|United States Department of Agriculture
The Rockwell International "Precision Lightweight GPS Receiver" (PLGR) with the new firmware (PLGR+96 FED) was evaluated on the GPS hardwood test site in Indiana in mid-August, 1996. This evaluation covers two new features of the firmware, Wide Area GPS Enhancement (WAGE) and an area calculation feature.
|The PLGR+96 FED is a five-channel, single-frequency Precise Positioning System (PPS) P"Y" Code receiver with built-in antenna.|
WAGE is a method to increase the horizontal accuracy of the P"Y" Code receiver, by adding additional range correction data to the satellite broadcast navigation message. The navigation message for each satellite is updated once daily or as needed. With WAGE, this daily update of each satellite navigation message would contain the range corrections for every other satellite in the constellation. Thus, more timely range correction information would be available for each satellite, resulting in increased horizontal accuracy. In the future the military plans to simplify the upload procedure, upload the data more often, and add more monitor stations for better range correction.
WAGE is available only to the Precise Positioning Service (PPS) or P"Y" Code receivers such as the PLGR. When the receiver is first turned on, it takes about 12-1/2 minutes to obtain the most recent WAGE data. The receiver always uses the most recent WAGE data available to calculate position. It will not use the data that is over 6 hours old (if the receiver has been off more than 6 hours, it will require 12-1/2 minutes to collect the data). After that, the process of using the data is automatic and transparent to the operator. Also, any time the receiver is on, it continually collects WAGE data (whether the WAGE mode is on or off). It only calculates position using WAGE data when WAGE is on; a "W" appears in the last line of the position display on the PLGR. When WAGE is turned off the receiver does not use the data to calculate position.
The area calculation feature in the PLGR+96 FED firmware provides for calculating the area of a polygon from selected waypoints (WP's). The operator selects the waypoints in the proper sequence that represent the polygon, and the receiver calculates the area. This can be a time-consuming process if a large number of waypoints are required to define the polygon. The operator has to designate waypoints one at a time in the proper sequence.
The GPS Hardwood Test Site is on the Hoosier National Forest near Bedford, IN. It consists of a closed polygon with seven stations on gently rolling terrain. It contains 3.32 acres. It is under a heavy oak, hickory, and beech hardwood canopy with moderate undergrowth, typical of an Eastern hardwood forest. The canopy tops are from 70 to 100 feet above the ground. The heaviest foliage occurs from May to August.
The PLGR is a five-channel, single-frequency Precise Positioning Service (PPS) P"Y" Code receiver with built-in antenna. It contains a security module that can eliminate the positional error intentionally introduced by selective availability (SA) and decode the encrypted signal which prevents its unauthorized use, called anti-spoof (AS). The receiver is not classified, but it is an accountable property item and should remain in control of the authorized user.
To evaluate the effects of WAGE on horizontal accu-racy, data were collected as point and walk files with WAGE on and off. The results were compared to the known value for the course. Data were collected over a 2-day period. Before each day's activities, the PLGR was allowed to collect the Daily Crypto Variable, the WAGE data, the latest almanac (1 day old), and to obtain a Figure of Merit (FOM) of 1--a process that required about 15 minutes each morning. The FOM is a dimensionless number indicating the potential horizontal accuracy of the displayed coordinates. A FOM of 1 means the position should be accurate to 25 meters or less.
Data for the point files were collected by stopping at each station on the traverse and recording a single position fix as a WP. This was repeated three times with WAGE ON and two times with WAGE OFF. These results were compared to the known position for the station and the horizontal error was calculated. Assuming averaging would improve accuracy, the above procedure was repeated for what was thought to be the average of 10, 30, 60, 120, and 180 position fixes at each station. After the data had been collected and analyzed, the manufacturer explained that the current PLGR could not electronically store the average as a WP. The only way to record the average is with paper and pen.
|Table 1--Evaluating the Rockwell PLGR+96 with WAGE ON using point files collected at the Bedford, IN, GPS Hardwood Test Site (FOM = 1).|
|Table 2--Evaluating the Rockwell PLGR+96 with WAGE OFF using point files collected at the Bedford, IN, GPS Hardwood Test Site (FOM = 1).|
The results are shown in Table 1 with WAGE ON and in Table 2 with WAGE OFF. Run No. 1 corresponds to the single position fix obtained at each station. Runs 2 through 6 correspond to what were thought to be the average of 10, 30, 60, 120, and 180 position fixes at each station. In fact, they represent only the last position fix obtained while averaging. Comparing the overall average of 9.1 meters using WAGE (Table 1) with the overall average of 8.5 meters without WAGE (Table 2) indicates no real advantage to WAGE when using a single position fix. Had the PLGR been able to record the average value, the results may have been different. WAGE did improve the horizontal accuracy of more than half of the individual station errors, but it also produced some of the largest errors: 50.3 meters for Run No. 2, Station E; and 30.7 meters for Run No. 4, Station D (Table 1). These errors compare to a maximum error of only 17.1 meters with WAGE OFF during Run No. 1, Station C (Table 2).
The reason for the large individual errors when using WAGE is unknown. It could be multipath, but that error should also appear when WAGE was not used, because the data were collected over several days at about the same time of day. This data would indicate no clear advantage to using WAGE when operating under a canopy. In fact, you could get very large position errors (from 30 to 50 meters), without being aware of them.
When attempting to record averages, the number of position fixes averaged generally equaled the number of seconds required to collect them (for instance, 120 fixes took 120 seconds). Several times when the receiver was turned on under the canopy, it had trouble acquiring satellite signals. This required the operator to move to an opening and acquire the signal, before moving under the canopy.
|Table 3--Evaluating the Rockwell PLGR+96 in the open (no canopy) using point files collected at a station near Missoula, MT (FOM=1).|
To compare the results with WAGE operating in the open, the test was repeated at a control station near Missoula, MT, where there were no obstructions to the signal. The same test procedure was followed except that no attempt was made to record a single position fix as a WP. The results are shown in Table 3. Run Nos. 1 through 5 were thought to be the average of 10, 30, 60, 120, and 180 position fixes, but really represented just the last position fix. These data indicate no real advantage (increased accuracy) to using WAGE as it is currently configured. Perhaps more frequent uploading of range correction data to the satellite will help. That improvement is in the planning stage by the military.
The PLGR area calculation feature was used to calculate the area of the polygon using the point files (single position record) obtained for each station on the hardwood test site and stored as waypoints. This was accomplished by selecting the desired waypoints in the proper sequence to create the polygon from which the receiver calculated the area. This was done for a series of six runs with WAGE ON and with WAGE OFF (Table 4). These results indicate that WAGE improved accuracy with a 3.8% average error in area determina-tion, compared to 5.9% average error without WAGE. Even though some of the horizontal positions used to calculate the area contained large errors, WAGE still produced better results. This would indicate that the offset caused by the large horizontal errors were in a direction that did not affect the area computation. This should not be expected to occur frequently when using a point file.
|Table 4--Comparison of area calculation methods from point files using the Rockwell PLGR+96 firmware and the Trimble PFINDER software.|
For comparison purposes, the area was also calculated using the Trimble Pathfinder PFINDER Software (ver. 3.00-11). This required some manipulation of the data but there was very little difference between areas determined with the PFINDER software and the PLGR+96 FED firmware (Table 4).
|Table 5--Comparison of areas calculated with the Trimble PFINDER software from walk files collected with the Rockwell PLGR+96 at the Bedford, IN, GPS Hardwood Test Site.|
Walk files were also used to determine area. They were collected as the operator traversed the course by using the Automark feature of the PLGR and recording a position as a waypoint every 5 seconds. This was repeated several times with WAGE ON and WAGE OFF. No attempt was made to use the PLGR to calculate the area. Such a calculation would have required manually selecting and designating over 100 waypoints to define the polygon before the area could have been calculated. Instead, the data were manipulated and loaded into the PFINDER software and the area was calculated. The results are shown in Table 5 for WAGE ON and WAGE OFF. There is no difference in the areas obtained with WAGE ON and WAGE OFF except for one area (3.73 acres) obtained with 121 position fixes with WAGE ON. This is an error of over 12% in area. The other errors are 3% or less. The reason for this large error (12%) is unknown.
Users should be cautious about using an autonomous GPS receiver to determine position--there is no method for checking the integrity of the data. Multipath and solar flares can cause large errors in position which affect the area calculation.
These tests were conducted under the assumption that when averaging position fixes the PLGR would record the results as a WP. In reality, the last position fix is recorded as a WP, not the average. These results are from single position fixes only. With averaging, the results could be completely different. Further testing of WAGE using averaging is anticipated.
This evaluation using single position fixes indicates no clear advantage to using WAGE in its current configuration. The results are too unpredictable. For determining horizontal position, WAGE generally produced good results, but it also produced some very large errors (30.7 and 50.3 meters). Its overall average error of 9.1 meters was worse than when WAGE was not used. The reason for these large errors is unknown, but the user should be aware of them and realize that they could seriously affect the accuracy of the data.
When these data were used to calculate the area of the traverse, the large errors did not seem to seriously affect the area determination. Apparently, the offsets due to the horizontal errors were in a direction that did not affect the area. The user should not plan on such a fortunate outcome. When the walk files were used to determine area, WAGE again produced a large error of 12.3% in area. Users should be very cautious about using WAGE under a canopy, because it can produce some very large errors in position that can affect area calculations.
Autonomous PPS receivers like the PLGR (with or without WAGE) should not be used for area determinations that are the basis for financial transactions. There is no integrity check of the data.
The military is working on methods to improve WAGE, such as providing more monitoring stations, and more frequent uploads of corrected data. They indicate results derived from WAGE data can degrade very rapidly and probably should not be used if they are more than 2 hours old (rather than the 6 hours currently recommended). WAGE technology should be reevaluated after the improvements have been made.
Additional single copies of this document may be ordered from:
USDA Forest Service
Missoula Technology & Development Center
Building 1, Fort Missoula
Missoula, MT 59804-7294
Phone: (406) 329-3900
Fax: (406) 329-3719
For further technical information, contact Tony Jasumback at MTDC.
Phone: (406) 329-3922
Fax: (406) 329-3719
About the Author--Tony Jasumback, the GPS Project Leader at MTDC, has been involved in the development and evaluation of GPS equipment for Forest Service use since 1984.
Visitor since October 1, 1999