Guys i am working on ofdm simulation.after taking the ifft of my crriers (having baseband data in the form of matrix) mycarriers are not orthogonal. As the ifft is responsible for making carriers orthogonal in order to acheive OFDM. I am using built in command of ifft in matlab. For eg x=ifft (carrier_matrix,128) then when i take dot product of my vectors the result is a non-zero number or when i perform x*x' the result is not identity matrix. It means that orthogonality is not acheived among subcarriers. please help me

# orthogonality not acheived after IFFT!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Started by ●March 8, 2007

Posted by ●March 8, 2007

On Thu, 08 Mar 2007 18:47:03 -0000, awais <awais.engr@gmail.com> wrote:> Guys i am working on ofdm simulation.after taking the ifft of my > crriers (having baseband data in the form of matrix) mycarriers are > not orthogonal. As the ifft is responsible for making carriers > orthogonal in order to acheive OFDM. I am using built in command of > ifft in matlab. > For eg x=3Difft (carrier_matrix,128) > then when i take dot product of my vectors the result is a non-zero > number > or when i perform x*x' the result is not identity matrix. > It means that orthogonality is not acheived among subcarriers.What do you mean by "baseband data in the form of matrix"? The only thi= ng = I can assume this means is that each column of carrier_matrix contains t= he = data for one OFDM symbol. If this is the case, then x*x' will not be the identity matrix (unless t= he = original data was especially chosen). -- = Oli

Posted by ●March 10, 2007

origional data is random and is not especially chosen by baseband data i mean a stream of 1's and -1's in the form of complex no for eg 1 -> 1+ j0 -1 -> 1 - j0 this data is then transformed into matrix which has 4 columns. i have to take ifft of this matrix in which my columns act as sub_carriers so the problem is that when i take ifft my resulting suncarriers are not orthogonal thats where the problem lies Should i write my own function for taking DFT and dont use matlab function help me thanks

Posted by ●March 10, 2007

awais wrote:> origional data is random and is not especially chosen > by baseband data i mean a stream of 1's and -1's in the form of > complex no > for eg 1 -> 1+ j0 > -1 -> 1 - j0 > this data is then transformed into matrix which has 4 columns. > i have to take ifft of this matrix in which my columns act as > sub_carriersIf you arrange your input data like this, then the result of y = ifft(x) will be a matrix; each column of y will be the IFFT of the corresponding column in x. Therefore: * Each column of x and y is a separate OFDM symbol. * The rows of x correspond to sub-carriers. * Neither the rows nor the columns of y correspond to sub-carriers. * y contains time-domain data. * In general, the columns of y won't be orthogonal.> so the problem is that when i take ifft my resulting suncarriers are > not orthogonal > thats where the problem lies >-- Oli

Posted by ●March 12, 2007

how can i acheive orthogonality in frequency or time domain i-e how can i make my subcarriers orthogonal in time domain

Posted by ●March 12, 2007

On Mar 12, 5:19 pm, "awais" <awais.e...@gmail.com> wrote:> how can i acheive orthogonality in frequency or time domain > i-e how can i make my subcarriers orthogonal in time domainThey are orthogonal. But neither the columns nor the rows of y (in y=ifft(x)) correspond to sub-carriers, so neither the columns nor the rows of y will be orthogonal. -- Oli