 |
 |
|
|
|
|
|
|
||||||||||||||||||
|
|
|||||||||||||||||||
|
|
|||||||||||||||||||
|
|||||
 Previous: JY1's North Carolina Connection |  Next: H&K LiteHelpful hints from other amateurs |
By Robert Gonsett, W6VR
March 15, 1999
Aircraft pilots and boaters alike use GPS receivers to determine their X,Y positions on the face of the earth. Problem is, positioning errors of 100 feet or more are not uncommon. If greater accuracy could be achieved, GPS could be used for precision navigation purposes. Here's a concept that uses Amateur Radio to derive the necessary correction data for a conventional GPS.
This is a story of one Amateur Radio operator's ongoing efforts to improve the accuracy of conventional GPS by building something called a DGPS system. The "D" stands for "differential," and it refers to the kind of correction data that are employed. DGPS already is used in commercial circles, but this is the first time that I have seen such a system constructed with low-cost Amateur Radio gear.
DGPS consists of a reference station and one or more rover
stations. The reference station is stationary. It observes
positioning errors in the incoming GPS satellite signals, and
transmits correction factors (called differential factors)
to the rovers. The rovers then use those factors to improve the
accuracy of both latitude and longitude readings.
The box marked WA6ZUH is the reference station. The other marked WB6ZUH (the call sign of Pete Prossen's XYL) is the rover. Each consists of an H-T, a TNC, and a battery pack. The reference station has the Motorola Oncore GPS receiver and control circuits in a die cast box. Prossen says the rover can return corrected position readings. Both the reference station and the rover can transmit and receive in half-duplex mode. "It is all subject to the normal arbitration procedures of packet," he says. "The reference station sends corrections at predetermined intervals (5 or 6 seconds or so). The rover receives these and via its TNC delivers them to the GPS III, which in turn incorporates the correction data into the NMEA position sentences that it can output." |
Pete Prossen, WA6ZUH, of Fallbrook, California, has constructed
a compact DGPS system which uses conventional Amateur Radio packet
gear. The system was first demonstrated last September during
the Palomar Amateur Radio Club's picnic in Escondido, California.
A ham frequency was used to send the packetized differential correction
data from the reference station to the rover (a Garmin GPS III).
Data bursts were spaced four seconds apart.
As a ham radio operator, I wanted to know something about the precision of Pete's DGPS system. It wasn't enough just to see it work. So, a small group of us hams picked four reference points within easy walking distance in the picnic area. Three of the points were picked at random while the fourth was chosen at the reference station itself. You'll see why in just a moment.
With the rover in the DGPS mode, we walked the four-point circuit twice and logged the rover's position reading at each location. (I wrote down the first coordinate pair that was displayed and did not attempt to average the readings.) Using the figures collected in the first loop as a baseline, I wanted to see how close the figures collected the second time around would match. That is, if Point A read coordinates X,Y the first time, would the DGPS system still read X,Y when we returned to the same physical location later on? And if there was a discrepancy, just how much was it?
The WB6ZUH rover package. The rover uses a Garmin GPS III. The RTCM code received from the reference station is fed to the GPS III, which is used in the normal manner. The rover also can transmit the NMEA sentences generated by the GPS II. |
Pete's DGPS system told us that we had returned to Points A, B, C, and D with an average X,Y position error of 9 feet. However, his system may be more precise than that. The rover displayed X,Y position data to a resolution of 0.001 minutes of latitude and longitude. Often in a digital system, the least significant digit will bounce around plus or minus one digit due to internal noise. We witnessed this happening here. The logged data clearly suggested that most of the position error observed was due to the last digits bouncing around by plus and minus one digit, and on one occasion two digits.
For a second test, we revisited the same four points twice, with the same GPS receiver used in the conventional (nondifferential) mode. The X,Y position figures collected on the first loop served as a baseline. The X,Y figures collected on the second loop were then compared against the baseline, and position errors were calculated. As before, I logged the first coordinate pair that was displayed and did not attempt to average the readings. I found that the average X,Y position error upon returning to each of the four points was 127 feet.
By now, it was obvious that DGPS had a lot to offer. DGPS had improved the precision by a factor of 14 (a 127-foot error versus a 9-foot error).
As a third test (back in the DGPS mode), we compared the coordinates stored in the reference station against those displayed on the rover when it was co-located with the reference station. They read the same (plus or minus the 0.001 minute noise factor) indicating that there was no major "DC offset" in Pete's system.
As a fourth test, we examined the ability of DGPS to stabilize
the altitude readings displayed on the rover unit. (GPS altitudes
can easily wander several hundred feet and are notoriously inaccurate.)
Here, the use of DGPS calmed down the altitude variations (as
seen in a one minute observation period at each point) by a factor
of 4.5. There was some indication that we should have waited a
little longer before starting to log our data at each location.
Had we done so, it is possible that the altitude improvement would
have been greater than 4.5.
The use of DGPS, as opposed to conventional GPS, improved the X,Y position precision by a factor of 14 and calmed down the altitude variations by a factor of 4.5 in our cursory tests. The real improvement may be more than these figures reflect for the reasons cited above. Congratulations to Pete Prossen for constructing a fine DGPS system out of simple Amateur Radio gear. Pete is still developing and testing his system and would like to hear from others with similar interests. He may be reached at: prossen@znet.com. For additional details, see "A Differential GPS Reference Station for Amateur Radio Use," below.
For additional information on this concept, contact Peter J. Prossen, WA6ZUH, 261 Del Valle Dr, Fallbrook, CA 92028-9427; tel 760-728-0014; fax 760-728-0015; prossen@znet.com.
A Differential GPS Reference Station for Amateur Radio UseThe NAVSTAR GPSThe NAVigation Satellite Timing and Ranging (NAVSTAR) Global Positioning System (GPS) is an all-weather, radio-based satellite navigation system. The overall system consists of the space segment, the ground control segment, and the user segment. Proper application of the feature of these segments enables users to accurately determine three-dimensional position, velocity, and time. The space segment is a constellation of 24 satellites (21 active plus three spares) 20,183 km above the earth. The ground control segment consists of a master control center and various monitoring stations. The user segment consists of all GPS receivers and their support equipment.
A GPS receiver's position is determined by the geometric intersection
of several simultaneously observed ranges (satellite-to-receiver
distances) from satellites with known coordinates in space. The
receiver measures the transmission time required for a satellite
signal to reach the receiver. Because this measurement includes
various propagation delays as well as satellite and receiver clock
errors, it is not a true geometric range, so it's called a pseudorange.
The receiver processes these pseudorange measurements along with
the received ephemeris (satellite orbit) data to determine
its three-dimensional position. Measurement Errors
There are five major sources of pseudorange error in GPS measurements.
Differential GPSA solution to the error problem involves the use of a specially-equipped GPS receiving station located at a fixed and precisely known position. This is called a reference station. To the extent that the errors are common to both the reference station and any given roving receiver, they can be ameliorated. For Amateur Radio applications, pseudorange correction is the only practical choice to implement this correction. Here's how it works. The reference station receiver measures the pseudoranges in the normal manner. Then, since the location is precisely known, it calculates the actual ranges to the same set of satellites. The differences between the measured pseudorange and the calculated true range for each satellite can then be found. Other GPS receivers in the area will encounter similar error conditions, so this difference becomes a correction that can be disseminated to them.
For users near the reference station, the respective signal paths to the satellites are sufficiently close so that compensation is almost complete. As the separation is increased, propagation delays lead to an error in the differential measurement called spatial decorrelation. The accuracy of the correction is significantly diluted when separations approach 200 km.
A standard message format for the Differential GPS correction
data is RTCM SC-104. DGPS corrections are currently provided
by the US Coast Guard within their radiolocation beacons. The
service is also available by subscription from private organizations,
propagated on subcarriers of FM broadcast stations. Additionally,
the FAA is currently installing a system called Wide Area Augmentation
System (WAAS) for use in aircraft navigation. All of these systems
require the use of specialized and somewhat expensive beacon receivers
in addition to the GPS receiver they discipline. DGPS in Amateur RadioHams have at their disposal a magnificent facility to disseminate DGPS corrections. There is almost no location in the country where a VHF or UHF repeater is not available to retransmit weak signals over large distances using digital packet protocols. In addition, hams like to tinker with technical things, and GPS is a very hot item. Two design goals I kept in mind were accuracy achieved at low cost, and minimal additional traffic burden for the repeater. Unlike commercial DGPS services that transmit corrections in the RTCM SC-104 format continuously over dedicated channels, Amateur Radio applications must wedge the transmissions into digital data packet channels carrying other traffic. Therefore an effort must be made to minimize the quantity of correction data transmitted, consistent with the accuracy needs of rovers. Because of the dynamic nature of GPS, the usefulness of a DGPS correction decreases with its age. An additional parameter, the pseudorange rate of change, can be used to extrapolate subsequent pseudorange corrections in the absence of fresh information, but accuracy continues to decline. If you require a 1-meter maximum position error, the error growth will exceed tolerance in about 6 seconds. Beyond that, and depending heavily on the intensity of selective availability, the error growth begins to rise exponentially. A correction is of questionable value by an age of about 40 seconds.
There are conflicting requirements here, so the reference station
must provide a means to meter the correction messages at intervals
and for durations requested by rovers who need the correction
service. Furthermore, transmissions should cease when no rovers
are in need of them. To accomplish this, the reference station
receives requests from rovers via a packet radio data link. The
request specifies the interval between correction transmissions
needed for the required degree of accuracy. The reference station
then transmits corrections for a limited time, determined by the
system operator. RTCM Data MessagesAs of version 2.2 of the RTCM Recommended Standards, 30 types of data messages have been defined within the RTCM format. Two of these, types 1 and 9, can convey the correction data to rovers. In addition, a type 4 message may be needed to convey the reference station's currently selected datum. The type 1 message contains correction data for all satellites in the view of the reference station (maximum of 8 for this project). The smaller type 9 message contains data for up to three of the satellites in view. Multiple type 9 messages are needed to deliver correction data for the entire set of satellites.
The type 9 message was developed as a means to reduce the application
latency in systems with a slow but continuous delivery rate. The
US Coast Guard beacons, for example, deliver at 50 bits per second.
With Amateur Radio packet deliveries at 1,200 baud or higher,
there seems to be no advantage offered by type 9 messages. HardwareA specific GPS receiver, the Motorola Oncore VP model, is unique among most in that it has the capability to calculate range from a known position to any GPS satellite that is being tracked. It is therefore capable of generating the correction parameters needed by roving receivers to improve position accuracy. The VP is an OEM device, provided as a circuit card less power supply, I/O signal conditioning, and packaging. It can output a wide variety of digital messages in several formats, none of which is compatible with the RTCM SC-104 standard. This receiver has been selected to be the core of the reference station. An additional microcontroller circuit card must be developed to transform the VP's messages into standardized RTCM SC-104 format, to provide RS-232C conditioning for the input/output channels, to process requests for service, and to manage the transmission of correction data packets.
TAPR (Tucson Amateur Packet Radio) supplies a DGPS reference station
interface board that connects to a Motorola Oncore VP OEM GPS
receiver to create a low cost 8-channel DGPS reference station.
The reference station provides pseudorange differential GPS corrections
that conform to the RTCM SC-104 Type 1 message format. Corrections
can be transmitted via data link to remote users. See the TAPR
Web site
for details
and ordering info.--Peter J. Prossen, WA6ZUH |
| Previous: JY1's North Carolina Connection | Next: H&K Lite Helpful hints from other amateurs |
