Hi Bernard-
First, let me say that I have read all of the posts in your profile (will probably get into LOR at some time) and am even more duly impressed with both your knowledge and your geniality than before. (Definitely a model for some of us "zoo" animals). I wish I could read them further back than SI saves them, as it would be time well spent, IMO.
<..at 38GHz, the signal to noise ratio (SNR) is significantly affected by rain or in the case of paths affected by foliage, wind conditions. So the signal to noise ratio changes with wheather conditions. Thus plugging in measured SNR figures in Shannon's formula would be rather meaningless...>
What about ranges? In ideal real world conditions (not a vacuum), and at the edges? LOS is critical, so let's avoid foilage or other LOS obstruction considerations. But what about worst case precipitation? And since power increases are supposed to be able to handle these, we'll increase both the signal and noise power spectrums to accomodate precipitation conditions. That should define approx Shannon limits for 38GHz.
<..Note also that within a cell, under LOS conditions, attenuation increases as the square of the distance to the receiver. So capacity is distance dependent...>
Again, perhaps some ranges and averages could be determined.
<..I believe however that OFDM schemes have been implemented for some satellites in the Ku band (around 12-18GHz), so that OFDM may not be totally far out for LMDS. Another idea which has been tested in Japan
consists in using adaptive modulation patterns, where depending on the prevailing SNR conditions, the modulation pattern is varied between QAM (requires high SNR) and QPSK (low SNR)...>
Very interesting. Thanks! I searched INSPEC for B. Levy. Is that how you publish? I was looking for your academic focus, thinking there is likely "gold in those hills" (above), and wondering if you are positioning.
<..The need for high frequency circuits comes in from the fact that if you are processing a signal with a 100MHz bandwidth..>
I was thinking the GHz range is what required this. So 100MHz channels
still require higher freq circuits.
<..and need to perform an A/D coversion, the A/D converter will typically need to operate at about 1GHz to achieve a sufficient bit accuracy...>
10 times is ideal Nyquist, but couldn't you get away with say, 600MHz?
What is the approx cutoff for GaAs technology?
<.. Note that several non GaAs technologies are just around the corner: SiGe and SOI...>
I've read a little about SiGe, but wasn't familiar with SOI.
(here are some articles I just pulled up on SOI (silicon-on-insulator):
techweb.com
techweb.com
techweb.com
And did the following address any of your P-P concerns?
Message 3901941
Thanks again-
Steve |
