Fr Robert Ballecer, SJ love the segment on Cognitive Radios (CR), but if I may offer a minor po...

[G+_John Mink]

Fr. Robert Ballecer, SJ love the segment on Cognitive Radios (CR), but if I may offer a minor point (hey, everyone is pedantic when it's their area of focus, no?) on Cognitive Radios, the processing isn't really the "hard" part...at least not directly.  It's really the analog side of things that limits most RF radios.

 

That's cause most of the processing is FFT and simple filtering/decoding/demodulating which all fits rather well onto an FPGA (meaning it runs in parallel well)  but even something as common and "low" radio frequency (RF) as 2.4 GHz can't be digitized directly (why? see below) ...but instead has to be stepped down to a lower frequency via analog components.

 

These analog components are the largest limitation, as they have all sorts of limitations that don't exist in digital realm.  I can go into those for sure...but it's well beyond the quick post I intended here!

 

 

 

 

Why can't 2.4GHz be digitized directly?....after all, your processor runs at 2.4 GHz (or faster)!

 

1) Yes, but does it run at 4.8 GHz (or 5 GHz to get some overhead)? Nyquist sampling (wikipedia has plenty of articles on this) means that to be able to recreate a signal you need to sample at twice the max frequency. so we're already at 5 GHz!

 

2) The way Analog-to-Digital Converters work is by guess and check (sad but true, I know) meaning they

 So if you're converting to a 10 bit number, you need 10 cycles (yes, there are 1024 posible values, 0-1023 but you need not try each one, rather step down through each bit, adjusting as necessary with a 'simple' analog comparator)

 

So now for 10 bit accuracy (which really isn't that great) you need an ADC which can do the comparison at 50 GHz (or 0.02 nanoseconds) not so easy!

 

And keep in mind 2.4 GHz is low... 10-20 GHz is common military band and they're looking at 60GHz for wireless HDMI!

[G+_John Mink]

Good catch Lee Ball!