The NEXT TTA project enhances the structure, functionality and dependability of the time-triggered architecture (TTA) to meet the austere cost structure of the automotive industry, while satisfying the rigourous safety requirements of the aerospace industry. By placing the safety-relevant algorithms, that are formally analyzed, into intelligent replicated star couplers, NEXT TTA reduces the cost and generalizes the failure assumptions of the node computers. Event-triggered communication services are integrated into the TTA to increase the required flexibility. The synchronous programming environment LUSTRE and its toll set are extended for the TTA and automated worst-case-execution-time analysis is explored. CORBA compliant interfaces are provided in order to make TTA systems interoperable with the open information infrastructure. The limits of implementing the TTA with COTS components in the gigabit range are investigated.
It is the objective of NEXT TTA project to further improve the dependability and functionality of the time-triggered architecture (TTA). In particular the NEXT TTA project aims to: (1) enhance the dependability of the TTA by developing a new physical interconnection structure with intelligent star couplers; (2) integrate event-triggered and time-triggered communication within a single communications network while maintaining composability; (3) integrate the synchronous programming environment LUSTRE into the TTA; (4) develop highly automated WCET (worst case execution time) estimation algorithms; (5) provide a CORBA compliant interface of the TTA such that the real-time data within the TTA can be monitored remotely; (6) explore the limits of a COTS implementation of the TTA in the gigabit range.
DESCRIPTION OF WORK
The NEXT TTA project will design a (replicated) star coupler with centralized bus guardian to isolate arbitrary node failures. This allows node computers to be composed of COTS hardware that need not exhibit restricted failure modes (the present implementation requires fail-silent nodes). All facilities needed to isolate arbitrary node failures will be contained within the interconnection network, particularly within the star coupler. To support certification, the critical software in the star coupler will be formally analyzed. The integration of time-triggered and event-triggered communication will provide data channels with configurable quality of service while maintaining composability. As a proof of concept, the NEXT TTA project will provide a prototype implementation of a CAN protocol emulation within the TTA. This emulation enables the use of legacy CAN software in the TTA. Integrating LUSTRE within the TTA framework with appropriate tool support will give LUSTRE users more capabilities for distribution, thus widening the scope of LUSTRE usage. Extending LUSTRE partitioning techniques for dealing with non periodic events will also enhance the scope of its usage. The automatic scheduling studies and WCET tools foreseen in the project will be yield code meeting fine grain real-time constraints. The NEXT TTA will develop CORBA standardization proposals to both the real-time service interface and the diagnostic and maintenance interface of the TTA. Having these interfaces integrated into CORBA allows accessing TTA interfaces via the global Internet. A TTA prototype cluster consisting exclusively of high-end COTS hardware will be set up to explore the fundamental performance limits existing with state-of-the-art hardware in the gigabit range. The project will pinpoint the performance bottlenecks and suggest approaches for implementing TTA systems beyond the one gigabit barrier.
Funding SchemeCSC - Cost-sharing contracts
75794 Paris Cedex 16
38031 Grenoble Cedex 1
38400 Saint Martin D'heres
YO10 5DD York