Dual-Frequency GPS/GLONASS RTK: Experimental Results Mark Zhodzishsky, Michael Vorobiev, Alexander Khvalkov, Lev Rapoport, and Javad Ashjaee. Javad Positioning Systems
BIOGRAPHY
1. INTRODUCTION INTRODUCTION
Mark Zhodzishsky received his Ph.D. and Dr.Sc. in 1965 and 1984, respectively. He is a Professor at Moscow State Avia Aviati tion on Inst Instit itut ute. e. He is prese present ntly ly Rese Resear arch ch Grou Group p Manager at the Javad Positioning Systems. Michael Vorobiev is a Software Engineer at JPS where he researches researches in the area of signal processing processing for receivers of direct spread spectrum signals. Alexander Khvalkov is currently a Software Ingineer at JPS where he works on satellite navigation problem. received ed his DrSc DrSc degree degree in Autom Automati aticc Lev Rapoport Rapoport receiv Contro Controll in 1995 1995 from the Institu Institute te of Contro Controll Scienc Sciences, es, where he served as a leading researcher. Since 1998 he has been with Javad Positioning Systems. receiv iveed his his Ph.D Ph.D.. in EE from from the the Javad Javad Ashjaee Ashjaee rece University of Iowa. He has over 18 years of experience in the field of high precision GPS technology. He founded Ashtech Ashtech in 1986 1986 and led it to impres impressiv sivee techni technical cal and financial success till 1995. Javad is currently the President and CEO of Javad Positioning Systems that was founded in 1996.
Diff Differ eren enti tial al GPS has has been been succ succes essfu sfully lly used used worl worldw dwid idee to provi provide de real real-t -tim imee rela relativ tivee posi positi tion oning ing accuracy at the centimeter level. To obtain sub-centimeter level accuracy for real-time applications it is necessary to employ differential carrier phase measurements from two receivers. One of the receivers with known coordinates is call called ed a base base stat statio ion n and and the the othe otherr is call called ed a rove roverr receiver. The base station should be placed at the point where the most possible number of satellites is available to be tracked, whereas the rover position, both stationary and movable, may be absolutely optional, namely, in the open sky, in urban canyons, in the woods and so forth. The determination of the carrier-cycle ambiguities on the the fly fly is key key to any any appl applic icat atio ion n where here prec precis isee posi positi tion onin ing g at the the cent centim imet eter er leve level, l, in real real time time,, is required. The approach to integer ambiguity estimation is bas based on prel prelim imin inar ary y inte intege gerr res resolu olution tion of float loat ambiguities ambiguities for carrier carrier phase phase measuremen measurements. ts. Resolution Resolution speed depends on the number of measurements and their quality. In general, the ambiguities can be resolved more quickly and with greater certainty if observations from a lot of satell satellite itess are availa available ble.. Hence, Hence, the measu measurem rement ent number is determined by both the number of the satellites in view (both GPS and GLONASS systems) and signals in L1 and L2 ban bands bein being g trac tracke ked d by the the rece receiv iveer. Meas Measur urem emen entt qual qualit ity y is acco accoun unte ted d for for a numb number er of measurement errors such as noise, multipath, tropospheric and ionospheric errors. For very fast ambiguity resolution it is necessary to empl employ oy the the most most poss possib ible le numb number er of meas measur urem emen ents ts (GPS (GPS+G +GLN LN L1+L L1+L2) 2) and and mini minimi mize ze erro errors rs of thos thosee measur measures es.. Four Four featur features es were were used used in JPS-re JPS-recei ceiver verss to achieve this goal. They are: 1) Using Using joint-trackin joint-tracking g system system (co-op tracking) tracking) [1] to mini minimi mize ze nois noisee error errorss and and incr increa ease se the the numb number er of satellites tracked. 2) Empl Employ oyin ing g stro strobe be metho ethods ds of prev preven enti ting ng multipath [2] to decrease multipath errors. 3) Using RegAnt antenna with choke-ring [3]. 4) Applying tropospheric delay model to minimize tropospheric effects. One more thing is used used to increas increasee accura accuracy cy of coordinate estimates. In the case of the stationary rover, in
ABSTRACT
This paper describes results of testing JPS Legacy receivers with relative carrier phase measurements. Four different configurations of RTK engine have been studied: a) GPS+GLONASS L1+L2, b) GPS L1+L2, L1+L2, c) GPS+GLN L1, d) GPS L1. The test results comprise the following: Posit Position ionin ing g accu accura raci cies es with with eith either er only only GPS GPS or • GPS+GLN for both short (less 1 km) and large (more 10 km) baselines; Accur ccurac acy y and and spee speed d esti estim mates ates of init initia iall RTK RTK • ambiguity ambiguity fixing fixing for static static or kinemat kinematic ic modes of the engine; RTK operation operation under short-term short-term interrupti interruptions ons in the • line transmitting differential corrections from a base station ; RTK operation with with RegAnt choke-ring choke-ring antenna; • RTK operation with the CoOp tracking system. •
RTK of JPS receivers we use specific engine mode, namely, static. Experimental results of RTK engine operation will be given below. Four different configurations of this engine were considered: a) GPS+GLONASS L1+L2, b) GPS L1+L2, c) GPS+GLN L1, d) GPS L1. The structure and operating functions of the engine were given in [4]. Both the engine itself and methods employed [1, 2, and 3] have been improved for past year. However, here we restrict ourselves to some experimental results. 2. FIX ACCURACY AND RESOLUTION SPEED
Experimental results are summarized in comparative tables where the resolution time with confidence probability of 0.9 (that is time at which there exists ambiguity fixing for 90% of all experiments) and measuring accuracy (rms) of the baseline in threedimensional coordinates present the quality factors of RTK operation. The note ‘> 120 s’ means that ambiguity fixing time exceeds 2 minutes in the most part of experiments. The note ‘no fixing’ means that ambiguity fixing was not available during experiment time (5 minutes). There are a few test scenarios including different baseline lengths and environment for the rover. It should be noted that at undetermined baselines we used a “true” baseline estimate as an estimate of baseline coordinates. Scenario 1: 56m baseline, urban area.
GPS L1
GPS+ GLN L1
GPS L1+L2
Table 1 GPS+ GLONASS L1+L2
Accuracy, mm
15
10
11
8
Resolution time, s
>120
50
12
4
Scenario 3: 15 km baseline, open sky.
GPS L1
GPS+ GLN L1
GPS L1+L2
Table 3 GPS+ GLONASS L1+L2
Accuracy, mm
-
-
30
22
Resolution time, s
No fixing
No fixing
> 120
75
The “true” baseline was computed by post processing program PINNACLE for many hours [5]. All the results were obtained under the following conditions. The elevation mask was 5 degrees. The receivers were forced to operate in static mode. Reliability of fixing was chosen as 99.9%, static mode was on, cut-off elevation 0 angle was 15 , tropospheric model was on, and the co-op tracking system was off. It can be seen from these tables that using DualFrequency GPS/GLONASS RTK allows to get fixed solutions under acceptable duration for the most of scenarios, while using Dual-Frequency GPS (without GLONASS) in the woods does not allow to fix reliably within some minutes due to lack of GPS satellites in view. On the other hand, at the longer baselines SingleFrequency GPS+GLONASS can not provide fixing because of ionospheric delays. . 3. ADVANTAGES OF OPERATION IN STATIC MODE In this operational mode the filtration both floatambiguities and baseline coordinates are carried out. We did not observe any appreciable profit of this option for ambiguity resolution time. But difference between static and kinematic modes is in filtering the fixed estimate of the baseline length using the filter whose bandwidth is about 0.05 Hz. Such a filter enables to neutralize almost completely the impact of the noise component in measurement errors. 0.10
St a t ic mo d e Kin e ma t ic mo d e
Scenario 2: 500 m baseline, moderate wood Table 2 GPS GPS+ GPS GPS+ L1 GLN L1 L1+L2 GLONASS L1+L2
Accuracy, mm
-
Resolution time, s
No fixing
18
-
0.05
] m [ 0.00 Y
Fig. 1
16 -0.05
> 120
No fixing
40 -0.10 - 0 .1 0
- 0 .0 5
0 .0 0
X [m]
Fig. 1
0 .0 5
0 .1 0
Fig.1 shows errors between true and real coordinates of the baseline in X-Y plane for static mode on/off. The results were obtained by processing experimental data for Dual-Frequency GPS+GLONASS RTK (scenario 2: 500 m baseline, moderate wood). Another advantage of static mode we can observe in studying test results in the dense woods. In the dense woods the quality and number of measurements may not be enough for getting reliable fixed solutions at the centimeter level. In this case, when resolving the differential carrier phase problem we restrict ourselves to so-called “float estimation”. Such an estimate obtained from float (not fixed) ambiguities has decimeter-accuracy level if ambiguities have been filtered for 5-10 minutes. Fig.2 (static mode off) and fig.3 (static mode on) reveal errors in Dual-Frequency GPS+GLONASS RTK between estimated and true values of coordinates x, y, and z. The baseline length for this experiment was about 530 m. Rover’s antenna was located in the woods and was connected to the two receivers. The number of satellites used in computation varied from 5 to 9 during the experiment.
It can be seen from the graphs that the estimate converges faster to the true estimate if static mode is turned on. 4. RTK + REGANT (ANTENNA WITH CHOKERING)
In general, the measurement accuracy at short enough (less 1km) baseline lengths in the open sky, for example, in the field, is determined by multipath errors from underlying surface. To decrease those multipath errors JPS utilizes RegAnt antenna with choke-ring. The following experiment was performed to assess the impact of choke-ring antenna on RTK position accuracy. Two antennae (RegAnt1 and LegAnt1) were placed over water surface with 30 cm distance from each other. Two other antennae (RegAnt2 and LegAnt2) were situated on the bank (with the same distance from each other). The base lines (LegAnt1 – LegAnt2 and RegAnt1 – RegAnt2) were 30 m approximately. Fig.4 shows estimate errors of two such base lines for X and Y coordinates (static mode on, cut-off elevation mask was 0 10 ). Table 4 contains error rms for all coordinates. 0.010
2.50
LegAnt
X-coordinate
2.00
Y-coordinate
RegAnt
Z-coordinate
0.005
1.50
] 1 . 0 0 m [ r o r r e 0.50
] m [
0.000
Y
0.00
-0.005 -0.50 0. 0 0
3 0 0. 0 0
60 0. 00
9 0 0. 00
1 200 . 0 0
1 5 00 .0 0
1 800 . 0 0
t im e [ s ]
Fig. 2
-0.010 -0 .01 0
2.50
-0 .00 5
0.00 0
0 .005
0.01 0
X [m]
Fig. 4
X-coordinate
2.00
Y-coordinate Z-coordinate
X RMS, mm Y RMS, mm Z RMS, mm
1.50
] 1 . 0 0 m [ r o r r e 0.50
0.00
-0.50 0. 0 0
3 00. 0 0
6 0 0. 0 0
9 0 0.0 0
1 20 0 .0 0
time [s]
Fig. 3
1 5 00 .0 0
1 8 00 . 0 0
RegAnt 1.3 1.2 1.9
Table 4 LegAnt 2.1 1.8 3.2
The experiment corresponds to the case of very strong multipath from surface. Thus, we had a maximal advantage of choke-ring antenna in this case. In other cases, for example in field conditions, the advantage is smaller. Note, to reject multipath from reflected objects, which locate over the Earth’s surface JPS receivers employ strobe methods. Test results revealed the impact of those methods are given in [2]. .
5. RTK + COOP TRACKING SYSTEM
Under operation in the woods or in other severe environments, where signals of some satellites have noticeable attenuation, JPS suggests employing the co-op tracking system that enables to acquire weak signals and keep them and hence to provide the most number of measures for resolving RTK problems. The following experiment was performed to assess the impact of the co-op tracking system on operation of Dual-Frequency GPS+GLONASS RTK. Two rovers were connected to the LegAnt antenna being located in the moderate wood, one of the rovers having functioned with turned on co-op tracking. During the experiment (10 minutes) the number of RTK fixed solutions was 90% for the rover with co-op tracking on compared to 69% for the rover with co-op tracking off (static mode, cut-off elevation mask was 15 °). Fig.5 represents a dependence of the satellite number used for RTK estimate computation on time. The average value of the satellites used was 8.2 with co-op tracking in respect to 7 without co-op tracking. 9
8
s V S
7
time intervals without differential corrections were indicated as the intervals not having solution for the engine operated in delay mode. The experimental results have shown that JPS RTK in extrapolation mode is substantially insensitive (i.e. there were no lost fixed solutions) to 5-second breakdowns and can bear a part of interruptions whose duration is within 5 - 10 seconds (float solutions are possible). 7. CONCLUSION
The usage of Dual-Frequency GPS/GLONASS RTK allows to get the fast and reliable fixed base line estimate for the wide variety of environments. The executed experiments show that the static mode helps to improve the position accuracy of base lines both in the fixed mode and in the float mode. The static mode does not allow practically to decrease the ambiguity resolution time. Using the choke-ring antenna enables to diminish error rms of base line estimates 1.5 times (maximum). The use of Co-op tracking in severe environment enables to increase the number of fixed solutions (due to increasing satellites in view). Breakdowns in transferring differential corrections up to 5 s do not affect RTK operation. The experimental results were dependent on current conditions (SV constellation, environment etc.). So presented results are of interest for comparison of JPS RTK options rather than their absolute figures. 8. REFERENCE
6
with CoOp tracking without CoOp tracking
5 0
100
200
300 time [s]
400
500
600
Fig. 5 6. DUAL-FREQUENCY GPS/GLONASS RTK WITH EXTRAPOLATION MODE (FAST RTK)
For kinematic applications, JPS RTK has base measurement extrapolation mode. Such a mode allows to get estimates of the baseline for the current moment in the rover as well as to provide the stable engine operation if there exist short-term interruptions in the communication line between the base and rover. To assess maximum permissible duration of line interruptions without loss of fixed solutions we carried out the following experiment. The two rovers operated using the same navigation antenna and modem. Dual-Frequency GPS+GLONASS RTK of the first receiver functioned in delay mode while the second receiver operated in extrapolation mode. Short-term breakdowns (from 1 to 10 s) were simulated during the kinematic experiment by turning the modem antenna off. When post processed,
1. M.Zhodzishsky, S.Yudanov, V.Veitzel, J.Ashjaee. CoOp Tracking for Carrier Phase. Proc. Of ION GPS-98, pp. 653-664. 2. Veitsel V., Zhdanov A., Zhodzishsky M. The Mitigation of Multipath Errors by Strobe Correlators in GPS/GLONASS receivers // GPS Solutions, Volume 2, Number 2, Fall 1998. pp. 38-45. 3.V.Filippov, D.Tatarnicov, J.Ashaee, A.Astakhov, I.Sutiagin. The First Dual-Depth Dual-Frequency Choke Ring. Proc. Of ION GPS-98, pp. 1035-1040. 4. M.Zhodzishsky, M.Vorobiev, A.Khvalkov, J.Ashjaee. Real-Time Kinematic (RTK) Processing for DualFrequency GPS/GLONASS. Proc. Of ION GPS-98, pp. 1325-1331. 5. www.javad.com
