Referring to your code in the Matlab file named 'sim_PSK_passband.m' :

1)The following line is written in the code:
suT= su*T;
The unit basis signal here is multiplied by T so that when the correlation is calculated, the correlation will be an unbiased estimate and not just the raw unscaled value. Is the reasoning right ? Or if it is wrong, what is the purpose of multiplying by T ?

2) The following line is written
sgmsT=sqrt(sigma2/T);
What is the purpose of dividing by T here ? Can you please explain mathematically ? I know that the noise power is being varied for every iteration by changing the SNR. But why do we specifically divide the noise power(sigma2)by T before applying it to the random noise generated in the code ?

I am a graduate student who is currently using your book titled “Matlab/Simulink for Digital Communication” to program in Matlab an efficient equalizer for a bandlimited signal distorted by severe Inter Symbol Interference (ISI). I had followed your example code for symbol spaced adaptive linear and DFE LMS equalizers in Chapter 6: ‘Bandlimited Channel and Equalizer’ of your book.

Request 1: I would like to implement a fractionally spaced adaptive linear and DFE LMS equalizer. Can you kindly upload a Matlab code file showing how to realize a fractionally spaced (i.e. equalizer tap weights spaced at T/2, T being symbol period. The incoming signal is oversampled at 2 samples per symbol.) linear and DFE LMS equalizers for a baseband binary(+1/-1) pulse amplitude modulated signal.

Request 2: I would like to also implement a fractionally spaced (i.e. equalizer tap weights spaced at T/2, T being symbol period.) adaptive Fast RLS (Kalman) DFE equalizer for a baseband binary(+1/-1) pulse amplitude modulated signal. Can you also kindly upload a Matlab code file showing how to achieve this in Matlab.