Bookmark this page:
- Does-1ms-IQ-data-enough-for-PPT-output
- 05-04-2009
If you were Registered and logged in, you could reply and use other advanced thread options
I am verifying a software GPS receiver concept. I would like to know
if I can get PVT output by post-processing with only 1ms (0.001s) of I
and Q samples from ADC. The I and Q samples includes signals from 6
GPS satellites with elevation angle greater than 15 degrees. The
signals are collected from open sky. The ephemerise data will be
downloaded from IGS website.
Theoretically I thought 1ms is enough for the post-processing. Any
different opinion?
Thank you!
(see below). The in between answer is "it depends." Like, what are you doing
with the post-correlation data? If you immediately form the nonlinear
discriminants (e.g, delay lock loop), then you are going to have a
relatively high minimum C/N0 threshold (i.e., you'll lose lock easily). But,
you can test this...if you perform the correlation against a PRN you know is
visible (3 lines of Matlab code) you should see a recognizable peak. Try one
lower in the sky and it may get iffy.
There is a semi-classic book called IIRC "Software GPS Receivers" by
Cheng(?) that talks in detail about block processing approaches and whatnot.
I'm fairly sure it's available from Navtech GPS Supply (good place).
It's not too hard to string together multiple 1 millisecond integrations to
then form longer coherent and noncoherent sums (see book).
Good luck,
Marty
wrote:
>I am verifying a software GPS receiver concept. I would like to know
> > if I can get PVT output by post-processing with only 1ms (0.001s) of I
> > and Q samples from ADC. The I and Q samples includes signals from 6
> > GPS satellites with elevation angle greater than 15 degrees. The
> > signals are collected from open sky. The ephemerise data will be
> > downloaded from IGS website.
> > and Q samples from ADC. The I and Q samples includes signals from 6
> > GPS satellites with elevation angle greater than 15 degrees. The
> > signals are collected from open sky. The ephemerise data will be
> > downloaded from IGS website.
> The short answer is yes, that is enough. The long answer could fill a boo=
k
> (see below). The in between answer is "it depends." Like, what are you do=
ing
> with the post-correlation data? If you immediately form the nonlinear
> discriminants (e.g, delay lock loop), then you are going to have a
> relatively high minimum C/N0 threshold (i.e., you'll lose lock easily). B=
ut,
> discriminants (e.g, delay lock loop), then you are going to have a
> relatively high minimum C/N0 threshold (i.e., you'll lose lock easily). B=
> you can test this...if you perform the correlation against a PRN you know=
is
> visible (3 lines of Matlab code) you should see a recognizable peak. Try =
one
> lower in the sky and it may get iffy.
> There is a semi-classic book called IIRC "Software GPS Receivers" by
> Cheng(?) that talks in detail about block processing approaches and whatn=
ot.
> There is a semi-classic book called IIRC "Software GPS Receivers" by
> Cheng(?) that talks in detail about block processing approaches and whatn=
> I'm fairly sure it's available from Navtech GPS Supply (good place).
> It's not too hard to string together multiple 1 millisecond integrations =
to
> It's not too hard to string together multiple 1 millisecond integrations =
> then form longer coherent and noncoherent sums (see book).
> Good luck,
> Marty
> Good luck,
> Marty
Thank you very much for your help, Marty. I will test the CNo as you
have mentioned.
After a second thought, I think we will need 3ms at least, to avoid
the navigation bit turn-over. The navigation bit turn-over may happen
at any time, so we need to compare the results of 3 segments, 1ms
each, and choose the close two. What do you think?
As for the 1ms segment, the post-processing needs to search along both
the doppler and code phase, just as the real-time receiver does. I
believe it will generate a correlation peak.
Johnson
verifying a software GPS receiver concept. I would like to know
> > if I can get PVT output by post-processing with only 1ms (0.001s) of I
> > and Q samples from ADC. The I and Q samples includes signals from 6
> > GPS satellites with elevation angle greater than 15 degrees. The
> > signals are collected from open sky. The ephemerise data will be
> > downloaded from IGS website.
> > and Q samples from ADC. The I and Q samples includes signals from 6
> > GPS satellites with elevation angle greater than 15 degrees. The
> > signals are collected from open sky. The ephemerise data will be
> > downloaded from IGS website.
> The short answer is yes, that is enough. The long answer could fill a book
> (see below). The in between answer is "it depends." Like, what are you doing
> with the post-correlation data? If you immediately form the nonlinear
> discriminants (e.g, delay lock loop), then you are going to have a
> relatively high minimum C/N0 threshold (i.e., you'll lose lock easily). But,
> you can test this...if you perform the correlation against a PRN you know is
> visible (3 lines of Matlab code) you should see a recognizable peak. Try one
> lower in the sky and it may get iffy.
> (see below). The in between answer is "it depends." Like, what are you doing
> with the post-correlation data? If you immediately form the nonlinear
> discriminants (e.g, delay lock loop), then you are going to have a
> relatively high minimum C/N0 threshold (i.e., you'll lose lock easily). But,
> you can test this...if you perform the correlation against a PRN you know is
> visible (3 lines of Matlab code) you should see a recognizable peak. Try one
> lower in the sky and it may get iffy.
I'm missing something here :)
The C/A code has 1023 chips in one millisecond
How can you correlate if you have only one I/Q sample every
millisecond? (sampling @1KHz)
Don't you need to sample (at least) at a frequency > 1MHz? (that is
the frequency of the C/A code)
.>Iam verifying a software GPS receiver concept. I would like to know
> > > if I can get PVT output by post-processing with only 1ms (0.001s) of =
I
> > > and Q samples from ADC. The I and Q samples includes signals from 6
> > > GPS satellites with elevation angle greater than 15 degrees. The
> > > signals are collected from open sky. The ephemerise data will be
> > > downloaded from IGS website.
> > > GPS satellites with elevation angle greater than 15 degrees. The
> > > signals are collected from open sky. The ephemerise data will be
> > > downloaded from IGS website.
> > The short answer is yes, that is enough. The long answer could fill a b=
ook
> > (see below). The in between answer is "it depends." Like, what are you =
doing
> > with the post-correlation data? If you immediately form the nonlinear
> > discriminants (e.g, delay lock loop), then you are going to have a
> > relatively high minimum C/N0 threshold (i.e., you'll lose lock easily).=
But,
> > discriminants (e.g, delay lock loop), then you are going to have a
> > relatively high minimum C/N0 threshold (i.e., you'll lose lock easily).=
> > you can test this...if you perform the correlation against a PRN you kn=
ow is
> > visible (3 lines of Matlab code) you should see a recognizable peak. Tr=
y one
> > lower in the sky and it may get iffy.
> I'm missing something here :)
> The C/A code has 1023 chips in one millisecond
> How can you correlate if you have only one I/Q sample every
> millisecond? (sampling @1KHz)
> Don't you need to sample (at least) at a frequency > 1MHz? =A0 (that is
> the frequency of the C/A code)- Hide quoted text -
> - Show quoted text -
> The C/A code has 1023 chips in one millisecond
> How can you correlate if you have only one I/Q sample every
> millisecond? (sampling @1KHz)
> Don't you need to sample (at least) at a frequency > 1MHz? =A0 (that is
> the frequency of the C/A code)- Hide quoted text -
> - Show quoted text -
Yes, I forgot to mention. The sampling rates must be 2 times of the
Nyquist frequency.
- USB Data/ Power Cable to Etrex
- Garmin GPS
- 2006-01-17
- Gis Gps data converter
- Global Positioning System
- 2005-10-19
- Garmin Data Fields
- Global Positioning System
- 2012-05-16
- GPS: Centroid vs. Precision Polygon Based data for Latitude & Longitude geopairs
- Global Positioning System
- 2011-12-21
- PND Market Data
- Satellite Navigation
- 2012-01-17
> if I can get PVT output by post-processing with only 1ms (0.001s) of I
> and Q samples from ADC. The I and Q samples includes signals from 6
> GPS satellites with elevation angle greater than 15 degrees. The
> signals are collected from open sky. The ephemerise data will be
> downloaded from IGS website.