choosing the modulation technique

"barry_b" <bbuternowsky@gmail.com> writes:

> The channel frequency band is fixed at 2-20 Hz (maybe up to 30 Hz). The
> transmitter works in this range and cannot be changed. We have a working
> system that uses bpsk at 12 Hz (bit rate is 4 bps). I am looking into
> improving the performance of the system, thus I am wondering if other
> modulation techniques (msk, 4-cpfsk, etc) may provide a better error rate.
> The system works outdoors for up to 2 km.

Barry,

For bandwidth-limited systems, you may want to investigate Trellis-coded
modulation. See, e.g., [proakiscomm].

--Randy

@BOOK{proakiscomm,
  title = "{Digital Communications}",
  author = "John~G.~Proakis",
  publisher = "McGraw-Hill",
  edition = "fourth",
  year = "2001"}

-- 
%  Randy Yates                  % "With time with what you've learned, 
%% Fuquay-Varina, NC            %  they'll kiss the ground you walk 
%%% 919-577-9882                %  upon."
%%%% <yates@ieee.org>           % '21st Century Man', *Time*, ELO
http://www.digitalsignallabs.com

On Tue, 01 Apr 2008 22:43:41 -0500, "barry_b" <bbuternowsky@gmail.com>
wrote:

>The channel frequency band is fixed at 2-20 Hz (maybe up to 30 Hz). The
>transmitter works in this range and cannot be changed. We have a working
>system that uses bpsk at 12 Hz (bit rate is 4 bps). I am looking into
>improving the performance of the system, thus I am wondering if other
>modulation techniques (msk, 4-cpfsk, etc) may provide a better error rate.
>The system works outdoors for up to 2 km.

Is that your RF frequency or just the signal bandwidth?   If that's
the rf frequency your antenna is kinda big, no?  That's what the
military used to use for submarine communication since it penetrates
sea water reasonably well. 

FWIW, PSK/QAM provides the best power efficiency among common
modulation types, which is why they're used so often.  If you have
BPSK working and the phase noise is manageable, then bumping up to
QPSK shouldn't be too big of an issue if you have 3dB of link margin
to use up.   Even if you don't, putting coding on top of it when you
move from BPSK to QPSK will help, but your packets are so small that
you're not going to be able to get a lot of coding gain.

Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org

<snipped.>
> I'll second Vladimir's input on the noise, and also that phase noise
> becomes very problematic at such low bit rates.   So phase-modulated
> signals may not be the best choice for such low rates.
>
<snipped>

Phase noise is problematic at low bit rates?  How are the two
related?  Not enough pilot symbols available to estimate the phase?
What else am I missing?

col

On Apr 1, 3:38&#4294967295;pm, "barry_b" <bbuternow...@gmail.com> wrote:
> Can I get some feedback on my thought process below for choosing a
> modulation technique?
>
> I am in the process of choosing the modulation and coding for the
> following system:
> - frequency band: 2 - 20 Hz
> - bit rate: configurable 2 - 16 bps
> - transmitter: battery powered
> - receiver: PC based (i.e. lots of processing power)
> - channel: some burst noise
> - adjacent signals/channels: none
> - traffic: 1 direction, single user, bursty, asynchronous
> - data in payload: 1 to 150 bits
>
> I was considering either bpsk, qpsk, msk or 4-cpfsk. The transmitter can
> accomodate psk or fsk.
>
> I ranked the modulations as follows:
> - bspk/qpsk: good error rate performance curve
> - msk: smaller bandwith and same BER as bpsk; thus I can use a narrower
> filter and get better performance
> - 4-cpfsk (mod index = 0.5): better BER curve than bpsk and msk (according
> to Matlab)
>
> Given that 4-cpfsk has a better BER curve, does it make sense to use
> 4-CPFSK even for low bit rates? I.e which
> is better:
> 1) bspk/qpsk
> 2) msk
> 3) 4-cpfsk (mod index = 0.5)
> 4) bspk with spreading
> 5) msk with spreading
>
> From what I know, spreading doesn't provide any coding gain, but only
> provides immunity to frequency drop-outs
> frequency jamming, thus 3) should be the best. Correct?
>
> Given the short payload, I don't think convolutional coding is applicable.
> I will be using a reed-solomon
> code.
>
> In the receiver, I will be implementing a coherent detector.
>
> thanks,
> barry


Barry,

I hope you realize your band is on top of a couple of the Earth's
Schumann resonance modes. This is okay if you wish to detect lightning
anywhere on the Earth's surface.

Clay

On Apr 1, 11:43&#4294967295;pm, "barry_b" <bbuternow...@gmail.com> wrote:
> The channel frequency band is fixed at 2-20 Hz (maybe up to 30 Hz). The
> transmitter works in this range and cannot be changed. We have a working
> system that uses bpsk at 12 Hz (bit rate is 4 bps). I am looking into
> improving the performance of the system, thus I am wondering if other
> modulation techniques (msk, 4-cpfsk, etc) may provide a better error rate.
> The system works outdoors for up to 2 km.
>


You have to define IMPROVE..

Do you want faster throughput over the same distance...

or the same throughput over a longer distance

These are opposite requirments that will impact the choice of
modulation.

Mark

On Wed, 2 Apr 2008 07:28:01 -0700 (PDT), clay@claysturner.com wrote:

>On Apr 1, 3:38&#4294967295;pm, "barry_b" <bbuternow...@gmail.com> wrote:
>> Can I get some feedback on my thought process below for choosing a
>> modulation technique?
>>
>> I am in the process of choosing the modulation and coding for the
>> following system:
>> - frequency band: 2 - 20 Hz
>> - bit rate: configurable 2 - 16 bps
>> - transmitter: battery powered
>> - receiver: PC based (i.e. lots of processing power)
>> - channel: some burst noise
>> - adjacent signals/channels: none
>> - traffic: 1 direction, single user, bursty, asynchronous
>> - data in payload: 1 to 150 bits
>>
>> I was considering either bpsk, qpsk, msk or 4-cpfsk. The transmitter can
>> accomodate psk or fsk.
>>
>> I ranked the modulations as follows:
>> - bspk/qpsk: good error rate performance curve
>> - msk: smaller bandwith and same BER as bpsk; thus I can use a narrower
>> filter and get better performance
>> - 4-cpfsk (mod index = 0.5): better BER curve than bpsk and msk (according
>> to Matlab)
>>
>> Given that 4-cpfsk has a better BER curve, does it make sense to use
>> 4-CPFSK even for low bit rates? I.e which
>> is better:
>> 1) bspk/qpsk
>> 2) msk
>> 3) 4-cpfsk (mod index = 0.5)
>> 4) bspk with spreading
>> 5) msk with spreading
>>
>> From what I know, spreading doesn't provide any coding gain, but only
>> provides immunity to frequency drop-outs
>> frequency jamming, thus 3) should be the best. Correct?
>>
>> Given the short payload, I don't think convolutional coding is applicable.
>> I will be using a reed-solomon
>> code.
>>
>> In the receiver, I will be implementing a coherent detector.
>>
>> thanks,
>> barry
>
>
>Barry,
>
>I hope you realize your band is on top of a couple of the Earth's
>Schumann resonance modes. This is okay if you wish to detect lightning
>anywhere on the Earth's surface.
>
>Clay

Clay,

I was also curious about who else might use that band and what the
regulatory issues might be, but a quick scan of some relevant portions
of Part 15 suggests that the FCC may not care much about what happens
below 9kHz.  In any case, I couldn't find any relevant regulations for
ELF radiators.   If Barry's system interferes with a government system
he might hear about it, but other than that I don't know if it's
problematic other than the natural effects you mention.

Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org

On Wed, 2 Apr 2008 06:16:03 -0700 (PDT), cb135@hotmail.com wrote:

><snipped.>
>> I'll second Vladimir's input on the noise, and also that phase noise
>> becomes very problematic at such low bit rates.   So phase-modulated
>> signals may not be the best choice for such low rates.
>>
><snipped>
>
>Phase noise is problematic at low bit rates?  How are the two
>related?  Not enough pilot symbols available to estimate the phase?
>What else am I missing?
>
>col

As Vladimir mentioned, phase noise is relatable to the 1/f noise, or
if you want to look at it another way, the noise that is very close-in
to the reference oscillator.   This is easily seen on a spectrum
analyzer and a very common way of quantifying phase noise is to just
measure the noise level close-in to the oscillator, typically within
1kHz or so.

If the loop bandwidth of the carrier phase PLL is wider than that it
has no problem tracking out the phase noise, since it's in-band wrt to
the PLL.   Obviously, the carrier recovery PLL loop bandwidth has to
be much less than the symbol rate or it won't be able to lock onto the
symbols.   So, as the symbol rate goes down the 1/f noise, or the
oscillator phase noise, becomes a bigger and bigger part of the PLL
bandwidth, and, therefore, harder to track out.

Symbol rates of a few bps are problematic for that reason.   I'd think
it would require a different way of looking at the problem.

Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org

True. I'm looking to improve the distance and keep the bit rate the same or
higher. 

Can trellis coding be used with 4-CPFSK, or do I have to use PSK? (I just
started reviewing "Intro to TCM".)

I haven't seen anyone comment on 4-CPFSK which according to Matlab seems
to have a better BER vs Eb/No curve than MSK/BPSK. Could I not improve the
system by using 4-CPFSK?

Frequency band is below 30 Hz (e.g. for bpsk carrier frequency is at 12
Hz). We sample at 1000 samples/sec, downcovert to baseband, and decode.
There is various filtering as well.


>On Apr 1, 11:43=A0pm, "barry_b" <bbuternow...@gmail.com> wrote:
>> The channel frequency band is fixed at 2-20 Hz (maybe up to 30 Hz).
The
>> transmitter works in this range and cannot be changed. We have a
working
>> system that uses bpsk at 12 Hz (bit rate is 4 bps). I am looking into
>> improving the performance of the system, thus I am wondering if other
>> modulation techniques (msk, 4-cpfsk, etc) may provide a better error
rate.=
>
>> The system works outdoors for up to 2 km.
>>
>
>
>You have to define IMPROVE..
>
>Do you want faster throughput over the same distance...
>
>or the same throughput over a longer distance
>
>These are opposite requirments that will impact the choice of
>modulation.
>
>Mark
>
>
>
>

On Wed, 02 Apr 2008 11:59:44 -0500, "barry_b" <bbuternowsky@gmail.com>
wrote:

>True. I'm looking to improve the distance and keep the bit rate the same or
>higher. 
>
>Can trellis coding be used with 4-CPFSK, or do I have to use PSK? (I just
>started reviewing "Intro to TCM".)

You might be able to use trellis coding with 4-CPFSK, sort of, but I
don't think it'll be worth the effort.  As mentioned previously,
you're not going to be able to get much gain from channel coding since
your packets are so small.  If you can tolerate very low-rate codes
you can probably make it work, but your packet transmission time will
increase substantially at the rates you're using.  

Increasing range via coding also requires that the receiver have
adequate sensitivity and detection thresholds to exploit the
improvement, so that's a consideration as well.

>I haven't seen anyone comment on 4-CPFSK which according to Matlab seems
>to have a better BER vs Eb/No curve than MSK/BPSK. Could I not improve the
>system by using 4-CPFSK?

I don't have much experience with CPFSK, it's often avoided since it's
not bandwidth efficient.  Are your sims showing the same bandwidth
occupation for both cases?   The filtering applied for each, pulse
shaping for BPSK and any phase filtering for CPFSK, will be important
in considering the bandwidth requirements.   

>Frequency band is below 30 Hz (e.g. for bpsk carrier frequency is at 12
>Hz). We sample at 1000 samples/sec, downcovert to baseband, and decode.
>There is various filtering as well.
>
Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org

On Apr 2, 12:26&#4294967295;pm, Eric Jacobsen <eric.jacob...@ieee.org> wrote:
> On Wed, 2 Apr 2008 07:28:01 -0700 (PDT), c...@claysturner.com wrote:
> >On Apr 1, 3:38&#4294967295;pm, "barry_b" <bbuternow...@gmail.com> wrote:
> >> Can I get some feedback on my thought process below for choosing a
> >> modulation technique?
>
> >> I am in the process of choosing the modulation and coding for the
> >> following system:
> >> - frequency band: 2 - 20 Hz
> >> - bit rate: configurable 2 - 16 bps
> >> - transmitter: battery powered
> >> - receiver: PC based (i.e. lots of processing power)
> >> - channel: some burst noise
> >> - adjacent signals/channels: none
> >> - traffic: 1 direction, single user, bursty, asynchronous
> >> - data in payload: 1 to 150 bits
>
> >> I was considering either bpsk, qpsk, msk or 4-cpfsk. The transmitter can
> >> accomodate psk or fsk.
>
> >> I ranked the modulations as follows:
> >> - bspk/qpsk: good error rate performance curve
> >> - msk: smaller bandwith and same BER as bpsk; thus I can use a narrower
> >> filter and get better performance
> >> - 4-cpfsk (mod index = 0.5): better BER curve than bpsk and msk (according
> >> to Matlab)
>
> >> Given that 4-cpfsk has a better BER curve, does it make sense to use
> >> 4-CPFSK even for low bit rates? I.e which
> >> is better:
> >> 1) bspk/qpsk
> >> 2) msk
> >> 3) 4-cpfsk (mod index = 0.5)
> >> 4) bspk with spreading
> >> 5) msk with spreading
>
> >> From what I know, spreading doesn't provide any coding gain, but only
> >> provides immunity to frequency drop-outs
> >> frequency jamming, thus 3) should be the best. Correct?
>
> >> Given the short payload, I don't think convolutional coding is applicable.
> >> I will be using a reed-solomon
> >> code.
>
> >> In the receiver, I will be implementing a coherent detector.
>
> >> thanks,
> >> barry
>
> >Barry,
>
> >I hope you realize your band is on top of a couple of the Earth's
> >Schumann resonance modes. This is okay if you wish to detect lightning
> >anywhere on the Earth's surface.
>
> >Clay
>
> Clay,
>
> I was also curious about who else might use that band and what the
> regulatory issues might be, but a quick scan of some relevant portions
> of Part 15 suggests that the FCC may not care much about what happens
> below 9kHz. &#4294967295;In any case, I couldn't find any relevant regulations for
> ELF radiators. &#4294967295; If Barry's system interferes with a government system
> he might hear about it, but other than that I don't know if it's
> problematic other than the natural effects you mention.
>
> Eric Jacobsen
> Minister of Algorithms
> Abineau Communicationshttp://www.ericjacobsen.org- Hide quoted text -
>
> - Show quoted text -

Hello Eric,

I don't think there is much going on in the way of ELF transmissions
these days. I think the Navy turned off the site in the midwest. I
don't the status of the site in Australia.

My recent readings were based on a group at MIT who were doing global
lightning detection and correlating that with the Earth's temperature.
Apparently atmospheric temperature and the number of lightning strikes
per second have a very strong positive correlation. So the MIT group
is looking at other ways to verify global warming.

This as we all know is a very complicated phenomenon. Scientists were
recently suprised to learn that over the last 4 years the oceans have
cooled off. This is from a survey involving 3000 temperature taking
robots travelling in the oceans' currents.

Clay