Publication detail

Benchmark Based Method for Predicting Computing Time of Simulink Model Running on Embedded Target

LAMBERSKY V., GREPL R.,

Original Title

Benchmark Based Method for Predicting Computing Time of Simulink Model Running on Embedded Target

English Title

Benchmark Based Method for Predicting Computing Time of Simulink Model Running on Embedded Target

Type

conference paper

Language

en

Original Abstract

Developing a method which allow the predicting of the computing time of the developed algorithm, based on the specification in the higher level programming language is not yet satisfactorily solving challenge. This paper presents a new method for predicting the computing time of control algorithms implemented in a Simulink environment running on the target hardware. This method uses a refined benchmark database which achieves much better computing time prediction accuracy compared to other benchmark score-based methods used today. The benchmark score database can be created virtually for any target and no further information is required for this method to work. This makes the refined benchmark method very universal and portable compared to present virtual software benchmarks which require a very complex MCU description. This paper contains implementation details of the presented refined benchmark prediction method and demonstrates its accuracy on selected control algorithms (e.g. PID, state space controller).

English abstract

Developing a method which allow the predicting of the computing time of the developed algorithm, based on the specification in the higher level programming language is not yet satisfactorily solving challenge. This paper presents a new method for predicting the computing time of control algorithms implemented in a Simulink environment running on the target hardware. This method uses a refined benchmark database which achieves much better computing time prediction accuracy compared to other benchmark score-based methods used today. The benchmark score database can be created virtually for any target and no further information is required for this method to work. This makes the refined benchmark method very universal and portable compared to present virtual software benchmarks which require a very complex MCU description. This paper contains implementation details of the presented refined benchmark prediction method and demonstrates its accuracy on selected control algorithms (e.g. PID, state space controller).

Keywords

embedded target, benchmarks, predicting computing time, Simulink model

Released

09.12.2016

ISBN

9788001058831

Book

Mechatronika (ME), 2016 17th International Conference on Mechatronics

Pages from

222

Pages to

229

Pages count

8

Documents

BibTex


@inproceedings{BUT132721,
  author="Vojtěch {Lamberský} and Robert {Grepl}",
  title="Benchmark Based Method for Predicting Computing Time of Simulink Model Running on Embedded Target",
  annote="Developing a method which allow the predicting of the computing time of the developed algorithm, based on the specification in the higher level programming language is not yet satisfactorily solving challenge. This paper presents a new method for predicting the computing time of control algorithms implemented in a Simulink environment running on the target hardware. This method uses a refined benchmark database which achieves much better computing time prediction accuracy compared to other benchmark score-based methods used today. The benchmark score database can be created virtually for any target and no further information is required for this method to work. This makes the refined benchmark method very universal and portable compared to present virtual software benchmarks which require a very complex MCU description. This paper contains implementation details of the presented refined benchmark prediction method and demonstrates its accuracy on selected control algorithms (e.g. PID, state space controller).",
  booktitle="Mechatronika (ME), 2016 17th International Conference on Mechatronics",
  chapter="132721",
  howpublished="electronic, physical medium",
  year="2016",
  month="december",
  pages="222--229",
  type="conference paper"
}