Thanks a lot!
Statistical validation can be correct for different Doppler
implementations in general.
However the way they add Doppler shifts is different.
I took your code and implemented in matlab and compared with jakes model
of matlab’s rayleighchan function.
There is a big mismatch in how they add Doppler shift from sample to
sample. May be this is an interpolation/filter issue.
This happens when I add Doppler shift in a deterministic way. I keep the
phase constant and time coherent.
After that I take a look at the Doppler shift that the methods add. As
an input I take ones to see the difference.
You know better than me how Doppler is added from sample to sample.
Especially, for a short time period (coherent time) within one OFDM
symbol the Doppler shift is linear.
At least we can assume that it is almost linear which we should not feel
the difference too much.
Classical Doppler models like Jakes model would represent this property.
Now I am trying to test it in reality and find out how it is added to
the input data from sample to sample.
And which model is better to take into account. I need it since I am
trying to compensate it back in the receiver as a part of my research
I will appreciate your comments on this.
Четверг, 6 марта 2014, 8:56 +13:00 от Richard C.