[lug] Linux Journal article about software radio

[Nate Duehr]

Wayde's analysis of software radio is really good.

Some other comments I would add:

1. Currently the components to build in mass quantity still favor 
traditional radios.  Software radio is marginalized by its price.  This 
will continue to change over time, but right now the DSP's needed to do 
a large bandwidth software radio are still too expensive for much mass 
production.  Most of the early "software radios" you see today are 
actually hybrids... some of the filtering and other RF blocks in the 
block diagram are still done the old-fashioned hardware way.

2. Cognitive radio is VERY hotly debated.  One camp says they're the 
nirvana of radio -- radios that will figure out what frequencies and 
power levels to use automatically.  Another camp says that our 
traditional regulatory spectrum management will be thrown out the window 
and interference complaints (imagine a bunch of software radios ship 
with bad code in the DSP) will be a nightmare.

3. Like in many other areas of research, Linux and the whole GNU 
software radio project seem to have come about because Linux is open, 
and the people doing the research needed a common platform they could 
develop on easily.  The system is free, the toolchain is robust and 
mature, and generally -- I find that part really neat.  But...

4. As SDR matures in the future, software-defined radio will probably be 
done in embedded hardware.  Linux will become marginalized as 
manufacturers start hiding their specific software "secrets" inside the 
black boxes.

The math background required to really contribute is a large barrier to 
entry to this space.  And the math will be hidden in proprietary DSP 
code soon.  My opinion anyway. 

There's also a regulatory storm on the horizon when groups really start 
pushing for the cognitive radio systems -- and traditional trunking and 
cellular companies will probably want to protect their investments in 
infrastructure and will have a lot of clout with the FCC. 

(Anyone with a lot of money has clout with them these days it seems.  
Broadband over Powerlines is a horribly-engineered and very problematic 
technology, but the FCC is cowtowing to it thus far.  One commissioner 
going so far to say that it's "broadband nirvana".  Yeah, right... 
whatever.)

Nate Duehr, nate at natetech.com

J. Wayde Allen wrote:

>On Fri, 21 May 2004 robmohr at earthnet.net wrote:
>
>  
>
>>Is software radio just a parlor trick?  
>>    
>>
>
>In classical radio systems you'd build a tuner circuit followed by a
>circuit that could demodulate and amplify the signal.  The development of
>high speed digital samplers (A/D converters) has now made it possible to
>perform many of the radio functions mathematically in the computer.  
>
>The idea is that if you have a high speed broadband digitizer you could
>convert the signal picked up by an antenna into a digital data stream.  
>The digital data can then be operated on using digital filters, fast
>Fourier transforms (FFT's), etc. to realize a radio as an almost
>completely mathematical construct.  For a system like this most of the
>hardware exists at the radio's front-end leading up to the digitizer.
>
>  
>
>>Or does it have immediate applications?
>>    
>>
>
>There are immediate applications.  Think of it this way.  If you build a
>radio using hardware its capabilities are limited by the design of the
>hardware.  You'll get an AM radio or and FM radio for instance, but not
>both simultaneously.
>
>With a software defined radio how the radio works can be changed as fast
>as you can change the signal processing algorithm.  This can be very
>valuable if for example you have a radio system that needs to interact
>with a variety of other radios.  This can be ideal for a public safety
>or military radio system where you may need to talk to a cellular network,
>two-way police, fire department, or even spy on the enemy.  Let's see,
>there are number of examples but the one I turned up first is
><http://www.spectrumsignal.com/Wireless_systems/Signals_Intelligence.asp>.
>
>There is even a GNU Radio project in case you'd like to build your own
>Linux based system, check it out at
><http://www.gnu.org/software/gnuradio/>.
>
>  
>
>>I am guessing here, are cell phones more software than hardware or
>>equal parts of each?
>>    
>>
>
>Cell phones are a little of both.
>
>  
>
>>Pointless question?
>>    
>>
>
>Nope, it is a pretty darn hot topic these days.  The next level above
>software defined radio is the so-called "Cognitive
>Radio" <http://www.eetimes.com/showArticle.jhtml?articleID=18700443>.  These
>radios actively look at the radio environment and work out the parameters
>needed to communicate.
>
>- Wayde
>  (wallen at village.org)
>
>_______________________________________________
>Web Page:  http://lug.boulder.co.us
>Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
>Join us on IRC: lug.boulder.co.us port=6667 channel=#colug
>  
>