Prev: ADSP 21262 Booting
Next: A issue about DSP_fft16x16t of dsplib of ti
From: Eric Jacobsen on 3 Apr 2008 18:50
On Thu, 03 Apr 2008 17:32:27 -0500, "barry_b" <bbuternowsky(a)gmail.com>
wrote:

>
>According to Matlab for BER = 0.001
>- msk/bpsk/qpsk: Eb/No = 7 dB
>- 4-cpfsk h=0.5: Eb/No = 3 dB
>
>For 4 bps, carrier = 12 Hz:
>- msk: tones (11, 13 Hz); bw = 2 Hz
>- bpsk: bw = 4 Hz
>- qpsk: bw = 2 Hz
>- 4-cpfsk: tones (9, 11, 13, 15 Hz); bw = 6 Hz
>
>Since I have available bandwidth between 2 and 20 Hz, wouldn't 4-CPFSK
>would provide the best use of the available bandwidth, and the lowest
>EB/No?
>
>But, now it seems that we also want bit rates of 1 - 20 bps.
>
>For example for 12 bps, carrier = 12 Hz:
>- msk: tones (9, 15 Hz); bw = 6 Hz
>- bpsk: bw = 12 Hz
>- qpsk: bw = 6 Hz
>- 4-cpfsk: tones( 3, 9, 15 21 Hz); bw = 18 Hz
>
>Thus, it seems 4-cpfsk is out. Either msk or qpsk is probably a better
>bet.

Remember that when considering bandwidth you need to include the
effects of any filtering that you may need to do, e.g., Root Raised
Cosine. The occupied bandwidth will always be a bit more (or a lot
more for rectangular pulses) than the 3dB bandwidth, which is what you
seem to be using above.

Even so I don't think your conclusion will change from what you just
indicated. QPSK won't be much different from the BPSK you're using
now other than the 3dB reduction in link margin. If you also want to
increase range as you suggested earlier, you'll need to somehow make
up for this 3dB plus some, by either increasing the transmit power, or
adding more powerful coding, or increasing the gain of the antennas,
etc., etc.

Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org
From: Randy Yates on 3 Apr 2008 19:06
Eric Jacobsen <eric.jacobsen(a)ieee.org> writes:
> [...]
> QPSK won't be much different from the BPSK you're using now other than
> the 3dB reduction in link margin.

Huh? Why then does Proakis show a two-fold increase in spectral
efficiency for the same Eb/N0 for QPSK over BPSK?

Re: p.283, figure 5.2-17 of [proakiscomm].

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

--
% Randy Yates % "She has an IQ of 1001, she has a jumpsuit
%% Fuquay-Varina, NC % on, and she's also a telephone."
%%% 919-577-9882 %
%%%% <yates(a)ieee.org> % 'Yours Truly, 2095', *Time*, ELO
http://www.digitalsignallabs.com
From: Vladimir Vassilevsky on 3 Apr 2008 19:26


barry_b wrote:

> According to Matlab for BER = 0.001
> - msk/bpsk/qpsk: Eb/No = 7 dB
> - 4-cpfsk h=0.5: Eb/No = 3 dB

This doesn't look right. QPSK/BPSK should perform at Eb/No ~ 7dB, 4-FSK
slightly worse than that, and for the MSK it should be about 9dB.

> For 4 bps, carrier = 12 Hz:
> - msk: tones (11, 13 Hz); bw = 2 Hz
> - bpsk: bw = 4 Hz
> - qpsk: bw = 2 Hz
> - 4-cpfsk: tones (9, 11, 13, 15 Hz); bw = 6 Hz
>
> Since I have available bandwidth between 2 and 20 Hz, wouldn't 4-CPFSK
> would provide the best use of the available bandwidth, and the lowest
> EB/No?

Lowest Eb/No -> maximize the number of states and the bandwidth.

However, there is the other important consideration: the carrier and
symbol sync accuracy. The more advanced is the signal, the more
difficult is maintaining the accurate lock. Hence the performance gets
limited by the lock accuracy rather then Eb/No.

>
> But, now it seems that we also want bit rates of 1 - 20 bps.
>
> For example for 12 bps, carrier = 12 Hz:
> - msk: tones (9, 15 Hz); bw = 6 Hz
> - bpsk: bw = 12 Hz
> - qpsk: bw = 6 Hz
> - 4-cpfsk: tones( 3, 9, 15 21 Hz); bw = 18 Hz
>
> Thus, it seems 4-cpfsk is out. Either msk or qpsk is probably a better
> bet.
>
>>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.

I give you an advice: avoid using the words that you don't understand
and find someone who has a clue.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

From: Randy Yates on 3 Apr 2008 19:30
Randy Yates <yates(a)ieee.org> writes:

> Eric Jacobsen <eric.jacobsen(a)ieee.org> writes:
>> [...]
>> QPSK won't be much different from the BPSK you're using now other than
>> the 3dB reduction in link margin.
>
> Huh? Why then does Proakis show a two-fold increase in spectral
> efficiency for the same Eb/N0 for QPSK over BPSK?

I see. Ignoring the symbol error rate inconsistency for a
moment, QPSK at the same Eb/N0 as BPSK requires 3 dB more tx power.

Duh. But I still have a problem with ser vs ber.
--
% Randy Yates % "...the answer lies within your soul
%% Fuquay-Varina, NC % 'cause no one knows which side
%%% 919-577-9882 % the coin will fall."
%%%% <yates(a)ieee.org> % 'Big Wheels', *Out of the Blue*, ELO
http://www.digitalsignallabs.com
From: barry_b on 4 Apr 2008 13:44
Ouch! Let's not get personal.

>
>I give you an advice: avoid using the words that you don't understand
>and find someone who has a clue.
>
>
>Vladimir Vassilevsky
>DSP and Mixed Signal Design Consultant
>http://www.abvolt.com
>
>
First  |  Prev  | 
Pages: 1 2 3 4 5
Prev: ADSP 21262 Booting
Next: A issue about DSP_fft16x16t of dsplib of ti