University of Leicester
Browse
2011donglmphil.PDF (872.83 kB)

The application of scheduler agents in time-triggered embedded systems

Download (872.83 kB)
thesis
posted on 2011-08-25, 14:12 authored by Lei Dong
This thesis is concerned with the monitoring of embedded systems in which timing behaviour is the key concern. The focus of this work is on the development of a “scheduler agent” (SA) which is used to monitor the temporal behaviour of embedded systems. It is assumed that the system to be monitored employs a time-triggered software architecture. This thesis begins by providing a review of timing issues in embedded systems, and followed by a review of previous research on runtime monitoring techniques (including hardware, software and “hybrid” approaches). The SA is then introduced. It consists of two parts: an internal monitor (IM) and an external analyzer (EA). Both of the IM and the EA have to work cooperatively in order to obtain information from the target system. The communication between them relies on a GPIO interface. However, an encoding technique is required since modern microcontrollers may not have enough GPIO port pins to represent all tasks in the target system. A simple and effective encoding technique has been introduced in this thesis to address this issue. Two versions of the SA – Passive SA (PSA) and Active SA (ASA) – are implemented. PSA retrieves task information from the “instrumented” target system passively. ASA takes advantages from the TTC architecture employed by the target system, in which the monitoring process collects information from the target system at the time when a task is due to start and end. We also developed a SA automation tool which can automatically generate the SA code for the external analyzer and the target system. This tool is used in the case study to generate the source code for both PSA and ASA. In the case study presented in this thesis, we confirm that the functionality of SA has in line with its requirements, since it is capable to measure task execution time and detect temporal errors in the target system. Finally, the conclusions of this thesis with a discussion of the results and some suggestions for further work in this important area are presented.

History

Supervisor(s)

Pont, Michael; Short, Michael

Date of award

2011-07-01

Awarding institution

University of Leicester

Qualification level

  • Masters

Qualification name

  • Mphil

Language

en

Usage metrics

    University of Leicester Theses

    Categories

    No categories selected

    Keywords

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC