Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 8 views

    An Adaptive Framework For WCDMA System Analysis in The EGEE Grid Infrastructure

    Uploaded by

    Eithu Thutun
    This document summarizes research being conducted to analyze wireless communication systems using the EGEE Grid infrastructure. Monte Carlo simulations are used to model stochastic systems, but can be computationally intensive. The researchers have developed a task farming framework to parallelize simulations across the EGEE Grid. This allows simulation of more complex systems. Future work includes exploiting parallelism within single simulations using MPI to address maximum job time limits of the Grid resources.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as DOC, PDF, TXT or read online from Scribd

    An Adaptive Framework For WCDMA System Analysis in The EGEE Grid Infrastructure

    Uploaded by

    Eithu Thutun
    8 views2 pages
    This document summarizes research being conducted to analyze wireless communication systems using the EGEE Grid infrastructure. Monte Carlo simulations are used to model stochastic systems, but can be computationally intensive. The researchers have developed a task farming framework to parallelize simulations across the EGEE Grid. This allows simulation of more complex systems. Future work includes exploiting parallelism within single simulations using MPI to address maximum job time limits of the Grid resources.

    Original Description:

    abstract_ID28.doc

    Original Title

    Abstract ID28

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    DOC, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes research being conducted to analyze wireless communication systems using the EGEE Grid infrastructure. Monte Carlo simulations are used to model stochastic systems, but can be computationally intensive. The researchers have developed a task farming framework to parallelize simulations across the EGEE Grid. This allows simulation of more complex systems. Future work includes exploiting parallelism within single simulations using MPI to address maximum job time limits of the Grid resources.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    8 views2 pages

    An Adaptive Framework For WCDMA System Analysis in The EGEE Grid Infrastructure

    Uploaded by

    Eithu Thutun
    This document summarizes research being conducted to analyze wireless communication systems using the EGEE Grid infrastructure. Monte Carlo simulations are used to model stochastic systems, but can be computationally intensive. The researchers have developed a task farming framework to parallelize simulations across the EGEE Grid. This allows simulation of more complex systems. Future work includes exploiting parallelism within single simulations using MPI to address maximum job time limits of the Grid resources.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    of 2

    An Adaptive Framework for WCDMA System Analysis in the EGEE Grid Infrastructure

    T. Athanaileas, P. Gkonis, G. Tsoulos and D. Kaklamani National Technical University of Athens School of Electrical and om!uter En"ineerin" #, $eroon Polytechneioy str, %o"rafou, Athens, Greece E&mail' thathan(esd.ntua."r

    Description of the activity )e !lan to use the EGEE infrastructure in order to evaluate and analy*e various techni+ues and scenarios for the desi"n of mo,ile communication systems. Thus far, -e have !erformed simulations of various scenarios for )ide,and ode Division .ulti!le Access /) D.A0 systems. 1n the future, -e intend to use EGEE resources for !erformin" further simulations on other mo,ile communication techni+ues, such as 23D., smart antennas and multi!le&in!ut multi!le&out!ut /.1.20 techni+ues. Grid added value The systems -e are analy*in" and evaluatin" are stochastic4 .onte arlo methods are used in order to !erform system simulations. These .onte arlo simulations may ,ecome e5tremely com!utationally demandin" as the !hysical dimensions and the loadin" of the simulated system increase. 1ndicatively, a sin"le simulation may last a day or even more in a standard deskto! com!uter and for the conver"ence of the .onte arlo to ,e satisfied, hundreds of simulations may ,e eventually needed. The EGEE infrastructure !rovides vast resources that may ,e e5!loited in order to efficiently reduce e5ecution time. Use of these resources -ill ena,le the simulation and analysis of more com!le5 systems that -ould ,e infeasi,le to !erform in a deskto! com!uter or a conventional small cluster. .onte arlo methods are -idely used in research fields /ray&tracin", molecular dynamics, finance, etc0 and our !ro!osed frame-ork may ,e used under these conte5ts -ith minimal modifications. E!perience or proposed activity 1n order to s!eedu! the e5ecution of the simulations, -e have e5!loited the !aralleli*a,ility of .onte arlo methods and have develo!ed a task&farmin" frame-ork for their !arallel e5ecution. The frame-ork is tailored to the "6ite middle-are and im!lements and mana"es a task&farmin" -orkflo- for !arallel .onte arlo simulations. 7riefly, once the initiali*ation !arameters are "iven, the a!!lication dis!atches several 8o,s to the "rid, assi"nin" a num,er of sin"le simulations to each 8o,. Due to the hetero"eneity of the infrastructure and the ma5imum PU time availa,le for each 8o,, the num,er of simulations assi"ned to each 8o, is dynamically determined ,y a sim!le heuristics that takes into consideration e5ecution time and .a5 PUTime and that is incor!orated to the 8o, itself. Until no-, -e have e5ecuted several scenarios re"ardin" a ) D.A system, for various ,ase station confi"urations, !hysical dimensions and system loadin". Future evolution A restrainin" factor to our a!!roach is that in certain scenarios a sin"le simulation may e5ceed the .a5 PUTime limit mentioned a,ove. 1n order to sur!ass this !ro,lem, -e !lan to e5!loit any !aralleli*ation !attern may ,e a!!lica,le to a sin"le simulation and develo! .P1 !ro"rams for concurrent e5ecution at simulation level. This, ho-ever, restricts usa,le resources to those of an .P1

    su!!ortin" site, as the su!!ort for .P1 a!!lications across different sites is still missin". Acknowled"ements The -ork !resented in this !a!er has ,een funded ,y the !ro8ect PENED 9::;. The !ro8ect is co financed <=> of !u,lic e5!enditure throu"h E & Euro!ean Social 3und, 9=> of !u,lic e5!enditure throu"h Greek .inistry of Develo!ment & General Secretariat of ?esearch and Technolo"y and throu"h !rivate sector, under measure @.; of 2PE?AT12NA6 P?2G?A. A 2.PET1T1BENESSA in the ;rd ommunity Su!!ort Pro"ram.

    You might also like