A new platform library in Metropolis allows the use of the fault tolerant data flow (FTDF) model of computation. FTDF is amenable to specifying periodic feedback control applications. It is structured in such a way that it is possible to analyze it formally and to use automatic and semi-automatic synthesis techniques for obtaining an efficient fault-tolerant implementation of the application under design. The use of the platform is illustrated by a simple controller example: an inverted pendulum stabilizer.
Redundancy can be easily introduced using the Metropolis refinement mechanism. Finally it is possible to simulate and to inject faults in the design. Some of the properties of FTDF include self-synchronization and automatic degradation in case of faults.
1Outside Adviser (non-EECS), Cadence Berkeley Labs