GNSS pseudorange outliers severely disturb carrier-phase ambiguity resolution and positioning, particularly during signal blockages and recovery. Temporal Doppler consistency provides a reliable short-term range-change reference, but requires valid measurements at both endpoints and becomes unavailable upon signal re-acquisition. This paper presents a two-stage pseudorange screening framework that complements Doppler consistency with joint IMU-odometer motion prediction. The second stage compares predicted and measured temporal changes of double-differenced pseudoranges against an accepted anchor epoch without using current GNSS observations in the motion reference. A first-order error model separates anchor, relative-motion, and measurement uncertainties to propagate factor-graph covariance into the chi-square detection gate, and both stages adjust code weights through a closed-form minimum-inflation rule. Under continuous tracking, Doppler screening reduces RTK 3D RMSE by 96.18% and outperforms the a-contrario baseline by 58.56%. When Doppler checks are unavailable, joint IMU-odometer prediction detects a 10-m outlier change in 99.58% of trials after 30 s and 73.72% after 60 s, reducing RIO 3D RMSE by up to 78.51% after outages while RI screening outperforms an inertial-aided baseline by 70.42% in simulation and 39.11% in deep urban canyons.
Figures & tables
Fig. 1: Two-stage pseudorange screening using Doppler and motion-aided between-satellite single differences. (a) Valid Doppler predicts the pseudorange change between consecutive epochs. (b) When Doppler is unavailable, joint IMU-odometer prediction provides the antenna displacement from the anchor point to the current rover position. Red paths illustrate reflected signals alongside the direct paths. In (b), the two satellites and the base and rover receivers form the double difference. The antenna increment transfers the accepted anchor to the current epoch. Paths are schematic and not to scale.
Symbol
Meaning
Local Gaussian simulation
Gugong injection experiment
α
Per-test false-alarm probability
0.01
0.05
ng
Gyro noise density (rad/s/ Hz )
3.49×10−4
2.67×10−4
na
Accelerometer noise density (m/s 2 / Hz )
0.03
0.03
Δp^O,uvbu
Wheel translation SD (m)
[0.03,0.03,0.20]T
Propagated preintegration covariance
Δq^O,uv
Wheel rotation SD, local angles (rad)
[1.75,1.75,0.87]T×10−2
Propagated preintegration covariance
v^o
Wheel-speed noise density (m/s/ Hz )
Included in relative-pose noise
0.01
TABLE I: Method parameters for the two simulation experiments.
Fig. 2: LOS motion errors in three trials of the local Gaussian model. Blue bands show the joint-covariance 95% interval and grey bands show the accumulated standard-deviation bound. Coverage is evaluated over 20,000 trials. Both intervals contain only the motion contributions to Smot and Sˉmot , excluding anchor and code noise. The panels use different vertical scales.
Fig. 3: Detection probability without Doppler in the local Gaussian model. RI and RIO share nominal false-alarm probability and measurement noise. White contours indicate 50% and 90% detection. Probabilities follow ( 15 ) and are checked against 20,000 Gaussian trials. They do not depend on the code-weighting rule.
Fig. 4: Controlled code-pulse detection in the injection replays. (a) G25, 1C with continuous Doppler, where open circles mark injected outliers. (b), (c) G10:G25, 1C after 30-s and 60-s outages, where open circles show reference triple-difference from antenna truth and the actual detector anchor, solid curves show test residuals and dashed curves show gates. Grey marks the first injected epoch.
Total flagged / 60
First return / 2
Update time (ms)
Method
0 s
30 s
60 s
30 s
60 s
Mean
P95
RTK
0
0
0
0
0
2.39
3.66
RTK-OM1
60
58
58
0
0
2.18
3.05
Zair et al. [ 13 ]
60
58
58
0
0
2.94
4.23
RI
0
0
0
0
0
14.61
17.23
RI-OM1
60
58
58
0
0
17.33
26.43
TABLE II: Detection counts and estimator-update time in the injection replays. Total counts use 60 injected observations, and first-return counts use the two injected signals at the first epoch after GNSS recovery. Timing uses the 30-s outage experiment over 299 common epochs, where P95 denotes the 95th percentile.
Method
No outage
30 s outage
60 s outage
RTK
2.12 / 7.63 / 7.92
2.13 / 8.02 / 8.30
2.13 / 8.02 / 8.30
RTK-OM1
0.20 / 0.22 / 0.30
0.29 / 1.91 / 1.94
0.29 / 1.92 / 1.94
Zair et al. [ 13 ]
0.31 / 0.66 / 0.73
0.35 / 1.95 / 1.99
0.35 / 1.96 / 1.99
RI
0.18 / 0.63 / 0.66
0.50 / 2.09 / 2.15
0.42 / 2.79 / 2.82
RI-OM1
0.16 / 0.22 / 0.28
0.39 / 0.60 / 0.72
0.28 / 1.72 / 1.74
RI-OM2
0.16 / 0.23 / 0.28
0.23 / 0.30 / 0.38
0.21 / 0.34 / 0.40
TABLE III: Positioning errors in the code-pulse injection replays. Horizontal / Up / 3D RMSE (m) over 299 common epochs. Bold marks the minimum within each sensor group and component.
Fig. 5: Horizontal and signed vertical errors in the injection replays. Grey marks identify injected epochs. The inset enlarges screened vertical errors after 60 s. Each column uses the common epochs of its condition in Table III .
Method
Gugong
UrbanNav Medium
UrbanNav Deep
RTK
1.98 / 5.76 / 6.09
2.68 / 5.12 / 5.78
3.08 / 7.98 / 8.55
RTK-OM1
1.98 / 5.76 / 6.09
2.64 / 4.87 / 5.54
3.08 / 7.83 / 8.41
Zair et al. [ 13 ]
1.98 / 6.12 / 6.43
2.62 / 4.96 / 5.61
3.08 / 7.98 / 8.55
RI
0.36 / 1.00 / 1.06
1.65 / 0.57 / 1.75
1.48 / 3.61 / 3.90
RI-OM1
0.36 / 1.00 / 1.06
1.65 / 0.53 / 1.74
1.50 / 3.47 / 3.78
RI-OM2
0.35 / 0.89 / 0.96
1.63 / 0.48 / 1.70
1.13 / 2.08 / 2.37
TABLE IV: Positioning errors on unmodified field recordings. Horizontal / Up / 3D RMSE (m) on common epochs. Bold marks the minimum within each sensor group and component. N/A denotes unavailable odometer data.
Fig. 6: Horizontal and signed vertical errors on unmodified field recordings. Rows cover the common supports in Table IV . No trajectory alignment or mean-error removal is applied.
In Global Navigation Satellite System (GNSS)-degraded environments, pseudolites (PLs) provide additional signal sources to enhance positioning performance, but their integration in optimization-based frameworks remains limited. This paper presents a loosely coupled factor graph optimization (FGO) framework that fuses the GNSS/PL least-squares (LS) solutions with inertial measurement unit (IMU) data. The evaluation considers low GNSS visibility scenarios with four high-elevation GNSS satellites and up to two PL transmitters over an 80~s window. FGO achieves a 22.8% to 41.3% reduction in mean 3D error compared to standard LS methods. Compared to a GNSS-IMU baseline, incorporating PL transmitters further improves positioning accuracy, with performance depending on geometry.
Reliable positioning in dense urban environments remains challenging due to frequent GNSS signal blockage, multipath, and rapidly varying satellite geometry. While factor graph optimization (FGO)-based GNSS-IMU fusion has demonstrated strong robustness and accuracy, most formulations remain offline. In this work, we present a real-time tightly coupled GNSS-IMU FGO method that enables causal state estimation via incremental optimization with fixed-lag marginalization, and we evaluate its performance in a highly urbanized GNSS-degraded environment using the UrbanNav dataset.
Radu-Andrei Cioaca, Paul Irofti, Cristian Rusu +3
University Politehnica of Bucharest, Romania · Three Tensors S.R.L., Romania · University of Bucharest, Romania +2
Accurate positioning, navigation, and timing (PNT) is fundamental to the operation of modern technologies and a key enabler of autonomous systems. A very important component of PNT is the Global Navigation Satellite System (GNSS) which ensures outdoor positioning. Modern research directions have pushed the performance of GNSS localization to new heights by fusing GNSS measurements with other sensory information, mainly measurements from Inertial Measurement Units (IMU). In this paper, we propose a loosely coupled architecture to integrate GNSS and IMU measurements using a Factor Graph Optimization (FGO) framework. Because the FGO method can be computationally challenging and often used as a post-processing method, our focus is on assessing its localization accuracy and service availability while operating in real-time in challenging environments (urban canyons). Experimental results on the UrbanNav-HK-MediumUrban-1 dataset show that the proposed approach achieves real-time operation and increased service availability compared to batch FGO methods. While this improvement comes at the cost of reduced positioning accuracy, the paper provides a detailed analysis of the trade-offs between accuracy, availability, and computational efficiency that characterize real-time FGO-based GNSS/IMU fusion.
Radu-Andrei Cioaca, Cristian Rusu, Paul Irofti +3
Three Tensors S.R.L., Romania · Navigation Systems Definition Section (TEC-SEN), European Space Agency · Romanian InSpace Engineering S.R.L. (RISE), Romania