Pith. sign in

REVIEW 4 major objections 6 minor 30 references

Tram Positioning with Map-Enabled GNSS Data Reconciliation

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Adding per-satellite Gaussian-mixture pseudorange reconciliation and a soft track constraint to an iterated extended Kalman filter reduces tram positioning RMSE from 13.3 m to 6.0 m in simulation and from 8.55 m to 0.91 m on real data.

desk verdict New per-satellite moment-matching pseudorange reconciliation with a soft track constraint; the simulation is solid, but the real-data validation is confounded and needs a proper independent reference. read the letter →

arxiv 2506.08032 v1 pith:SSJIDIV3 submitted 2025-06-02 eess.SY cs.SYeess.SP

Jakub Kašpar , Vít Fanta , Vladimír Havlena This is my paper
classification eess.SYcs.SYeess.SP
keywords tramlocalizationGNSSmultipathnon-line-of-sightiteratedextendedKalmanfilterGaussianmixturemapconstraintpseudorange
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that urban tram localization can be made substantially more accurate by reconciling raw GNSS pseudorange measurements with the filter's own predicted pseudoranges before the measurement update, and by adding a soft constraint that pulls the posterior position toward the track map. The reconciliation is done per satellite by moment-matching a Gaussian mixture of the measured and predicted pseudorange, which shifts biased measurements toward the prediction and inflates their covariance when they disagree. Combined with the track constraint inside an iterated extended Kalman filter, the method reduces simulated ground-truth RMSE from $13.3$ m to $6.0$ m and real-data distance-to-track RMSE from $8.55$ m to $0.91$ m compared with the baseline. This matters because multipath and non-line-of-sight reception in dense urban areas corrupt several satellites at once, and classical integrity monitoring with multiple fault hypotheses scales combinatorially, whereas the proposed scheme keeps linear complexity in the number of measurements.

What carries the argument

The object that carries the argument is the distribution-mixing identity of Eqs. (42)-(45) applied to each satellite independently: it collapses the two-component Gaussian mixture of raw measurement and filter prediction into one Gaussian with matched first and second moments. This gives a cheap, linear-in-satellites way to turn a suspected multipath bias into an inflated variance rather than a hard outlier decision. The second load-bearing mechanism is the soft track constraint, a fabricated position measurement near the track network that enters the IEKF update as an additional independent measurement. Together they feed a Gauss-Newton/IEKF master filter whose update is a nonlinear least-squares problem with a constrained line search, so the map constraint and the reconciled GNSS readings are optimized in one step.

What would settle it

Take a vehicle with known ground truth, place it under persistent non-line-of-sight reception from one contiguous sky segment for several minutes, and compare the mixing-based filter with and without map aid against a baseline that rejects satellites whose pseudorange residual exceeds a threshold. If the mixing filter without map aid diverges faster or recovers more slowly than the baseline after the NLOS episode ends, the prediction-feedback entrenchment is real; the paper's Table II already hints at this with $9.29$ m vs $8.55$ m.

Watch

Extended reading notes

Core claim

The central claim is that a per-satellite distribution-mixing step, rather than rejecting or reweighting outliers, is enough to make an iterated extended Kalman filter resilient to urban GNSS multipath when the vehicle is known to move on a tram track. For each satellite, the measured pseudorange $\bar{z}_k$ and the predicted pseudorange $\hat{z}_k=g_k(\hat{x})$ are fused into a mixed Gaussian with mean $\tilde{z}_k=\mu_a\bar{z}_k+\mu_b\hat{z}_k$ and variance $\tilde{R}_k=\mu_a(\bar{R}_k+(\bar z_k-\tilde z_k)^2)+\mu_b(\hat R_k+(\hat z_k-\tilde z_k)^2)$, where the weights are inverse-variance proportions. This makes inconsistent pseudoranges contribute less to the position fix while their inflated covariance widens the GNSS position uncertainty, which the IEKF then combines with a fabricated track measurement through a Gauss-Newton MAP update with line search. The paper reports that both features together give the lowest error in simulation and in a 1430-second real tram run.

Load-bearing premise

The load-bearing assumption is that the filter's predicted pseudorange for each satellite, taken from the current state estimate, is an accurate enough reference to pull raw measurements toward; if the state prior is already biased, the mixing loop can lock in that bias instead of correcting it.

Editorial extensions

If this is right

  • In simulation, the full method lowers ground-truth RMSE from $13.3$ m (baseline) to $6.0$ m; on real data, the distance-to-track RMSE drops from $8.55$ m to $0.91$ m, below the $5.32$ m of the commercial receiver solution.
  • Using the mixed distribution without the map aid is worse than the baseline on real data ($9.29$ m vs $8.55$ m), so the track constraint is not a refinement but a necessary counterweight to the low-pass effect of mixing.
  • The resulting position estimates are smoother, with distances between consecutive samples that resemble a realistic tram speed profile.
  • Because the per-satellite mixing is independent and linear in the number of measurements, the approach scales to constellations with many satellites without the combinatorial explosion of multiple-fault RAIM.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • One implication the authors do not pursue is that the mixing step is essentially a shrinkage estimator toward the predicted pseudorange, so its strength could be tuned by a single shrinkage parameter instead of per-satellite covariance ratios, which would make the feedback risk easier to diagnose.
  • The method's failure mode suggests a testable guard: when the filter's own predicted position is biased, mixing entrenches rather than corrects. A natural extension is to gate the mixing with a $\chi^2$ test on the residual $\bar{z}_k-\hat{z}_k$ and skip mixing for satellites that are consistent, which the authors list as future work.
  • The same machinery could transfer to other fixed-route vehicles, such as buses or trains, where the track or route map provides the same low-dimensional constraint that compensates for mixing-induced drift.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes a tram localization method combining an iterated extended Kalman filter (IEKF) with two enhancements: per-satellite Gaussian-mixture reconciliation of GNSS pseudoranges (Section III-B.2, Eqs. (42)-(47)) and a soft constraint that pulls the state estimate toward a predefined track map (Section II-B, Eqs. (21)-(23)). The authors claim that the combined method improves positioning accuracy over a baseline IEKF in both simulation and real-world experiments, while maintaining linear complexity in the number of GNSS measurements. The simulation (Section IV-A, Table I) uses independent ground truth and reports RMSE reduction from 13.3 m to 6.0 m. The real-data experiment (Section IV-B, Table II) reports a reduction from 8.55 m to 0.91 m in RMS distance to an OpenStreetMap track, using the same map as both the soft-constraint input and the accuracy reference. The paper concludes that both features together give the best performance, and it suggests future extensions such as IMM filtering and chi-squared-based outlier rejection.

Significance. If the central claim holds, the contribution is a moderately useful, computationally efficient heuristic for urban GNSS multipath mitigation in rail vehicles. The derivation of the IEKF and the moment-matching equations is internally consistent, and the simulation with independent ground truth constitutes a genuine, if limited, positive result. The linear-complexity property is a practical advantage over multiple-hypothesis RAIM approaches. However, the real-data evaluation is confounded by the circular use of the track map as both constraint and reference, and the paper does not provide independent validation of the mixing step's predicted-pseudorange reference. As a proof-of-concept the paper is acceptable, but the evidence for the headline real-world accuracy gain is not yet established.

major comments (4)
  1. [Section IV-B, Table II] The real-data accuracy metric is the RMS perpendicular distance of position estimates from the OpenStreetMap track, but the same track map is used to construct the soft constraint in Section II-B (Eqs. (21)-(23)). The comparison between 'with soft constraint' and 'without soft constraint' is therefore partly circular: pulling estimates toward the track trivially reduces their perpendicular distance to it. The large gap (8.55 m vs 1.19 m) is a structural consequence of the constraint, not evidence of positioning accuracy. To substantiate the real-data claim, the authors should evaluate against an independent reference (e.g., surveyed ground truth, or a map source not used in the filter) or at least report along-track vs cross-track errors to separate the constraint effect from genuine localization improvement.
  2. [Section III-B.2, Eqs. (42)-(47)] The distribution-mixing step uses the filter's own predicted pseudorange z_hat_k as the reference for each satellite, with weights determined by the ratio of covariances. If the state prior is biased, the mixing pulls raw measurements toward the biased prediction, forming a feedback loop that can entrench the error. The paper itself observes this in Table II: without the map aid, the mixed filter gives 9.29 m versus 8.55 m for the unmixed baseline. This indicates that the mixing step is not independently beneficial; its apparent advantage in the 'with map aid' column (0.91 m vs 1.19 m) may be an interaction with the soft constraint rather than a standalone improvement. The authors should validate the predicted pseudorange distribution as a reliable reference, for example by evaluating the mixing alone against independent ground truth in a scenario with prolonged multipath.
  3. [Table II and Figure 6] The key real-data comparison between the full method (0.91 m) and the map-aided baseline without mixing (1.19 m) is a difference of only 0.28 m, obtained from a single 1430 s run with no error bars, no repeated trials, and no statistical significance test. Given the circularity of the metric and the single-run nature of the experiment, this difference may not be meaningful. The paper should provide either multiple runs with variability statistics, or a clear demonstration that the difference is reproducible and exceeds the noise level of the measurement.
  4. [Section IV-A, Table I] The simulation uses a single straight-line scenario with five satellites (three LOS, two NLOS) and does not represent the dense-urban environment (narrow streets, prolonged multipath, multi-constellation reception) that motivates the paper. The independent-ground-truth simulation is a strength, but its scope is narrow: it shows the method works in a simple synthetic case, not that it addresses the urban-canyon regime. A more realistic simulation with an urban canyon geometry and a larger constellation would materially strengthen the central claim.
minor comments (6)
  1. [Section III-B.1] Typo: 'interated' should be 'iterated' in the sentence 'An interated extended Kalman filter was used...'.
  2. [Section V] In the final paragraph, 'constraint the state prediction step' should be 'constrain the state prediction step'.
  3. [Section III-B.2, Eq. (47)] Equation (47) writes the predicted pseudorange variance as L^T P_{k|k-1} L, but P is the full state covariance (8x8) and L is a 3-vector. The expression should use the full measurement Jacobian (as in Eq. (39)) with appropriate zero blocks for velocity and clock states, or the notation should be clarified.
  4. [Section III-B.2, Eq. (41)] The symbol 'CN o' appears both in text and in Eq. (41) with inconsistent formatting; it should be written as C/N0 (carrier-to-noise density ratio) throughout.
  5. [Section II-B] The soft-constraint hyperparameters are said to be tuned for performance, but their values are not reported in this paper. For reproducibility, the authors should either specify the values used in the experiment or explicitly state that they are taken from [15] and give the relevant values there.
  6. [Section IV-B] The text refers to 'GPS' in Figure 3 while the real-data experiment uses GPS, BeiDou, and Galileo observables. The figure label should be updated to 'GNSS' to avoid confusion.

Circularity Check

1 steps flagged · score 6.0 of 10

Real-data evaluation is self-referential: the track map is both the soft-constraint input and the accuracy reference, so Table II's map-aided gain is partly by construction.

  1. self definitional [Section II-B, Section IV-B, Table II and abstract]
    "Perpendicular distances of the position estimates from the track according to the map were calculated. Table II shows the root-mean-squared distance from the track map. ... The aforementioned heuristic technique fabricates a new position measurement which always lies in the vicinity of tracks. When such measurement is incorporated into the filtering scheme according to (21)-(23), the posterior estimate is driven closer to the track network."

    The same OpenStreetMap track is used in two roles: (i) in Section II-B the soft constraint fabricates a position measurement near the track to pull the posterior estimate toward the track network, and (ii) in Section IV-B the reported real-data 'accuracy' is the perpendicular distance from the estimates to that same track map. The with-constraint configurations by definition minimize distance-to-track, so the Table II gap (8.55 m without vs 1.19 m/0.91 m with constraint) measures compliance with the constraint, not independent positioning error. The abstract concedes this by using orthogonal projections onto the track 'as the true position, which is unavailable in the realistic scenario.' This makes the real-data validation of map aid partially tautological.

full rationale

The paper has independent content: the simulation (Section IV-A, Table I) evaluates RMS distance from known ground-truth positions, and all IEKF, mixing (Eqs. 42-47), and constraint equations are explicitly specified, so the algorithmic concept does not reduce to its inputs by construction. The circularity is confined to the real-data evaluation: Table II uses distance to the OpenStreetMap track as the accuracy metric while Section II-B uses that same track map to generate the soft constraint that pulls estimates toward the track. The large improvement from 'without soft constr.' to 'with soft constr.' in Table II is therefore a structural consequence of the constraint's definitional goal rather than evidence of absolute positioning accuracy. The remaining comparison relevant to the paper's novel mixing contribution is 0.91 m vs 1.19 m, both with the constraint active; that difference is not itself forced by construction, but it is still measured against the constraint's own reference map. The mixing block also uses the filter's predicted pseudorange as the reconciliation reference (Eqs. 44-47), creating a feedback between predicted and reconciled measurements; the paper itself observes that without map aid mixing degrades RMSE (9.29 m vs 8.55 m). This is a legitimate limitation and a self-referential risk, but it is a performance concern rather than a derivation-circularity, so it is not counted as a separate circular step. Overall: partial circularity of the real-data headline claim, with independent simulation support keeping the score at 6.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The central method adds no new physical entities and fits no constants to the evaluation data, but it depends on several modeling choices: the moment-matched Gaussian mixture, the soft map constraint from [15] with undisclosed hyperparameters, and the standard but simplified motion and tropospheric models. The constants in the measurement variance model are taken from prior literature. The main free parameters of the filter are not reported, which limits the ledger's completeness and reproducibility.

free parameters (3)
  • Empirical GNSS variance model constants (T, F, A, a) = T=50, F=10, A=30, a=40
    Eq. (41) sets pseudorange noise variance using these hand-chosen constants from [27,28]; they directly control measurement weights and the mixing behavior.
  • Soft constraint hyperparameters from [15] = not disclosed
    Section II-B says the track-attraction heuristic introduces additional hyperparameters which have to be tuned; their values are not given, so the map constraint behavior is not fully specified.
  • Process noise covariance and prior covariance = not reported
    Required by Eqs. (16)-(18) for the constant-velocity and clock model; no values or tuning procedure are provided, which affects the predicted pseudorange uncertainty in Eq. (47).
assumptions (5)
  • standard math Gaussian distributions and conditional independence of measurement sources in the MAP fusion and IEKF derivation (Eqs. (1)-(4), (24))
    The master fusion and IEKF derivation rely on Gaussian noise and independence of measurement sources; this is standard but unstated in the experiment setup.
  • domain assumption Constant velocity plus receiver clock error model for tram motion (Eq. (37))
    Tram motion is not constant-velocity, especially at stops and curves; the model is used in both simulation and real data without verification against actual dynamics.
  • domain assumption Tropospheric delay is neglected in the pseudorange model (Section III-A.1)
    The paper justifies the simplification as small relative to multipath, but at low elevation angles tropospheric delay can be meters and may bias pseudoranges.
  • ad hoc to paper Multipath and NLOS errors can be represented by an equivalent Gaussian mixture with matched first and second moments (Section III-B.2, Eqs. (42)-(45))
    The reconciliation step replaces a non-Gaussian NLOS error with a moment-matched mixture; the paper does not validate this approximation against actual NLOS error distributions.
  • domain assumption The OpenStreetMap track map is sufficiently accurate and the nearest-waypoint projection is a valid constraint (Section II-B)
    The map is used both as a constraint and as the real-data evaluation reference; map errors or track branches, mentioned only as future work, are not handled.
invented entities (1)
  • Artificial track-constraint position measurement
    purpose: Acts as a soft virtual sensor attracting the IEKF posterior toward the track network
    Section II-B, from [15], fabricates a position measurement from nearby waypoints; it has no physical counterpart and requires tuned hyperparameters, yet it is central to the reported map-aid improvement.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tram Positioning with Map-Enabled GNSS Data Reconciliation." pith.science (2026). https://pith.science/paper/SSJIDIV3

@misc{pith2026250608032,
  author       = {Pith},
  title        = {Pith review of: Tram Positioning with Map-Enabled GNSS Data Reconciliation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SSJIDIV3}},
  note         = {Machine review of arXiv:2506.08032}
}
abstract

This paper presents an approach to tackle the problem of tram localization through utilizing a custom processing of Global Navigation Satellite System (GNSS) observables and the track map. The method is motivated by suboptimal performance in dense urban environments where the direct line of sight to GNSS satellites is often obscured which leads to multipath propagation of GNSS signals. The presented concept is based upon the iterated extended Kalman filter (IEKF) and has linear complexity (with respect to the number of GNSS measurements) as opposed to some other techniques mitigating the multipath signal propagation. The technique is demonstrated both on a simulated example and real data. The root-mean-squared errors from the simulated ground truth positions show that the presented solution is able to improve performance compared to a baseline localization approach. Similar result is achieved for the experiment with real data, while treating orthogonal projections onto the tram track as the true position, which is unavailable in the realistic scenario. This proof-of-concept shows results which may be further improved with implementation of a bank-of-models method or $\chi^2$-based rejection of outlying GNSS pseudorange measurements.

Figures

Figures reproduced from arXiv: 2506.08032 by the authors.

Figure 1
Figure 1. Illustration of the filtering step of the localization scheme without [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. Proposed filtering scheme utilizing GNSS measurements and track [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 5
Figure 5. Illustration of the simulated effect of multipath on the position [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figures from the paper (2 more)
Figure 7
Figure 7. Figure 7: Distance between consecutive position estimates. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Visualized trajectories of the state estimates from the four filters. [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

30 extracted references · 30 canonical work pages

  1. [1]

    Parkinson et al., eds

    Bradford W. Parkinson et al., eds. Global Positioning System: Theory and Applications, Volume II . Washington DC: American Institute of Aeronautics and Astronautics, Jan. 1, 1996. 143-165. ISBN : 978-1- 56347-107-0

    work page 1996

  2. [2]

    GPS Failure Detection by Autonomous Means Within the Cockpit

    Grover Brown and Patrick Y . C. Hwang. “GPS Failure Detection by Autonomous Means Within the Cockpit”. In: Navigation 33.4 (Dec. 1986), pp. 335–353. ISSN : 00281522

    work page 1986

  3. [3]

    A New Improved RAIM Method Based on the Optimally Weighted Average Solution (OW AS) Under the Assumption of a Single Fault

    Y . C. Lee. “A New Improved RAIM Method Based on the Optimally Weighted Average Solution (OW AS) Under the Assumption of a Single Fault”. In: Jan. 20, 2006

    work page 2006

  4. [4]

    NIORAIM Integrity Mon- itoring Performance In Simultaneous Two-Fault Satellite Scenarios

    Patrick Y . Hwang and R. Grover Brown. “NIORAIM Integrity Mon- itoring Performance In Simultaneous Two-Fault Satellite Scenarios”. In: Proceedings of the 18th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2005). Sept. 16, 2005, pp. 1760–1771

    work page 2005

  5. [5]

    A Portfolio Approach to NLOS and Multipath Mitigation in Dense Urban Areas

    P. D. Groves et al. “A Portfolio Approach to NLOS and Multipath Mitigation in Dense Urban Areas”. In: Proceedings of the 26th Inter- national Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013). Sept. 20, 2013, pp. 3231–3247

    work page 2013

  6. [6]

    Multi- Constellation GNSS Performance Evaluation for Urban Canyons Using Large Virtual Reality City Models

    Lei Wang, Paul D. Groves, and Marek K. Ziebart. “Multi- Constellation GNSS Performance Evaluation for Urban Canyons Using Large Virtual Reality City Models”. In: The Journal of Navigation 65.3 (July 2012), pp. 459–476. ISSN : 1469-7785, 0373- 4633

    work page 2012

  7. [7]

    An Optimized Multiple Hypothesis RAIM Algorithm for Vertical Guidance

    Juan Blanch et al. “An Optimized Multiple Hypothesis RAIM Algorithm for Vertical Guidance”. In: Proceedings of the 20th Inter- national Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2007). Sept. 28, 2007, pp. 2924–2933

    work page 2007

  8. [8]

    Demonstrations of Multi-constellation Ad- vanced RAIM for Vertical Guidance Using GPS and GLONASS Signals

    Myungjun Choi et al. “Demonstrations of Multi-constellation Ad- vanced RAIM for Vertical Guidance Using GPS and GLONASS Signals”. In: Proceedings of the 24th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2011). Sept. 23, 2011, pp. 3227–3234

    work page 2011

Show all 30 references
  1. [9]

    Receiver Integrity Monitoring in Case of Multiple Failures

    I. Martini, R. Wolf, and G. W. Hein. “Receiver Integrity Monitoring in Case of Multiple Failures”. In: Proceedings of the 19th Interna- tional Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2006). Sept. 29, 2006, pp. 2608–2620

    2006

  2. [10]

    Integrity in Urban and Road Envi- ronments and Its Use in Liability Critical Applications

    J. Cosmen-Schortmann et al. “Integrity in Urban and Road Envi- ronments and Its Use in Liability Critical Applications”. In: 2008 IEEE/ION Position, Location and Navigation Symposium . 2008 IEEE/ION Position, Location and Navigation Symposium. May 2008, pp. 972–983

    2008

  3. [11]

    Lane- Level Integrity Provision for Navigation and Map Matching With GNSS, Dead Reckoning, and Enhanced Maps

    Rafael Toledo-Moreo, David Betaille, and Franc ¸ois Peyret. “Lane- Level Integrity Provision for Navigation and Map Matching With GNSS, Dead Reckoning, and Enhanced Maps”. In: IEEE Trans- actions on Intelligent Transportation Systems 11.1 (Mar. 2010), pp. 100–112. ISSN : 1558-0016

    2010

  4. [12]

    Map-Aided Integrity Monitoring of a Land Vehicle Navigation System

    Nagendra R. Velaga et al. “Map-Aided Integrity Monitoring of a Land Vehicle Navigation System”. In: IEEE Transactions on Intelligent Transportation Systems 13.2 (June 2012), pp. 848–858. ISSN : 1558-0016

    2012

  5. [13]

    P-RANSAC: An Integrity Monitoring Approach for GNSS Signal Degraded Scenario

    Gaetano Castaldo et al. “P-RANSAC: An Integrity Monitoring Approach for GNSS Signal Degraded Scenario”. In: International Journal of Navigation and Observation 2014 (Sept. 23, 2014), pp. 1–11. ISSN : 1687-5990, 1687-6008

    2014

  6. [14]

    Receiver Autonomous Integrity Monitoring of GNSS Signals for Electronic Toll Collection

    Daniel Sal ´os et al. “Receiver Autonomous Integrity Monitoring of GNSS Signals for Electronic Toll Collection”. In: IEEE Transactions on Intelligent Transportation Systems 15.1 (Feb. 2014), pp. 94–103. ISSN : 1558-0016

    2014

  7. [15]

    Tram Localization Using Soft-Constrained Iterated Kalman Filter with Optimal Step Size Control

    V ´ıt Fanta, Vladim´ır Havlena, and Zdenˇek Hur´ak. “Tram Localization Using Soft-Constrained Iterated Kalman Filter with Optimal Step Size Control”. In: 2024 IEEE International Conference on Multisen- sor Fusion and Integration for Intelligent Systems (MFI) . 2024 IEEE Intern...

    2024

  8. [16]

    Performance Evaluation of Iter- ated Extended Kalman Filter with Variable Step-Length

    Jind ˇrich Havl´ık and Ondˇrej Straka. “Performance Evaluation of Iter- ated Extended Kalman Filter with Variable Step-Length”. In: Journal of Physics: Conference Series 659 (Nov. 19, 2015), p. 012022. ISSN : 1742-6588, 1742-6596

    2015

  9. [17]

    Extended Kalman Filter Modifications Based on an Optimization View Point

    Martin A. Skoglund, Gustaf Hendeby, and Daniel Axehill. “Extended Kalman Filter Modifications Based on an Optimization View Point”. In: 2015 18th International Conference on Information Fusion (Fusion). 2015 18th International Conference on Information Fusion (Fusion). July 20...

    2015

  10. [18]

    Optimal State Estimation: Kalman, H [Infinity] and Nonlinear Approaches

    Dan Simon. Optimal State Estimation: Kalman, H [Infinity] and Nonlinear Approaches . Hoboken, N.J: Wiley-Interscience, 2006. 526 pp. ISBN : 978-0-471-70858-2

    2006

  11. [19]

    NAVSTAR GPS Space Segment/Navigation User Seg- ment Interfaces

    Tony Anthony. NAVSTAR GPS Space Segment/Navigation User Seg- ment Interfaces

  12. [20]

    Relativity in the Global Positioning System

    Neil Ashby. “Relativity in the Global Positioning System”. en. In: Living Reviews in Relativity 6.1 (Dec. 2003), p. 1. ISSN : 2367-3613, 1433-8351

    2003

  13. [21]

    Ionospheric Time-Delay Algorithm for Single- Frequency GPS Users

    John Klobuchar. “Ionospheric Time-Delay Algorithm for Single- Frequency GPS Users”. In: IEEE Transactions on Aerospace and Electronic Systems AES-23.3 (May 1987), pp. 325–331. ISSN : 0018- 9251

    1987

  14. [22]

    A new version of the NeQuick ionosphere electron density model

    B. Nava, P. Co ¨ısson, and S.M. Radicella. “A new version of the NeQuick ionosphere electron density model”. en. In: Journal of Atmospheric and Solar-Terrestrial Physics 70.15 (Dec. 2008), pp. 1856–1862. ISSN : 13646826

    2008

  15. [23]

    Aided navigation: GPS with high rate sensors

    Jay Farrell. Aided navigation: GPS with high rate sensors. Electronic engineering. OCLC: ocn212908814. New York: McGraw-Hill, 2008. ISBN : 978-0-07-149329-1

    2008

  16. [24]

    Influence of the inhomogeneous troposphere on GNSS positioning and integer ambiguity resolution

    Hongyang Ma et al. “Influence of the inhomogeneous troposphere on GNSS positioning and integer ambiguity resolution”. en. In: Advances in Space Research 67.6 (Mar. 2021), pp. 1914–1928. ISSN : 02731177

    2021

  17. [25]

    J.P. Collins. Assessment and Development of a Tropospheric Delay Model for Aircraft Users of the Global Positioning System. Technical report (University of New Brunswick. Department of Geodesy and Geomatics Engineering). University of New Brunswick, 1999

    1999

  18. [26]

    The TropGrid2 standard tropospheric correction model

    Torben Sch ¨uler. “The TropGrid2 standard tropospheric correction model”. en. In: GPS Solutions 18.1 (Jan. 2014), pp. 123–131. ISSN : 1080-5370, 1521-1886

    2014

  19. [27]

    goGPS: open source soft- ware for enhancing the accuracy of low-cost receivers by single- frequency relative kinematic positioning

    Eugenio Realini and Mirko Reguzzoni. “goGPS: open source soft- ware for enhancing the accuracy of low-cost receivers by single- frequency relative kinematic positioning”. In: Measurement Science and Technology 24.11 (Nov. 2013), p. 115010. ISSN : 0957-0233, 1361-6501

    2013

  20. [28]

    Improved weighting scheme using consumer- level GNSS L5/E5a/B2a pseudorange measurements in the urban area

    Hoi-Fung Ng et al. “Improved weighting scheme using consumer- level GNSS L5/E5a/B2a pseudorange measurements in the urban area”. en. In: Advances in Space Research 66.7 (Oct. 2020), pp. 1647–1658. ISSN : 02731177

    2020

  21. [29]

    Conversion of Earth-centered Earth-fixed coordinates to geodetic coordinates

    J. Zhu. “Conversion of Earth-centered Earth-fixed coordinates to geodetic coordinates”. In: IEEE Transactions on Aerospace and Electronic Systems 30.3 (July 1994), pp. 957–961. ISSN : 00189251

    1994

  22. [30]

    Planet Dump Retrieved from https://planet.osm.org . 2024

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.