3rd International ICST Conference on Performance Evaluation Methodologies and Tools

Research Article

End-to-end delays in polling tree networks

Download572 downloads
  • @INPROCEEDINGS{10.4108/ICST.VALUETOOLS2008.4248,
        author={P. Beekhuizen and T.J.J. Denteneer and J.A.C. Resing},
        title={End-to-end delays in polling tree networks},
        proceedings={3rd International ICST Conference on Performance Evaluation Methodologies and Tools},
        publisher={ICST},
        proceedings_a={VALUETOOLS},
        year={2010},
        month={5},
        keywords={Polling Concentrating tree networks HoL-based service dis- ciplines Networks on Chips},
        doi={10.4108/ICST.VALUETOOLS2008.4248}
    }
    
  • P. Beekhuizen
    T.J.J. Denteneer
    J.A.C. Resing
    Year: 2010
    End-to-end delays in polling tree networks
    VALUETOOLS
    ICST
    DOI: 10.4108/ICST.VALUETOOLS2008.4248
P. Beekhuizen1,*, T.J.J. Denteneer2,*, J.A.C. Resing3,*
  • 1: Philips Research, Eindhoven, The Netherlands and EURANDOM, Eindhoven University of Technology
  • 2: Philips Research, Eindhoven, The Netherlands
  • 3: Eindhoven University of Technology, Department of Mathematics and Computer Science, Eindhoven, The Netherlands
*Contact email: beekhuizen@eurandom.tue.nl, dee.denteneer@philips.com, resing@win.tue.nl

Abstract

We consider a tree network of polling stations operating in discrete-time. Packets arrive from external sources to the network according to batch Bernoulli arrival processes. We assume that all nodes have a service discipline that is HoL-based. The class of HoL-based service disciplines contains for instance the Bernoulli and limited service disciplines, and hence also the classical exhaustive and 1-limited. We obtain an exact expression for the overall mean end-to-end delay, and an approximation for the mean end-to-end delay of packets per source. The study is motivated by Networks on Chips where multiple processors share a single memory.