Adaptive Equalization Tutorial: Trained and Blind Algorithms

  • Published on
    19-Oct-2015

  • View
    27

  • Download
    0

DESCRIPTION

Adaptive equalizers compensate for signal distortion attributed to intersymbol interference (ISI), which is caused by multipath within time-dispersion channels. This tutorial reviews trained and blind adaptive equalization algorithms including LMS, GSA and CMA.

Transcript

  • RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    Kevin Banovic

    October 14, 2005

    Department of Electrical and Computer Engineering,

    University of Windsor, Windsor, Ontario, Canada N9B 3P4

    ADAPTIVE EQUALIZATION:A TUTORIAL

  • KEVIN BANOVIC Slide 2

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    Adaptive equalizers compensate for signal distortion attributed to intersymbol interference (ISI), which iscaused by multipath within time-dispersive channels.

    Typically employed in high-speed communication systems, which do not use differential modulation schemes or frequency division multiplexing

    The equalizer is the most expensive component of a data demodulator and can consume over 80% of the total computations needed to demodulate a given signal [01]

    Adaptive Equalization

  • KEVIN BANOVIC Slide 3

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    Channel

    EqualizerAdjustment

    FIREqualizer

    DecisionDevice

    ErrorComputation

    s k( )y k( )

    e k( )

    r k( )s k( )

    TrainingSequence

    SymbolStatistics

    Blind Mode

    Decision-DirectedModeTraining Mode

    Adaptive Equalization

  • KEVIN BANOVIC Slide 4

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    The following quantities are defined for a linear equalizer with a real input signal: Equalizer tap coefficient vector:

    Equalizer input samples in the tapped delay line:

    Equalizer output: (Lf = equalizer length)

    r(k) =r0(k) r1(k) . . . rLf1(k)

    T=

    r0(k) r0(k 1) . . . r0(k Lf + 1)

    T

    fT (k) =f0(k) f1(k) . . . f(Lf1)(k)

    y(k) =

    Lf1Xi=0

    fi(k) r0(k i) = fT (k)r(k)

    Adaptive Equalization

  • KEVIN BANOVIC Slide 5

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    Error signal:

    where d(k) is the desired signal

    e(k) = d(k) y(k)= d(k) fT (k)r(k)

    Adaptive Equalization

  • KEVIN BANOVIC Slide 6

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    The mean-squared-error cost function is defined as [02]:

    When the filter coefficients are fixed, the cost function can be rewritten as follows:

    Where p is the cross-correlation vector and R is the input signal correlation matrix

    JMSE = Ee2(k)

    = E

    d2(k) 2d(k)y(k) + y2(k)

    = E

    d2(k)

    2E

    d(k)fT (k)r(k)

    +E

    fT (k)r(k)rT (k)f(k)

    JMSE = Ed2(k)

    2fT E {d(k)r(k)}| {z }

    p

    +fT Er(k)rT (k)

    | {z }R

    f

    = Ed2(k)

    2fTp+ fTRf

    Minimum Mean-Squared-Error (MMSE) Equalization

  • KEVIN BANOVIC Slide 7

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    The gradient of the MSE cost function with respect to the equalizer tap weights is defined as follows:

    The optimal equalizer taps fo required to obtain the MMSE can be determined by replacing f with fo and setting the gradient above to zero:

    fJMSE =JMSE

    f =JMSEf0

    JMSEf1

    . . .JMSEfLf1

    = 2p+ 2Rf

    0 = 2Rfo 2p fo = R1p

    Minimum Mean-Squared-Error (MMSE) Equalization

  • KEVIN BANOVIC Slide 8

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    Finally, the MMSE is expressed as follows:

    Questions:Why is the MSE cost function so popular?

    Is the calculation of fo practical?

    min = Ed2(k)

    2fTo p+ fTo Rfo

    = Ed2(k)

    2

    R1p

    Tp+

    R1p

    TRR1p

    = E

    d2(k)

    2pTR1p+ pTR1p

    = Ed2(k)

    pTR1p

    Minimum Mean-Squared-Error (MMSE) Equalization

  • KEVIN BANOVIC Slide 9

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    In practical situations, an analytic description of the cost surface is not available

    However, points can be estimated by time-averaging and search algorithms are used to descend the surface

    The method of steepest descent is a gradient search algorithm that adjusts the equalizer tap weights in direction of the negative gradient as follows [02][03]:

    Where is constant stepsize that controls the speed and accuracy of the equalizer tap adaptation.

    f(k + 1) = f(k) + fJMSE

    Method of Steepest Descent

  • KEVIN BANOVIC Slide 10

    RESEARCH CENTRE FOR INTEGRATED MICROSYSTEMS UNIVERSITY OF WINDSOR

    EQUALIZATION TUTORIAL

    For convergence, is chosen as follows [02][03]:

    Where max is the maximum eigenvalue of R At the minimum, this method requires a noisy estimate of

    the gradient during each iteration, which hinders its application in real applications

    However, it serves as the basis for an entire class of practical algorithms, including the algorithms to follow

    0

Recommended

View more >