Turbo Multiuser Detection Architectures

Gerben Heinen

    Research output: ThesisPhD Thesis - Research UT, graduation UT

    24 Downloads (Pure)


    The discovery of Turbo Codes in 1996 by Berrou et. al. proved to be a huge boost for the research of channel coding. The Turbo Principle behind turbo codes was found to be applicable in other areas. One of these areas is Multiuser Detection. In this thesis, Turbo Multiuser Detection is investigated in order to answer two main questions. The questions concern the performance gain that is obtained when turbo multiuser detection is used instead of non-turbo multiuser detection and the convergence behavior of turbo multiuser detection. The performance gain is determined by comparing the bit-error-rate (BER) chart of a turbo multiuser detection architecture with the BER chart of a non-turbo multiuser detector. It was found that turbo multiuser detection results in a dramatical performance gain when Eb/N0 > 3 dB and more than one iteration is performed. The convergence behavior of turbo multiuser detection is analyzed with the help of EXIT charts. EXIT charts are recently proposed by S. ten Brink as a tool to analyze the convergence behavior of turbo architectures. EXIT charts are discussed in this thesis. An EXIT chart of a turbo multiuser detection architecture is created. From this chart, the minimum number of iterations to obtain the lowest BER possible are found. EXIT charts are also used to analyze the difference of iterating aposteriori and extrinsic information in a turbo architecture. The analysis shows that EXIT charts of a-posteriori information give results, which contradict the results of simulations on turbo architectures.
    Original languageEnglish
    • Slump, Cornelis Herman, Supervisor
    • Hoeksema, Fokke Wiert, Supervisor
    • Schiphorst, Roelof , Supervisor
    • Hofstra, K.L., Supervisor
    • Potman, J., Supervisor
    Place of PublicationEnschede
    Publication statusPublished - 2003


    Dive into the research topics of 'Turbo Multiuser Detection Architectures'. Together they form a unique fingerprint.

    Cite this