tom1379
Newbie level 2
Hi,
I am trying to simulate LMS adaptive filtering in matlab but am having some problems forming the equalisation matrix which needs to be of dimension nTx by nRx (4x4 in this case)
My error signal is a 4x1 matrix which is the result of x
,i)- W' *y,i) where x,i) is a 4x1 matrix comprising 4 training symbols and y,i) is the received vector for those training symbols. When trying to update the equalisation matrix using W=W+2*mu*e,i)*y,i), an error is given because obviously e and y are both 4x1 matrices and cannot be multiplied. Should e be multiplied by the transpose of y?
I have attached my LMS code to show you exactly what I mean. It does run and produce a graph which is incorrect but only because I have transposed the y signal in the error calculation.
Any help would be greatly appreciated!
I am trying to simulate LMS adaptive filtering in matlab but am having some problems forming the equalisation matrix which needs to be of dimension nTx by nRx (4x4 in this case)
My error signal is a 4x1 matrix which is the result of x
,i)- W' *y,i) where x,i) is a 4x1 matrix comprising 4 training symbols and y,i) is the received vector for those training symbols. When trying to update the equalisation matrix using W=W+2*mu*e,i)*y,i), an error is given because obviously e and y are both 4x1 matrices and cannot be multiplied. Should e be multiplied by the transpose of y?I have attached my LMS code to show you exactly what I mean. It does run and produce a graph which is incorrect but only because I have transposed the y signal in the error calculation.
Any help would be greatly appreciated!