GPS Receiver Test Plan for Week Roll Over and Y2K Critical Dates Compliance M.G. Petovello, G. Fotopoulos, M.E. Cannon and, G. Lachapelle, Department of Geomatics Engineering The University of Calgary S. Ryan, Canadian Coast Guard BIOGRAPHIES M. Petovello is a M.Sc. student in the Department of Geomatics Engineering, The University of Calgary. He obtained a B.Sc. in Geomatics Engineering from the same department in Spring 1998. G. Fotopoulos is a M.Sc. student in the Department of Geomatics Engineering, The University of Calgary. She obtained her B.Sc. from the same department in Spring 1998. Dr. M.E. Cannon is a professor in Geomatics Engineering at The University of Calgary. She has been involved with GPS research since 1984 and has published numerous papers on static and kinematic GPS positioning. She is also the author of several GPS related software programs. Dr. G. Lachapelle is a professor in the Department of Geomatics Engineering where he is responsible for teaching and research related to positioning, navigation and hydrography. He has been involved with GPS developments and applications since 1980 and is the author of numerous papers on the subject. S. Ryan has a B.Eng. in Electrical Engineering from Memorial University, St. John’s, Newfoundland, and is an Electronic Systems Engineer in the Technical and Operational Services Directorate of the Canadian Coast Guard. He is currently on educational leave at the University of Calgary where he is pursuing a M.Sc. degree in Geomatics Engineering. ABSTRACT Proper GPS positioning, navigation and time transfer capabilities are essential to many industries in terms of economic benefits and public safety. With the new millennium nearly upon us, the urgency to confirm proper operation of various GPS receivers during all Year 2000 (Y2K) epochs is increasing. This urgency is the Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) motivation behind a series of tests designed to identify if a GPS receiver is Y2K compliant. Several critical epochs apply to GPS including; (i) GPS Week Roll Over 1024 (WRO1024) in August, 1999, (ii) century roll over from 1999 to 2000 (Y2K), (iii) leap year day of February 29, 2000 (Y2KLY), (iv) year roll over from 2000 to 2001 (new millennium) and, (v) various end-of-record (EOR) dates. A test plan designed for evaluating a receiver’s use of almanac and ephemeris data before, during and after all of the critical dates described above using a GPS signal simulator is presented. Finally, a procedure for assessing receiver compliance in various domains is included. INTRODUCTION The anticipation and excitement surrounding the new millennium is partially shrouded by the fear that many electronic devices, including GPS receivers, will begin malfunctioning causing widespread problems. The belief is that devices performing time and date computations will incorrectly interpret the new year as 1900 instead of 2000 thus producing unexpected results or even failure. This stems from the fact that older devices only store the last two digits of the year (e.g. 95 to represent 1995) in order to save memory. Questions also arise as to the ability of such devices to correctly interpret the year 2000 as a leap year. The problem is compounded for GPS since the number of weeks since January 6, 1980 (necessary for proper time and date determination) will rollover from 1023 to 0 in the navigation message in August, 1999. The heavy reliance on GPS for safety of navigation by many users, including those in the marine industry, has brought the issue of receiver compliance to the forefront. This was the motivation behind a joint effort by the Canadian Coast Guard (CCG) and the Department of Geomatics Engineering, the University of Calgary to develop a thorough test plan for assessing receiver compliance during various critical dates through the use of a GPS signal simulator. Tests are designed for two 1/9 purposes, namely to assess the compliance of the CCG’s DGPS Stations and a variety of marine user receivers under single point and differential conditions. The test plan was developed by critically analyzing the various scenarios that could cause problems at each critical epoch. Other test plans developed by agencies such as the US Coast Guard, the US Air Force and various receiver manufacturers were also consulted. Discussion begins with a summary of critical dates tested and why they are considered important. Specific receiver issues addressed by the test plan are presented. A description of the test plan proper is included, both for the DGPS Station and the individual marine user receivers. Finally, a brief discussion on how to analyze the data in the time/date, observation, residual, position and differential correction domains is described. DESCRIPTION OF CRITICAL DATES A variety of critical dates arise when considering GPS receiver compliance. These dates can be broken down into three main categories; (i) those dealing specifically with GPS, (ii) those dealing with the new millennium and, (iii) end-of-record dates. Each of these categories are discussed below. A time line of all critical dates tested is given in Figure 1. GPS Specific Dates The only date that is truly GPS specific involves the roll over of the number of weeks since the start of GPS time (January 6, 1980) from 1023 to 1024. The week number is stored as a ten bit integer in the satellite and thus can take a maximum value of (210-1=) 1023. Consequently, when the week rolls over from 1023 to 1024 on August 22, 1999 the satellites will again transmit a value of zero and “users must account for the previous 1024 weeks” (ICD-GPS-200, 1993). Failure to correctly account for the previous 1024 weeks may result in time/date errors, position errors or failure of the receiver. New Millennium Dates Four critical dates are associated with the new millennium. Each is listed below with a brief description of why they are important: 1. Year 2000 roll over (Y2K): December 31, 1999 to January 1, 2000. Since receivers may still utilize two digits to represent the year, the receiver may incorrectly determine the year to be 1900. 2. Year 2000 leap year logic (Y2KLY): February 29, 2000. Since the year 2000 is evenly divisible by four and four hundred, it is a leap year. However, it is believed some software programs implement an algorithm that checks if a year is divisible by one Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) 3. 4. hundred in which case it is not considered a leap year. Year 2001 roll over (New Millennium - NM): December 31, 2000 to January 1, 2001. Since the Roman calendar originally began with year one (and not year zero) the actual beginning of the new millennium is January 1, 2001. It is believed that some receivers may experience difficulties with date computations at this epoch. Year 2001 leap year logic: February 28 to March 1, 2001. Although the year 2001 is not a leap year, some believe that receivers may assume all years after 1999 are leap years. End-of-Record Dates End-of-record (EOR) dates are those dates which form a series of numbers (typically 9999) that are believed to be used by programmers to indicate the end of a record. Consequently, when the date is computed, the receiver may incorrectly interpret it as an EOR thus causing malfunctions. The EOR dates addressed in this plan include: 1. April 9, 1999. This could be interpreted as an EOR if receivers count the number of days into the year since April 9, 1999 is the 99th day of the 99th year. 2. September 9, 1999. This could be interpreted as an EOR if receivers use the Roman calendar for computation purposes since September 9 can be represented as 9/9/99. From a GPS perspective, WRO1024 appears the most critical epoch since the determination of the GPS week affects most of the important receiver functions including satellite prediction, orbit computation, and time and date calculation. Furthermore, it is the only epoch that involves a “discontinuity” in time because the week number transmitted by the satellites will actually revert to a previous value. All other critical epochs are incremental in nature and, in theory, should be more easily accounted for. RECEIVER ISSUES Proper functioning of a GPS receiver requires that the information transmitted by each satellite be correctly interpreted and utilized. Ephemeris and almanac data are typically most important for a GPS receiver. Tests must therefore be designed to test the various ways in which the receiver utilizes this information. A brief explanation of how ephemeris and almanac data are used and what problems may occur is given below. 2/9 01 20 r” ea pY Mar “L ea Feb M Jan Nov Oct Sept Aug Jul May Jun Apr 2000 Dec Ne w LY Y2 K Mar Feb K Y2 Jan Dec Nov Oct Sept ille 9/9 RO EO 102 R 4 (S ep t Aug Jul May Jun W EO R Apr nn ium 9) ) / 99 (A pr 9 `9 9 Mar NT M Feb IO N Jan 1999 2001 Figure 1 – Time Line of GPS Critical Dates Ephemeris Ephemeris data contains information for precisely computing the position of a single GPS satellite. The ephemeris data contains a set of parameters describing a particular satellite’s orbit over time and a reference time (TOE) to which these parameters apply. These parameters are typically updated every two hours and are valid for two hours on either side of the reference time, however deviations from this schedule are possible. Although the GPS week number is not explicitly included in the ephemeris message, the receiver must be able to account for a change in week number in the event that the reference time falls in a week different from the current week. Failure to do so may result in a receiver error or failure. Almanac Almanac data consists of information describing the approximate locations of all satellites as a function of time. The reference time for the almanac parameters (TOA) currently being transmitted can differ by up to 3.5 days (in the past or future) from the current time. However, the parameters are valid for much longer periods, namely up to six months. The main purpose of the almanac is to assist the receiver in acquiring signal lock at the start of and/or during a data collection campaign. By knowing an approximate user location, periods of satellite visibility and the Doppler shift for an individual satellite can be computed, thus limiting the signal search to code bins only. If satellites are not visible, the receiver may power down some tracking channels in order to conserve power. When a satellite is deemed to be visible, the tracking channels can then be powered up and the new satellite can be tracked. If the reference week number for the almanac differs from the current week the receiver must be able to correctly interpret the difference. Failure to do so may result in Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) newly visible satellites (i.e. satellites rising over the horizon) not being tracked causing degraded position accuracy and reliability. Note that since the TOA can be up to 3.5 days into the future, problems may arise as early as noon on Wednesday, August 18, 1999 when an almanac with a reference week of 1024 begins broadcasting. Information Stored in Receiver at Power-up Since receiver behaviour may differ depending on which information (i.e. almanac and ephemeris) is available at power-up, this is the last major receiver issue addressed herein. For example, certain information may indicate that certain critical dates have already elapsed (e.g. WRO1024) thus aiding the receiver in performing various computations. By varying the reference time of the previously downloaded almanac parameters (and to a lesser extent, the ephemeris parameters), receiver performance can be evaluated as function of how much time has elapsed since the receiver was last used. TESTING OBJECTIVES To realistically evaluate a receiver’s performance all components, starting with the RF front end, should be subject to test. This can only be done using a GPS signal simulator, which can be programmed to simulate any period of time in the past and/or future. Simulators further allow for the simulation of atypical occurrences (e.g. late uploads to various satellites) thus providing the flexibility to thoroughly evaluate a receiver. The development of the test plan focused on addressing the three receiver issues presented in the previous section. Three corresponding testing objectives have been identified and are described below. 3/9 The first objective involves ensuring receivers properly interpret ephemeris information referenced before and/or after a critical epoch. Although cutovers to data sets referenced after the critical epoch typically occur two hours prior to that epoch, this is not mandated by ICDGPS-200 (1993). Therefore, other combinations of ephemeris data cutover schedules are investigated. Second, tests must ensure receivers properly track satellites rising above the horizon before and after the critical epoch through the use of its almanac, regardless of when the almanac parameters are referenced (e.g. before or after the critical epoch). If the receiver is correctly using the downloaded almanac, new satellites should be easily tracked. However, if rising satellites are not tracked, position accuracy and reliability will suffer. Simulating satellites to rise above the horizon both before and after the critical epoch allows for an assessment of whether the receiver is correctly using its almanac data. The final objective is to ensure receivers will correctly transit the critical epoch regardless of what information, if any, is stored in memory at power-up. By varying the age of the previously downloaded almanac parameters, any period elapsed since the last time the receiver was used can be simulated. This may be important if one or more critical dates have elapsed since the last time the receiver was used. Under such a situation, the receiver may fail to function properly because it was forced to handle several critical dates at once despite being capable of handling each critical date in succession. Completely erasing all memory before power-up simulates the scenario where a full reset of the receiver has occurred, a typical occurrence if problems are encountered. TEST DESCRIPTION This section begins with a description of the general setup of all tests, followed by a more in-depth discussion of the specific tests to be performed for each of the critical dates. GPS WRO1024 tests are discussed first, as they are the most complicated from a testing perspective and are most relevant to GPS. The remaining tests (new millennium tests and EOR tests) will follow. The section concludes with a discussion of what parameters should be considered when performing each test. All tests are performed on a GSS 4760 DGPS signal simulator using software version 6.30. Each test is broken down into three stages, namely a loading period, an initialization period and a testing period. Prior to the loading period, the receiver has all previous almanac data, ephemeris data, and date information nulled/erased (herein referred to as performing a full reset). Testing has showed that depending on the GPS receiver manufacturer, this may be necessary since some receivers do not operate correctly if consecutive tests are not sequential in time Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) (i.e. appear to go back in time) and therefore is performed as a safeguard. The loading period is used to download desired almanac and ephemeris parameters into the receiver. These parameters are those to be stored in memory at the beginning of the test (i.e. at power-up). Once the receiver is tracking its first satellite the loading period is 15 minutes in length, 12.5 minutes to download an almanac plus 2.5 minutes of contingency. After loading the necessary data, the receiver is power cycled before the initialization period begins. Note that the loading period is omitted if the receiver’s memory is to remain nulled at the beginning of the test. The initialization period is a 15-minute period prior to the test, which is used to allow the receiver to acquire the almanac parameters currently being transmitted. This is needed because the almanac parameters stored in the receiver from the loading period do not necessarily match those currently transmitted, thus simulating an extended period when the receiver was not used. The testing period is a 30-minute period beginning 15 minutes before the critical epoch and continuing until 15 minutes after the critical epoch. Note that this 30 minutes follows immediately after the initialization period, see Figure 2, and constitutes the period of data to be analyzed to determine receiver compliance. Most tests begin with six visible satellites with a seventh satellite rising five minutes before and an eighth satellite rising five minutes after the critical epoch. Again, this is used to evaluate the receiver’s use of its almanac. A diagram depicting satellite visibility as a function of time is given in Figure 2. Because the normal satellite constellation has to be modified to obtain satellites rising at the desired times, care is taken to ensure that a good Position Dilution of Precision (PDOP) is maintained throughout the test. Otherwise, if the PDOP becomes too large it may become difficult to separate receiver malfunctions from the effects of poor geometry. Consequently, a PDOP lower than or equal to six (PDOP≤6) is maintained throughout the test period. Some tests are also performed using only four satellites in order to evaluate receiver performance under reduced satellite geometry. GPS WRO1024 Tests As previously discussed, the GPS WRO1024 epoch is the most difficult from a testing perspective because of the various combinations of almanac and ephemeris reference times that could be simulated. Consequently, some “modification” of the navigation data transmitted by each satellite must occur, relative to the typical case. Since this is a time consuming process, a balance between test reliability, that is the ability to identify if a receiver has 4/9 passed/failed, and testing time needs to be achieved. Therefore, only a representative subset of all possible tests is selected that still allows for a confident assessment of receiver performance. A list of all WRO1024 tests performed is given in Table 1 along with a brief description of each. In the table, tests occurring before, during and after WRO1024 are identified by shading of the corresponding rows. Figure 2 – Satellite Visibility as a Function of Time for Tests Using Eight Satellites For WRO1024, three separate sets of tests were devised, namely those before, during and after the critical epoch. Tests before the critical epoch serve two purposes. The first is to obtain baseline behaviour of the receiver under test during a “normal” week roll over (e.g. week 1022 to 1023). These results are to be compared to those obtained during WRO1024 to aid in the assessment of compliance. Second, it allows for the investigation of receiver performance when an almanac with a reference week of 1024 is downloaded prior to the actual week roll over (recall this could happen as early as noon on August 18, 1999). Tests performed during WRO1024 are designed to evaluate receiver compliance under various scenarios. The primary variable between all of the tests during WRO1024 is the ephemeris information transmitted by the satellites. Several combinations of ephemeris parameters referenced to week 1023 and/or 1024 are included. Although tests B2 and B5 represent the typical cases, the other tests could possibly be realized under the specifications of ICD-GPS-200 (1993). Finally, tests simulating post-WRO1024 time periods serve two purposes. The first is to ensure that the next “normal” week roll over, and presumably all subsequent week roll overs, will be properly handled by the receiver. The second purpose is to ensure the receiver will operate properly regardless of which information is stored in memory at power-up. In particular, can the receiver Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) properly interpret the week number if no information is available at power-up? Proper computation of the date may occur if the receiver has apriori knowledge that the week number has already rolled over but the receiver may fail if such information is unavailable. This scenario is important since memory resets are commonly used in practice when difficulties are encountered. New Millennium and EOR Tests The remaining critical dates are less concerned with what is being transmitted by the satellites than with: 1. The general performance of the receiver during the transition of the critical epoch. This is of minor importance since the conversion from GPS time to a UTC/local time and date is relatively trivial. Consequently, these transitions are only tested with a nulled memory (presumably the most challenging scenario for a receiver). 2. The performance of the receiver after the critical epoch as a function of the information available in receiver memory at power-up. The concern is that performance after the critical epoch may be dependent on whether the receiver is aware that various other critical dates have elapsed since the last power-up. For example, the receiver may fail if an attempt to handle more than one critical epoch at a time is made. The first week roll over (WRO) after the critical epoch is used since WROs presumably pose the largest problems for receivers. A list of all remaining critical epochs tested is given in Table 2 along with a brief description of each. Again, row shading is used to discriminate between critical epochs. Those epochs believed to be of less significance (e.g. EOR and “Leap Year” 2001) are given less attention accordingly but are included for completeness. All of the above tests can be performed in single point or differential mode. This decision was based on how the receiver is most likely to be used in practice. For typical marine requirements, differential GPS is most commonly used and therefore is tested more thoroughly. Simulation Parameters One of the major benefits of using a GPS signal simulator lies in the flexibility of selecting which errors are to be generated. For example, should tropospheric and ionospheric delays be included? Should a Selective Availability-like (SA-like) ranging error be superimposed on the satellite ranges? Testing has shown that some receivers automatically apply some form of atmospheric correction to the measured ranges that cannot be disabled. Consequently, simulated signals always contain the normal atmospheric delays/advances as this most closely simulates operational conditions. 5/9 However, the corruption of satellite signals with an SAlike error is only performed if testing the receiver in differential mode. The reason for this is SA-like signals will contaminate the measured ranges and computed single point positions to such a large extent that any small discontinuity resulting from receiver non-compliance may pass unnoticed. Test A1 Date 14/08/99 to 15/08/99 Almanac Rx: week 1022 Tx: week 1023 A2 14/08/99 to 15/08/99 Rx: week 1022 Tx: week 1023 A3 20/08/99 to 21/08/99 Rx: week 1023 Tx: week 1024 B1 21/08/99 to 22/08/99 Rx: week 1023 Tx: week 1023 B2 21/08/99 to 22/08/99 Rx: week 1023 Tx: week 1024 B3 21/08/99 to 22/08/99 Rx: week 1023 Tx: week 1024 B4 21/08/99 to 22/08/99 Rx: week 1023 Tx: week 1024 B5 21/08/99 to 22/08/99 Rx: week 1024 Tx: week 1024 C1 28/08/99 to 29/08/99 Rx: Nulled Tx: week 1025 C2 28/08/99 to 29/08/99 Rx: week 1023 Tx: week 1025 C3 28/08/99 to 29/08/99 Rx: week 1024 Tx: week 1025 Evaluation of a DGPS reference station always uses SAlike corrupted signals. The accuracy of differential corrections generated by a base station are more reliably determined when using SA-like corrupted signals because the corrections have to track larger range and range rate errors. Without the corrupted ranges, errors in the differential corrections may be difficult to separate from receiver noise. Table 1 – GPS WRO1024 Tests Ephemeris Test Description Rx: week 1022 Four satellites only. Obtain baseline behaviour of Tx: week 1023 receiver WRO performance during a normal week roll over under reduced satellite visibility. Rx: week 1022 Obtain baseline behaviour of receiver WRO Tx: week 1023 performance during a normal week roll over under good satellite visibility. Rx: week 1023 Obtain baseline behaviour of receiver WRO Tx: week 1023 performance when the satellites transmit an almanac referenced to week 1024. Rx: week 1023 Investigate whether the receiver can properly Tx: week 1023 interpret ephemeris information referenced before the critical epoch during the epoch transition. Rx: week 1023 This is the typical scenario to be expected. It tests Tx: week 1024 that the receiver can properly cross the critical epoch when old almanac parameters are stored in memory. Rx: week 1023 Investigate how the receiver handles ephemeris Tx: week 1023 and information reference both before and after the 1024 critical epoch under good satellite coverage. Rx: week 1023 Four satellites only. Investigate how the receiver Tx: week 1023 and handles ephemeris information reference both 1024 before and after the critical epoch under reduced satellite coverage. Rx: week 1024 This is the typical scenario to be expected. It tests Tx: week 1024 that having an almanac referenced to week 1024 in memory does not adversely affect receiver performance. Rx: Nulled Investigate receiver WRO performance after Tx: week 1025 WRO1024 without any information stored in memory. This simulates a full reset of the receiver after WRO1024. Rx: week 1023 Investigate receiver WRO performance after Tx: week 1025 WRO1024 if it has stored week 1023 almanac parameters. Rx: week 1024 Investigate receiver WRO performance after Tx: week 1025 WRO1024 if it has stored week 1024 almanac parameters. Rx – Stored in the receiver Tx – Transmitted by the satellites Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) 6/9 Table 2 – New Millennium and EOR Tests Almanac Ephemeris Test Description Rx: Nulled Rx: Nulled Investigate receiver performance during the Tx: week 1005 Tx: week 1004 transition into the potential EOR indicator date of April 9, 1999. D2 09/04/99 Rx: Nulled Rx: Nulled Investigate receiver performance during the to 10/04/99 Tx: week 1005 Tx: week 1004 transition out of the potential EOR indicator date of April 9 to April 10, 1999. E1 08/09/99 Rx: Nulled Rx: Nulled Investigate receiver performance during the to 09/09/99 Tx: week 1027 Tx: week 1026 transition into the potential EOR indicator date of September 9, 1999. E2 09/09/99 Rx: Nulled Rx: Nulled Investigate receiver performance during the to 10/09/99 Tx: week 1027 Tx: week 1026 transition out of the potential EOR indicator date of September 9 to September 10, 1999. F1 31/12/99 Rx: Nulled Rx: Nulled Investigate receiver performance during the to 01/01/00 Tx: week 1043 Tx: week 1042 transition of Y2K when memory has been nulled. F2 01/01/00 Rx: Nulled Rx: Nulled Investigate receiver WRO performance after Y2K to 02/01/00 Tx: week 1043 Tx: week 1043 when memory has been nulled. F3 01/01/00 Rx: week 1023 Rx: week 1023 Investigate receiver WRO performance after Y2K to 02/01/00 Tx: week 1043 Tx: week 1043 when almanac data is referenced before WRO1024. F4 01/01/00 Rx: week 1042 Rx: week 1042 Investigate receiver WRO performance after Y2K to 02/01/00 Tx: week 1043 Tx: week 1043 when almanac data is referenced between WRO1024 and Y2K. F5 01/01/00 Rx: week 1043 Rx: week 1043 Investigate receiver WRO performance after Y2K to 02/01/00 Tx: week 1043 Tx: week 1043 when almanac data is referenced after Y2K but before Y2KLY. G1 28/02/00 Rx: Nulled Rx: Nulled Investigate receiver performance during the to 29/02/00 Tx: week 1051 Tx: week 1051 transition of Y2KLY when memory has been nulled. G2 29/02/00 Rx: Nulled Rx: Nulled Investigate receiver performance during transition to 01/03/00 Tx: week 1051 Tx: week 1051 from February 29 to March 1, 2000 when memory has been nulled. G3 04/03/00 Rx: Nulled Rx: Nulled Investigate receiver WRO performance after to 05/03/00 Tx: week 1052 Tx: week 1052 Y2KLY when memory has been nulled. G4 04/03/00 Rx: week 1051 Rx: week 1051 Investigate receiver WRO performance after to 05/03/00 Tx: week 1052 Tx: week 1052 Y2KLY when almanac data is referenced between Y2K and Y2KLY. G5 04/03/00 Rx: week 1052 Rx: week 1052 Investigate receiver WRO performance after to 05/03/00 Tx: week 1052 Tx: week 1052 Y2KLY when almanac data is referenced between Y2KLY and new millennium. H1 31/12/00 Rx: Nulled Rx: Nulled Investigate receiver performance during the to 01/01/01 Tx: week 1095 Tx: week 1095 transition of NM when memory has been nulled. H2 06/01/01 Rx: Nulled Rx: Nulled Investigate receiver WRO performance after NM to 07/01/01 Tx: week 1096 Tx: week 1096 when memory has been nulled. H3 06/01/01 Rx: week 1094 Rx: week 1094 Investigate receiver WRO performance after NM to 07/01/01 Tx: week 1096 Tx: week 1096 when almanac data is referenced between Y2KLY and NM. H4 06/01/01 Rx: week 1096 Rx: week 1096 Investigate receiver WRO performance after NM to 07/01/01 Tx: week 1096 Tx: week 1096 when almanac data is referenced after NM. I1 28/02/01 Rx: Nulled Rx: Nulled Investigate receiver performance during the to 01/03/01 Tx: week 1103 Tx: week 1103 transition from February 28 to March 1, 2001 when memory has been nulled. Rx – Stored in the receiver Tx – Transmitted by the satellites Test D1 Date 08/04/99 to 09/04/99 Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) 7/9 All receivers not being tested as DGPS reference stations are simulated in kinematic mode to check the validity of the GPS time stamp. Since the reference position, that is the position to which the simulated ranges apply, is available from the simulator at every epoch a truth trajectory can be generated. If static data is used, only one reference position is generated. This means that comparisons between the reference position (generated by the simulator) and computed positions (generated by the receiver) for a given epoch may appear correct while the time stamp is incorrect. By contrast, in kinematic mode, a separate reference position is generated by the simulator for each epoch. Therefore, for a given epoch in kinematic mode, if the reference and computed positions coincide, the receiver is compliant in terms of time and position for the critical epoch being tested. However, if the reference and receiver positions do not coincide, the receiver is deemed non-compliant. Finally, all tests simulate the transmission of L1 and L2 data. Receivers capable of tracking the P-Y code signal on L2 are forced (if possible) to use semi-codeless tracking techniques despite the fact that the simulated L2 signal is not encrypted. This realistically recreates what will be experienced by the receiver under normal operating conditions. EVALUATING RECEIVER COMPLIANCE Receiver compliance is evaluated in the position, time/date, residual, observation, and differential correction domains. Failure of the receiver in any of these aspects is sufficient to deem the receiver noncompliant. However, if a receiver has failed a test, the test is redone before designating the receiver noncompliant. This helps to ensure that operator errors or simulator inconsistencies do not artificially skew test results. If the test fails a second time, a non-compliant status is confidently assigned. Data analysis is performed on the appropriate NMEA messages output from the receiver and on results obtained in post-mission. Postmission results are computed using the University of Calgary’s C3NAVG2™ GPS/GLONASS processing software. Comparison of reference and receiver positions was discussed at the end of the last section. By analyzing position errors versus time, especially near the critical epoch, position domain compliance can be ascertained. Time domain analysis was also briefly discussed above, however verification of the time stamp alone, usually given as seconds into the week, is insufficient. The UTC/local time and date output by the receiver may also be of use in certain applications. Therefore, an investigation of the appropriate NMEA messages is performed. Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) Analysis of measurement residuals obtained from a leastsquares procedure typically demonstrate low frequency trends (within the limits of SA) superimposed with high frequency noise. Comparison of results with those observed during the baseline test should reveal any atypical behaviour, possibly resulting from receiver noncompliance. This may include discontinuous residuals or data gaps. A GPS simulator also provides access to the true ranges that should be measured to each satellite. This information is used to evaluate receiver performance in the observation domain. The problem is that the offset of the receiver clock from GPS time is large and will mask all other errors unless it is accounted for. The difference between the true range and the measured range is an estimate of the receiver’s internal clock offset (all other errors are cancelled in the difference). Since this error is common to all satellites for a given epoch, averaging the range errors of all satellites will effectively smooth receiver noise thus allowing for an accurate determination of the receiver’s internal clock error: dt Rx ≈ 1 N ∑ [P̂ − P ] N i i Eq. (1) i =1 where: dt Rx N P̂i Pi is the receiver clock error [m] is the number of satellites tracked is the measured range to the ith satellite [m] is the true range to the ith satellite [m] The observation error can then be obtained using: ErrorP = Pi − P̂i − dt Rx Eq. (2) where: ErrorP is the observation error [m] Note that Pi in Eq. (1) should be interpreted as the C/A code range. This is because an estimate of the true range to the satellite is needed and the C/A is the most reliable. However, once the clock offset is obtained from Eq. (1), Pi in Eq. (2) could symbolize any type of range measurement, including L1 and L2 carrier phases. By analyzing the errors produced from Eq. (2), the receiver’s performance in the observation domain can be evaluated. If the results differ significantly from the baseline behaviour, the receiver may be experiencing problems. If a receiver is to act as a reference station for generating differential corrections, the accuracy of these corrections 8/9 must be investigated. This is done by comparing the true range and range rate errors to the generated corrections. Any large or sudden deviations between the two values may constitute receiver non-compliance. Finally, if an entire differential system is to be evaluated, the testing should proceed in an incremental fashion. For example, one CCG DGPS Station contains, among other things (Ryan et. al., 1997): 1. Two Ashtech Z12-R receivers acting as reference stations (RS). 2. Two Ashtech Super C/A receivers acting as integrity monitors (IM). 3. Two control station computers (CS) to monitor and control the operation of the DGPS Station. Testing all components simultaneously limits the ability to identify problems and more importantly limits the ability to isolate a problem if one occurs. The incremental procedure used to evaluate the CCG DGPS Station includes: 1. Testing the RS and IM receivers in single point mode to ensure proper standalone operation. 2. Testing one pair of RS and IM receivers operating differentially, without any interaction from the CS computers. 3. Testing the entire DGPS Station including the CS software testing. Note that failures at this stage of testing would isolate the problem to the interaction of the CS with the RS and IM or to the CS software. Ryan, S., F. Forbes, and S. Wee, (1997). Avoiding the Rocks – The Canadian Coast Guard Differential GPS System, Proceedings of the International Symposium on Kinematic Systems in Geodesy, Geomatics and Navigation. Department of Geomatics Engineering, The University of Calgary. pp. 367-375. Spilker, J.J.Jr., (1996). GPS Navigation Message, Global Positioning System: Theory and Applications, Volume I, American Institute of Aeronautics and Astronautics, Inc., pp. 121-176. Year 2000 & Week Rollover Compliance Test Description – DGPS, United States Coast Guard. The following Internet site was used as a reference for information provided in this paper: GPS Receiver Boundary Rollover Test Plan. http://www.laafb.af.mil/SMC/CZ/homepage/y2000/ rollover.htm, September 28, 1998. The above procedure is included to illustrate how to easily identify problems in complex systems. Although testing of the entire system at once may save time, this must be balanced with the ability to confidently detect failures within various components of the system. CONCLUSIONS The test plan described herein is being used to test the Canadian Coast Guard DGPS Stations and a variety of marine receivers used on CCG vessels. REFERENCES Cannon, M.E., (1998). ENGO 561 – Satellite Positioning, Department of Geomatics Engineering, The University of Calgary. Global Positioning System Standard Positioning Service Signal Specification, (1995). ICD-GPS-200, (1993) Arinc Research Corporation, USA. Marine Technical and Support Services Directorate DGPS Specification - Annex A. Canadian Coast Guard. August 2, 1995. Proceedings of ION NTM 99, San Diego, CA, January 25-27, 1999 (pp. 891-899) 9/9
© Copyright 2026 Paperzz