Doppler frequency shift and frequency offset ?

In a normal receiver frequency offset can be detected and compensated by AFC(Automated Frequency Controller). If Doppler frequency shift exists, can this frequency offset be detected by AFC in the same way? Or we have to do something special to cancel Doppler effect?

well it depend purely on how much shift in frequency ( which mainly depends on velocity and direction of moving transmitter or reciever ) is if it is with in correctable range then AFC will have no problems in correcting it i don't think so we have to do something special if u are working in ghz and and realative velocity of reciever is around 80km/hr because its gonna introduce only frequency shift of around 100hz

for more details read
article 2.2 of
WIRELESS COMMUNICATIONS
Andrea Goldsmith

Another aspect is the bandwidth of the signal. Different parts will be shifted the same relative amount but will produce different absolute shifts.

In my mind Doppler frequency shift is in fact some thing like frequency spread, not just a SHIFT. Typically the doppler frequency spread is U-shape where the frequency at (fc+fd)(fc: carrier frequency, fd: doppler shift) dominates, therefore what AFC detect is not a pure frequency offset. If the frequency spread is in the working range of AFC, the problem is how does AFC detect it ?

analog AFC circuits are designed for a single carrier modulation. It changes the tuning of the receiver so that the mean/median peak of the power spectrum is in the center of the IF stages.

There may be more sophisticated DSP methods of doing this.

If AFC in the baseband processor CAN detect the doppler frequency shift ( I think AFC cannot distinguish Doppler frequency shift and carrier frequency offset in this case, right?), AFC will adjust the LO to cancel the freq offset. However, considering Jakes U-shape Doppler frequency model, there are two major tone (fc+fd) and (fc-fd) at the front-end of the receiver( Assume that signal bandwidth B>>fd and fc >> fd), LO equals any one will miss the other tone. And it is obvious that is this case the equivalent channel baseband model will occupy bandwith of 2*fd. Is that right?

There are two Doppler situations. The classical one from the past is a single shift because there is a direct link to the other terminal, such as a space satellite and a ground station with a clutter free line of sight propagation path.

I think you are discussing the land mobile situation where there are scattering surfaces all around and the movement of the terminal will produce a plus shift of the frequency caused by movement from some local reflector and a minus shift from some other reflector at the same time.

This latter case will not work with the analog AFC methods. The DSP will have to be very sophisticated to model the transmission path and take the inverse of it. I am not a DSP expert and cannot go into detail on how this could be implemented.

I think you are right, after checking some other papers and books. Doppler frequency shift can be compensated directly by AFC when LOS path dominates such as satellite communication. However when LOS does not dominate and deep scattering environments will spread the frequency from fc-fd to fc+fd, which is the basics of Jakes U-shape frequency model. The receiver will try to lock the LO to fc and baseband processor engineers will focus on the frequency spread from -fd to fd.