Distributed Embedded ARchitectures using COTS components

Home Summary Members Archive

Project Summary:

The main purpose of the DEAR-COTS project is the specification of an architecture based on the use of commercial off-the-shelf (COTS) components, able to support distributed computer controlled systems where safety and timeliness are major requirements. The DEAR-COTS project will address two main problems: 

How to apply the distributed computing technology to the solution of computer controlled problems using commercial off-the-shelf field-buses? 

How to support inter-working and interconnection of several "field-bus clusters" and the possibility of remotely access these "clusters interface" for supervision and configuration? 

A structured approach to the first problem needs to address several non-trivial issues, both at the communication and processing levels, such as: the impact of real-time and dependability requirements on the field-bus architecture, distributed fault-tolerant concurrent applications and the real-time support environment. 

In order to solve the second problem, the proposal foresees the use of emerging information technologies to cope with heterogeneity issues while providing a dependable user-friendly man-machine interface. This calls for innovative solutions addressing issues such as replica determinism of WWW servers, the use of object oriented technologies, the downloading/uploading within field-bus clusters and ensuring fault-containment between the field-bus and the "open network" domain. The DEAR-COTS proposed architecture will be validated through the development and test of a prototype, which will be integrated in a semi-industrial environment, with stringent safety and timeliness requirements. 

The DEAR-COTS architecture will constitute a step forward. In fact, a considerable research effort has been devoted to the design and validation of dependable real-time systems architectures. At the European level, the most significant examples are DELTA-4, MARS and GUARDS. However, none of these approaches is supported by standard field-buses, such as Profibus, FIP and CAN. One reason for this is that well known paradigms and fault-tolerance techniques, such as state-machines and replication management protocols, are hard to be supported by field-buses. On the other hand, specialised communication architectures, such as TTP, are costly. Thus, DEAR-COTS is motivated by the need for a fault-tolerant real-time distributed system based on standard field-buses, thus yielding highly available and low cost COTS technologies. 

The fundamental issue to be solved by DEAR-COTS architecture is how to apply the nowadays mature technology of distributed computing to the resolution of computerized control problems using standard COTS field-buses?



For problems or questions regarding this web contact [pmartins@di.fc.ul.pt].
Last updated: 31/10/00.