Hi David
Oh yeah, I agree with you- OFDM, like what you have done for Free DV,
is certainly the best mode for performance.
On the computationally constrained platform, the SM1000 is fantastic.
The TD eq single carrier mode has to work hard. The OFDM modm gets its
multipath tolerance for free.
So the OFDM modem is a obvious choice if you have the peak power to spare.
If you have the computational power available and the peak power
available, OFDM is great.
Consider though, that if you are at 10dB peak to average , that is 10W
RMS for a 100W amplifier. Utilization of the 100W PA is not good !
If using a (near) constant envelope, single carrier mode, you can use
all the 100W RMS from the amplifier. 10dB ahead is alot.
Of course, many of our HF paths are NOT SNR limited, and you can have
super high SNR signals and that will not help you with severe multiplath
delay spread exceeding the symbol rate.
The killer for the time domain equalized single carrier is the time
spent equalizing the channel.
Some systems will use a distributed "training sequence' others will not
use a dedicated sequence, but will use a blind equaliser (and there are
also hgihly computationaly expensive AI techniques abound)
The most simple method using a "training sequence" must be at least as
long as the channel time, and that can cost alot of the payload.
and with OFDM of course all you need is guard interval, and enough ECC
bits to survive the lots carriers bins.
Successful techniques use a 2d equaliser, and interpolate the channel
between the training channel sequence points.
So, the TD equalised single carrier is good, but with significant
caveats , and likely computational complexity for the time and
frequency interpolating equalizer.
The OFDM modem is simple, however, adding in the PAPR reduction to make
better UTILIZATION of the 100W PA -I think that is where the
development would be beneficial . And a good project yes- finding the
lowest computational cost on the constrained platform for effective
PAPR reduction.
-glen
On 18/06/2020 18:14, David Rowe wrote:
Hi Glen,
it has the least chance of being set up incorrectly and generating
substantial wideband interference.
Plenty of people run FreeDV cleanly.
I've seen splatter in a few cases with over driven FreeDV, however on my
radio I couldn't make it happen. I guess the ALC was kicking in on the
peaks, preventing any clipping and splatter. YMMV.
The reason PAPR reduction is not in the standard specification for DVB
and DRM is that PAPR reduction was very much a 2nd thought, as DVB and
DRM came around the beginning of OFDM in widespread use (apart from DAB) .
Well also in DVB/DRM are broadcast modes - there is only 1 Tx for every
million receivers, so it's OK to have a single pricey super linear PA,
for cheap and reliable operation of the link.
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2
