PRN133 ranging now useable for SoL, at non precision approach level

On 2011-11-01 17:47 , HIPAR wrote:
On Nov 1, 9:25 am, Alan
wrote:

...

L5 isn't exactly usable yet with all of 2 sats in orbit. Will be a long
wait before any advance with any system. For simplicity sake (a good
thing in avionics) mixing GPS with EGNOS in a system won't be seen in
avionics for quite a while yet.

--

I couldn't agree more that we need simplicity .. too many
constellations transmitting signals that are compatible only by the
definition of not interfering with each other. My head would be
spinning if I were tasked to perform a trade study defining the next
generation of avionics. But the GNSS community thinks this kind of
diversity is great so those geniuses like Marcelo (just joking) can
sort it out.

Would it have been nice if Galileo L5 and NAVSTAR L5 shared a common
ICD? Would it have been nice if there were a common L1 modernized
signal. That would be 'bound' the problem.

That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.

Regarding WDGPS, I really don't understand who actually controls
access to the system. If NASA operates the core system, what kind of
agreement does the US government have with Deere allowing them
exclusive commercial marketing rights under the Starfire trademark?
NASA/JPL doesn't say much about that.

No idea. But with the network of ground stations collecting the data
for GDGPS that data can be "sold" to J-D for further use. In that sense
JD depend on the network, but they package the data for Starfire (and to
finer resolution and accuracy than WAAS).

I looked over a few of the easier to read references concerning the
JPL system. This one addresses the expected performance for a GDGPS
corrected C/A code system:

http://www.gdgps.net/system-desc/pap...leFreqCorr.pdf

Receiving L1 only, I'd say it might provide WAAS grade performance.
Getting back to simplicity, the need to receive the corrections from
another satellite system would complicate the actual operations.
Along with the other issues discussed, WAAS remains a more practical
system for airplanes.

But EGNOS provides SBAS for both GPS and GLONASS...

--
gmail originated posts filtered due to spam.

On Nov 1, 2:47*pm, HIPAR wrote:

Regarding *WDGPS, *I really don't understand who actually controls
access to the system. *If NASA operates the core system, what kind of
agreement does the US government have with Deere allowing them
exclusive commercial marketing rights under the Starfire trademark?
NASA/JPL doesn't say much about that.

--- *CHAS

JPL alludes to commercial opportunities on their site, but don't give
specifics of licensing, fees, etc. Clicking on the "Customer Portal"
tab yields a certificate error message.

http://www.gdgps.net/system-desc/network.html

"The core of the GDGPS network is the NASA Global GPS Network (GGN), a
JPL-owned and operated network of roughly 70 geodetic-quality, dual
frequency receivers, distributed globally. Additional real-time sites
are contributed by a variety of U.S. and international partner
organizations. The result is the world's largest real-time GPS
tracking network, with more than 100 global sites (as of October
2006). All these sites stream their GPS measurements at 1 Hz to the
GDGPS Operation Centers (GOCs), where it is processed and analyzed in
real-time.

.. . . The GDGPS system is proud to count 4 national timing
laboratories among it contributing network partners. In particular,
the United States Naval Observatory (USNO) contributes two monitoring
sites driven by its Master Clock, allowing the GDGPS System to provide
its global users the most accurate real-time realization of USNO UTC.

.. . . We continue to expand our network, and welcome contributions
from interested organizations. We offer our network partners a variety
of benefits, including real-time positioning, timing, and
environmental monitoring, as well as data archiving and data
distribution through the NASA CDDIS facility."

http://www.gdgps.net/applications/index.html

"The GDGPS System produces differential corrections to the GPS
broadcast ephemeris with unparelleled accuracy and seamless global
validity. Various GDGPS technology components and data products are
being used by nearly all of the providers of premium global
differential corrections. The underlying software and algorithms are
being used by the FAA's Wide Area Augmentation System (WAAS), and its
Japanese counterpart, MSAS. "

http://www.gdgps.net/system-desc/references.html

Kevin Dixon, "StarFi A Global SBAS for Sub-Decimeter Precise Point
Positioning," ION GNSS 2006, Fort Worth, TX, September 2006,
http://www.gdgps.net/system-desc/papers/starfire.pdf

"The central processing hubs are based upon a version of the Real Time
GIPSY (RTG) suite, originally developed by the Jet Propulsion
Laboratory for precise real time orbit and clock determination of
GNSS. This has been refined to optimize positioning accuracy of NavCom
developed GNSS hardware.

.. . . The StarFire correction stream consists of the RTG generated
GNSS precise orbit and clock values differenced with respect to the
GNSS broadcast ephemeris.

, , , The RTG code is the latest state of the art implementation from
NASA's Jet Propulsion Laboratory."

More on RTG:

http://gipsy.jpl.nasa.gov/igdg/system/od/index.html

Some older publications imply that JPL and NavCom (or John Deere --
think they were spun off once, then brought back) may have had a joint
venture at one time.

The world map in this 2004 paper shows roughly equal number of JPL and
NavCom reference stations (NavCom in green, the John Deere color):

http://www.gmat.unsw.edu.au/wang/jgp...2/v3n12p19.pdf

This 2001 NavCom press release (reprinted in GPS World) describes a
joint ventu

http://www.navcomtech.com/News/PressReleases.cfm?id=8

On Nov 1, 3:21*pm, Alan Browne
wrote:

That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.


There are actually around 500 balanced (roughly equal number of 0s and
1s) Gold codes in GPS, if you ignore the 2-tap mechanization shown in
the ICD. The 4-asterisk footnote was added to Tables 3-I and 3-II a
few years ago when the first list of expanded codes was published in
the ICD ( " **** The two-tap coder utilized here is only an example
implementation that generates a limited set of valid C/A codes.").

With zero Doppler difference between two PRNs, any pair of the 500 or
so balanced Gold codes would have the same peak cross-correlations.
The cross-correlation peak comes up a few dB at some Doppler
differences. Gold's 1967 papers showed that the zero-Doppler peak for
a 10-bit code is limited to 20*log(65/1023) of 20*log(63/1023), where
63 and 65 represent the excess of bit by bit agreements over
disagreements (or vice versa) between the two codes at a given time
offset. His papers also showed the probability of occurrence.

The log works out to about -24 dB. But user antenna gain, as well as
differences in satellite power, can bring the peak up more.

The bigger issue is that the broadcast ephemeris doesn't include the
PRN number. It's just assumed by the receiver that it's tracking the
one it wanted. But that's not a deficiency of the Gold codes, nor
really a deficiency of the original signal design, which is quite
elegant -- an awful lot of information packed into very few bits.

All just a quibble -- you're right that the newer signals with longer
codes will work better.

On Nov 1, 7:54*pm, "Ed M." wrote:
On Nov 1, 3:21*pm, Alan Browne
wrote:



That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.

There are actually around 500 balanced (roughly equal number of 0s and
1s) Gold codes in GPS, if you ignore the 2-tap mechanization shown in
the ICD. *The 4-asterisk footnote was added to Tables 3-I and 3-II a
few years ago when the first list of expanded codes was published in
the ICD ( " **** The two-tap coder utilized here is only an example
implementation that generates a limited set of valid C/A codes.").

With zero Doppler difference between two PRNs, any pair of the 500 or
so balanced Gold codes would have the same peak cross-correlations.
The cross-correlation peak comes up a few dB at some Doppler
differences. *Gold's 1967 papers showed that the zero-Doppler peak for
a 10-bit code is limited to 20*log(65/1023) of 20*log(63/1023), where
63 and 65 represent the excess of bit by bit agreements over
disagreements (or vice versa) between the two codes at a given time
offset. *His papers also showed the probability of occurrence.

The log works out to about -24 dB. *But user antenna gain, as well as
differences in satellite power, can bring the peak up more.

The bigger issue is that the broadcast ephemeris doesn't include the
PRN number. *It's just assumed by the receiver that it's tracking the
one it wanted. *But that's not a deficiency of the Gold codes, nor
really a deficiency of the original signal design, which is quite
elegant -- an awful lot of information packed into very few bits.

All just a quibble -- you're right that the newer signals with longer
codes will work better.


I suppose if cross correlation becomes a problem, it can be mitigated
by placing the conflicting satellites in antipodal positions.


Current L5 PRN assignments:

http://www.losangeles.af.mil/shared/...070530-041.pdf

And L1 C/A:

http://www.losangeles.af.mil/shared/...101124-042.pdf

Of course, 'modernized signal' would be the operative concept for a
universal open L1 signal conforming with a common ICD. Shame it will
never happen.

It's a 'minor miracle' that SBAS has been standardized.

http://www.elisanet.fi/master.naviga...S_Coverage.jpg

--- CHAS

On Nov 1, 10:58*pm, HIPAR wrote:
On Nov 1, 7:54*pm, "Ed M." wrote:









On Nov 1, 3:21*pm, Alan Browne
wrote:

That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.

There are actually around 500 balanced (roughly equal number of 0s and
1s) Gold codes in GPS, if you ignore the 2-tap mechanization shown in
the ICD. *The 4-asterisk footnote was added to Tables 3-I and 3-II a
few years ago when the first list of expanded codes was published in
the ICD ( " **** The two-tap coder utilized here is only an example
implementation that generates a limited set of valid C/A codes.").

With zero Doppler difference between two PRNs, any pair of the 500 or
so balanced Gold codes would have the same peak cross-correlations.
The cross-correlation peak comes up a few dB at some Doppler
differences. *Gold's 1967 papers showed that the zero-Doppler peak for
a 10-bit code is limited to 20*log(65/1023) of 20*log(63/1023), where
63 and 65 represent the excess of bit by bit agreements over
disagreements (or vice versa) between the two codes at a given time
offset. *His papers also showed the probability of occurrence.

The log works out to about -24 dB. *But user antenna gain, as well as
differences in satellite power, can bring the peak up more.

The bigger issue is that the broadcast ephemeris doesn't include the
PRN number. *It's just assumed by the receiver that it's tracking the
one it wanted. *But that's not a deficiency of the Gold codes, nor
really a deficiency of the original signal design, which is quite
elegant -- an awful lot of information packed into very few bits.

All just a quibble -- you're right that the newer signals with longer
codes will work better.

I suppose if cross correlation becomes a problem, it can be mitigated
by placing the conflicting satellites in antipodal positions.

Current L5 PRN assignments:

http://www.losangeles.af.mil/shared/...070530-041.pdf

And L1 C/A:

http://www.losangeles.af.mil/shared/...101124-042.pdf

Of course, *'modernized signal' would be the operative concept for a
universal open L1 signal conforming with a common ICD. *Shame it will
never happen.

It's a 'minor miracle' that SBAS has been standardized.

http://www.elisanet.fi/master.naviga...S_Coverage.jpg

--- *CHAS

One interesting question, military and L5 signals both use a 10 mega
chip/sec signal, clearly better for jamming tolerance and ability to
acquire a signal under challenging conditions, but does 50 10 mega
chip/sec signals being received at the same band perform better than
50 1 mega chip/sec signals ?

As far as the common ICD, I believe this will take 20-30 years, and
will happen when L1 C/A gets retired and replaced by a brand new
signal, perhaps a 10 mega chip/sec signal, broadcast by all GPS
satellites (one can hope...).

Also there has been talk about a signal in the 5GHz band, maybe one
day we could have a sane, normalized signal there too. A 5GHz signal
would improve IONO corrections hugely, due to the multi GHz jump in
frequency.

SBAS has a common ICD due to FAA having no proprietary interest in the
signal, quite on the opposite, by making SBAS signal structure a
worldwide standard benefits the early manufacturers (Americans) to
come to market, and due to the obvious requirement that an American
aircraft needs to be able to fly elsewhere in the world and use the
other SBAS systems, much like ILS/VOR/NDB/DME allow that today and
vice-versa. The requirements for revision control, documentation and
testing of software onboard aircraft navigation systems is 100%
assinine and extremely expensive. Talk about burying yourself in
paperwork.

But are the ICD differences between GPS/Galileo and QZSS that big ? It
seems to me that the differences are 100% software stuff, nothing to
do with acquiring and tracking the signal, just in the higher level
functions, like a few ifs in the higher level software of a receiver.
I have not read those ICDs, I'm really asking.

Marcelo Pacheco

On 2011-11-01 19:54 , Ed M. wrote:
On Nov 1, 3:21 pm, Alan
wrote:

That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.


There are actually around 500 balanced (roughly equal number of 0s and

To avoid x-correlation there are only 35 or so. Don't recall the
correct number.

--
gmail originated posts filtered due to spam.

Alan Browne wrote:
On 2011-11-01 19:54 , Ed M. wrote:
On Nov 1, 3:21 pm, Alan
wrote:

That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.


There are actually around 500 balanced (roughly equal number of 0s and

To avoid x-correlation there are only 35 or so. Don't recall the correct
number.

I have read a white paper which stated that the number of available Gold
codes was (afair) in the 60-70 range.

For randomly generated 1023-bit codes we should expect collisions around
32 (sqrt(1024)), but since the codes can be selected by hand, they can
get away with approximately twice as many without having problems with
cross-correlation, with or without doppler corrections.

Several people have noted that you could theoretically get twice as many
sats if you set them up in pairs on opposite side of the globe, but
since there's no explicit sat nr in the transmitted msg, this won't work
with any currently deployed gps receivers.

(In theory, as long as the pairs were not exactly opposite, you should
be able to determine which hemisphere was the correct one by looking at
the residual errors for each of them?)

Terje
--
- Terje.Mathisen at tmsw.no
"almost all programming can be viewed as an exercise in caching"

On 2011-11-02 09:43 , Terje Mathisen wrote:
Alan Browne wrote:
On 2011-11-01 19:54 , Ed M. wrote:
On Nov 1, 3:21 pm, Alan
wrote:

That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.


There are actually around 500 balanced (roughly equal number of 0s and

To avoid x-correlation there are only 35 or so. Don't recall the correct
number.

I have read a white paper which stated that the number of available Gold
codes was (afair) in the 60-70 range.

The number of "suitable codes" is 37. Again, x-correlation is the issue.

http://www.kowoma.de/en/gps/signals.htm as we discussed here some months
ago.

As I've stated in the past (and HIPAR states in this thread) one could
put satellites in opposition in the same plane and use the codes twice.

A quick scan (meaning I might have read it wrong) of ICD-200 data frames
doesn't appear to name the transmitting SV. - which would help
"unconfuse" the receiver (would need two almanacs per PRN). Acquisition
of PRN, eg: 23, would require knowledge of both almanacs and careful
attention to the ephemeris used if there is no other means of
identifying which is which. (Of course with reliable init position and
time, this would be unambiguous. For a cold start it wouldn't matter:
search a PRN and if it comes up it's one of the two). Problem is how to
store two almanacs per PRN, acquire and track without screwing up.
Could require an analysis of several satellite positions to
unambiguously select.

In brief, the PRN is the identifier for the satellite. (Almanac data
identifies the satellite for the data set (In the convoluted definition
of subframes 4 and 5 which carry a lot of almanac and other data ...
over a 12.5 minute cycle)).

http://www.navcen.uscg.gov/pubs/gps/...D200Cw1234.pdf

--
gmail originated posts filtered due to spam.

Up front, I apologize for nit-picking. I agree with all the posts
about modernized signals (L2C, L5, L1C). They have all been designed
to correct various problems experienced with C/A code.

Meanwhile, the number of C/A codes is being expanded to 209, or about
40% of the balanced C/A codes. Many or most of these may never
actually be broadcast, but they are defined.

From Spilker's paper in the Summer 1978 special issue of the ION
Journal (alias "Vol. I Red Book"):

Spilker, J. J., "SIGNAL STRUCTURE AND PERFORMANCE CHARACTERISTICS
(SPACE SEGMENT)", NAVIGATION, Vol. 25, No. 2, Summer 1978, pp.
121-146.
http://www.ion.org/search/view_abstr...?jp=j&idno=680

"Thus the advantage of the Gold codes is not simply a low cross-
correlation between all members of the family but that there are a
large number of codes all of similar good properties.

.. . . Fig. 2-13 shows the cumulative probability of various cross-
correlation interference levels for the GPS C/A code for various
doppler shifts from fd = 0 to + 5 kHz. Note that the 4 kHz doppler
gives the worst cross-correlation sidelobe over this range; however,
the other doppler shifts give similar results. These cumulative
averages are formed by averaging results for alI 1023 of the Gold
codes of period 1023 in the GPS family. All possible code time offsets
are considered for each doppler offset and ail possible pairs of codes
in this family. . . .

Peak Cross-correlation (any doppler shift) -21.6 dB
Peak Cross-correlation (zero doppler) -23.8 dB
Probability of worst case or near worst case cross-correlation 0.25
"


Couldn't find a free on-line copy of that paper, but you can browse
the Vol. I "Blue Book" via Google Books:

Global positioning system: theory and applications, Volume 1, Bradford
W. Parkinson and James J. Spilker (eds.),
AIAA, 1996, ISBN: 156347106X, 9781563471063

http://books.google.com/books?id=lvI1a5J_4ewC

pg. 99: "The number of balanced Gold codes is . . . 513 . . . ."

pg. 102, Table 9

The table shows that the cross-correlation of any pair of Gold codes
is 1/1023 with probability 3/4, 63/1023 with probability 1/8, and
65/1023 with probability 1/8. Taking 20*Log10 of those values gives
worst case cross-correlation of -23.8 dB, and best case of -30 dB.

Again, these theoretical cross-correlation peaks are valid only for
zero Doppler. As Spilker notes in his 1978 ION paper, the worst case
Doppler raises that peak about 2.2 dB.


IS-GPS-200E shows C/A code generation in Figs. 3-8 through 3-10, and
Table 3-I.

http://www.gps.gov/technical/icwg/

http://www.gps.gov/technical/icwg/IS-GPS-200E.pdf

The table and Fig. 3-10 show a "2-tap" mechanization, in which 2
stages of the G2 shift register are tapped and added modulo 2 to the
output of the G1 register. Since the G2 register has 10 stages, there
are 45 ways ("10 choose 2") to do this.

Table 3-1 shows only 36 unique tap pairs. PRNs 34 and 37 are
identical on C/A code (though unique on P-Y code).

The reason is that the other 9 tap pairs produce unbalanced codes,
i.e., the number of 0s and 1s differ by more than 1. Note also that
the selection of a pair of stages is equivalent to a time delay of the
G2 sequence. The newly defined C/A codes are defined on the basis of
G2 time delay, so that information has been added to Table 3-I for
consistency.

C/A codes can easily be generated in a spreadsheet, which shows that
the other 9 2-tap codes are unbalanced.

IS-GPS-200E defines 173 new C/A codes, for a total of 209 unique ones,
or 210 total.

"6.3.6.1 Additional C/A-code PRN sequences. The PRN C/A-code is
described in Section 3.2.1.3 and 36 legacy C/A-code sequences are
assigned by SV-ID number in Table 3-I. An additional set of 173 C/A-
code PRN sequences are selected and assigned with PRN numbers in this
section as shown in Table 6-I. Among the 173 additional sequences; PRN
numbers 38 through 63 are reserved for future GPS SVs; PRN numbers 64
through 119 are reserved for future Ground Based Augmentation System
(GBAS) and other augmentation systems; PRN numbers 120 through 158 are
reserved for SBAS; and PRN numbers 159 through 210 are reserved for
other Global Navigation Satellite System (GNSS) applications."

This information will evntually migrate from Sec. 6 to Sec. 3.

The implication is that someone has gone through the tedious effort of
examining roughly 125,000 (513 choose 2) pairs of codes for
undesirable cross-correlation properties, and selected the 173 "best"
in some sense for the PRN expansion.