Generic Timing Model

• Start date: 01.01.2014
• Funded by: Universität Augsburg
• Local head of project: Prof. Dr. Theo Ungerer
• Local scientists: Florian Kluge

Abstract

Cyber-physical systems (CPS) typically exhibit a versatile timing behaviour. This behaviour emerges from complex interactions of physical processes with computations. It is constrained by requirements on the CPS. With the Generic Timing Model (GTM) we are developing a formalism to map timing properties and requirements within one model. One aim of GTM is to bridge the gap between an engineer's abstract view and a programmer's view in terms of scheduling models. GTM is designed such that it allows an intuitive mapping of parameters.

Utility functions derived from GTM help to specify timing requirements and visualise timing behaviour. Such utility functions can also be used to improve task scheduling, especially in overloaded real-time systems. There for, the Timing Models Scheduling Simulation (tms-sim) framework is being developed. tms-sim is available under the conditions of the GNU GPL v3.

Publications

2016

• tms-sim – Timing Models Scheduling Simulation Framework – Release 2016-07 
  Florian Kluge 
  Universitätsbibliothek der Universität Augsburg, Technical Report 
  Reference: 2016-04
• Notes on the Generation of Spin-Values for Fixed (m,k)-Patterns 
  Florian Kluge 
  Universitätsbibliothek der Universität Augsburg, Technical Report 
  Reference: 2016-01 
  The use of fixed (m,k)-patterns for scheduling of (m,k)-firm real-time tasks has the advantage of a higher predictability compared to other approaches. However, feasibility of task sets can be influenced negatively through high interferences between jobs classified as mandatory under the (m,k)-patterns. Rotating the (m,k)-patterns of single tasks by a spin value can improve feasibility of task sets. Due to minor errors and missing information in the original presentation of this approach, a reproduction of these results is difficult. In this report we provide the necessary corrections. Our evaluation shows that, after applying these corrections, the original results can be reproduced.

2015

• Utility-based scheduling of (m,k)-Firm Real-Time Task Sets 
  Florian Kluge, Markus Neuerburg, Theo Ungerer 
  28th International Conference on Architecture of Computing Systems (ARCS 2015), Porto, Portugal, March 24-27, 2015, Proceedings

2014

• tms-sim - Timing Models Scheduling Simulation Framework, Release 2014-12 
  Florian Kluge 
  Universitätsbibliothek der Universität Augsburg, Technical Report 
  Reference: 2014-07 
  tms-sim is a framework for the simulation and evaluation of scheduling algorithms. It is being developed to support our work on real-time task scheduling based on time-utility and history-cognisant utility functions. We publish tms-sim under the conditions of the GNU GPL to make our results reproducible and in the hope that it may be useful for others. This report describes the usage of the TMS framework libraries and how they can be used to build further simulation environments. It is not intended as a documentation of the single classes, which can be found in the Doxygen documentation.
• A Generic Timing Model for Cyber-Physical Systems 
  Florian Kluge, Mike Gerdes, Florian Haas, Theo Ungerer 
  Workshop on Reconciling Predictability and Performance (RePP'14)

2013

• History-Cognisant Time-Utility-Functions for Scheduling Overloaded Real-Time Control Systems 
  Florian Kluge, Florian Haas, Mike Gerdes, Theo Ungerer 
  Proceedings of 7th Junior Researcher Workshop on Real-Time Computing (JRWRTC 2013)