Links to document based files:

• Entire Document (95 pages, 1124 KBytes)
  
  
• Document in pieces
  • Abstract (1 page, 3 KBytes)
  • Title (1 page, 2 KBytes)
  • Acknowledgements (2 pages, 4 KBytes)
  • Table of Contents (2 pages, 8 KBytes)
  • List of Tables (1 page, 2 KBytes)
  • List of Figures (1 page, 4 KBytes)
    
    
  • Chapter 1: Introduction (6 pages, 14 KBytes)
  • Chapter 2: Background (29 pages, 64 KBytes)
  • Chapter 3: Target Environment (12 pages, 30 KBytes)
  • Chapter 4: Design and Implementation of the Tool (17 pages, 33 KBytes)
  • Chapter 5: Model of Execution Time and Overhead in a Real-time System (13 pages, 36 KBytes)
  • Chapter 6: Experimental Validation and Results (14 pages, 21 KBytes)
  • Chapter 7: Summary and Future Work (2 pages, 5 KBytes)
  • Bibliography (2 pages, 6 KBytes)

Abstract

Our objective is to automate the process of debugging the timing parameters of a real-time embedded system. As a major step towards this objective, we have developed a tool that predicts the best way of solving timing problems using its analysis and predictor units. Schedulability analysis of actual measured data from the system is used to pinpoint the timing problems of the underlying embedded system. This tool provides detailed analysis of the schedulability of each given taskset using five different scheduling algorithms, with their various overheads incurred, missed deadlines, and graphs for the time traces, using the actual measured data from the system. These results are used to verify and fine-tune the timing parameters of the system. The accuracy of the tool was evaluated by comparing the predicted maximum breakdown utilization factors of a taskset using the individual task execution times and periods, with the results obtained directly from our real-time operating systems. Preliminary results using three different tasksets indicate 97% and 98% accuracy of the tool, using breakdown utilization factors of execution times and periods, respectively.