Power saving and fault-tolerance in real-time critical embedded systems
Journal of Systems Architecture: the EUROMICRO Journal
| Hi-index | 0.00 |
IN THE NEXT GENERATION HARD REAL-TIME SYSTEMS, IT IS ANTICIPATED THAT MANY APPLICATIONS WILL CONTAIN COLLECTIONS OF TASKS RUNNING ON A DISTRI- BUTED SYSTEM, AND THAT MANY OF THESE TASKS MUST BE EXECUTED ACCORDING TO CERTAIN PRECEDENCE AND REAL-TIME CONSTRAINTS. THE SPACE STATION, NEW PROCESS CONTROL APPLICATIONS INTEGRATED WITH EXPERT SYSTEMS SOFTWARE, OTHER REAL-TIME AI APPLICATIONS SUCH AS AN AUTONOMOUS LAND ROVER, AND SDI ARE SOME EXAMPLES OF SUCH FUTURE APPLICATIONS. IN THESE NEXT GENERATION, LARGE- SCALE, HARD REAL-TIME SYSTEMS, ONE WILL FIND TASKS WITH DIFFERING FUNCTION- AL COMPLEXITY AND MULTIPLE LEVELS OF TIMING CONSTRAINTS. FOR EXAMPLE, PROCESSING DATA FROM A SENSOR IS SOMETIMES FUNCTIONALLY SIMPLE, AND MIGHT NEED TO OCCUR IN THE MICROSECOND RANGE, WHILE CONSISTENT UPDATES OF REPLI- CATED FILES IS MORE COMPLEX AND MIGHT NEED TO OCCUR IN THE SECONDS RANGE. AT EVEN HIGHER LEVELS, CONTROL AND COORDINATION TASKS MIGHT BE QUITE INTRI- CATE AND NEED TO OCCUR IN THE SECONDS OR MINUTES RANGE, AND MOVEMENT OF MATERIAL IN AN AUTOMATED FACTORY MIGHT BE BASED ON COMPLICATED COORDINATION ALGORITHMS, BUT HAVE DEADLINES IN THE RANGE OF MINUTES OR EVEN HOURS. THE WORK PRESENTED IN THIS PAPER IS APPLICABLE TO THE GROUPS OF TASKS WHERE THEIR DEADLINES ARE IN THE ONE HALF SECOND RANGE OR GREATER, AND, THUS, DOES NOT ADDRESS TIGHT TIMING CONSTRAINTS SUCH AS THOSE THAT OCCUR IN SENSOR PROCESSING. WE ASSUME THAT THE SYSTEM UNDER CONSIDERATION HAS LOWER