>dear all,
>
>I was trying to implement constant modulus algorithm for blind channel
>equalization. I found that the filter taps of the equalizer are modified
>according to W(k+1)=W(k)+ mu * conj(x(k)) * e(k)
>
>where mu= adaptation size
>W the filter taps.
>x(k)= transmitted symbol
>e(k)= error given by e(k)= y(k) * (b^2 - abs(y(k))^2)
>
>where b^2= 13.2
>and y(k)= equalized output.
>
>my question is.......How should I start with initial value of W(k) the
>first tap of the equalizer. I thought of starting with 1. Is it a valid
>assumption????? Any suggestions????
>
>Thanks
>
Initialize the equalizer with one tap equal to one. If you know the
impulse response of the channel, initilize first tap if the mass of the
impulse response is on the left, initialize the center tap if the impulse
response is symmetric and initilize the last tap if the mass of the
impulse response is on the right. In case you do not know the impulse
response of the channel, a center tap initialized to one would do.
_____________________________________
Do you know a company who employs DSP engineers?
Is it already listed at http://dsprelated.com/employers.php ?
Reply by aamer●February 14, 20072007-02-14
dear all,
I was trying to implement constant modulus algorithm for blind channel
equalization. I found that the filter taps of the equalizer are modified
according to W(k+1)=W(k)+ mu * conj(x(k)) * e(k)
where mu= adaptation size
W the filter taps.
x(k)= transmitted symbol
e(k)= error given by e(k)= y(k) * (b^2 - abs(y(k))^2)
where b^2= 13.2
and y(k)= equalized output.
my question is.......How should I start with initial value of W(k) the
first tap of the equalizer. I thought of starting with 1. Is it a valid
assumption????? Any suggestions????
Thanks