research Center, which is funded by the ITRC (Information Technology Research Center) support program of Ministry of Information and Communication in Korea. He is now one of the vice presidents of Korea Information Science Society and he is in charge of the division of OSS (Open Source Software) promotion. His research interests include distributed real-time systems and embedded systems.

Moon Hae Kim: Moon Hae Kim received his B.S. degree in Electrical Engineering at Seoul National University in 1979, M.S. degree in Computer Science and Engineering at University of South Florida in 1985, and Ph.D. degree in Computer Science at University of California, Berkeley in 1991. He was a visiting scholar of University of California, Irvine from 2000 to 2001. In 1991, he joined Konkuk University, Seoul, Korea, where he is now a professor in the Division of Computer Science and Engineering. Since 1999, he has been the director of Software Research Center, which is funded by the ITRC (Information Technology Research Center) support program of Ministry of Information and Communication in Korea. His research interests include distributed real-time systems, embedded systems, fault-tolerant computing, and software engineering.