The ambiguity-resolved detector (ARD) is developed for the validation of high-precision Global Navigation Satellite Systems (GNSS) mixed-integer observation models. The ARD releases the high success rate restriction of resolving the float ambiguities to integers, enabling ambiguity resolution to contribute to model validation even if the resolution success rate is not close to one. However, there are no closed-form expressions for the distribution of the ARD test statistic. The ARD critical value can only be obtained by Monte Carlo simulation, which requires heavy computations. To reduce the effort of applying the ARD, we provide a lookup table for the critical value of the ARD that utilizes the integer least-squares (ILS) estimator to resolve the ambiguities (ARDILS). We obtain 263 408 ARDILS critical values corresponding to GNSS observation models that vary in the precision of observables, satellite geometry, constellation, and the number of signal frequencies. We treat critical values as functions of the integer bootstrapping success rate and conduct curve fitting with second-order polynomials. The polynomial parameters are provided in the lookup table with which the ARDILS critical values can be simply computed rather than simulated. Numerical experiments demonstrate that the critical values computed with the lookup table provide significance levels close to the specified ones. The ARDILS can be implemented easily and efficiently with the lookup table. The lookup table with the polynomial coefficients to compute the ARD critical values can be downloaded from (Yin, 2024).

The analysis of positioning safety often employs a probability-based formulation. This approach quantifies the probability of positioning failure, which is the probability of the position estimator being outside a safety-region, and compares it against an application specific requirement. The design of positioning algorithms for safety-critical applications, such as automated/autonomous vehicles, should consider the dependence between parameter or state estimation and statistical hypothesis testing for model misspecifications in the evaluation of positioning safety. If this dependence is not considered, as this article shows, the conclusions drawn from the positioning safety analysis might be overly-optimistic. Therefore, this article focuses on the aforementioned dependence through a vehicle positioning scenario based on an Extended Kalman Filter (EKF) and the Detection, Identification, and Adaptation (DIA) method for misspecifications in the motion and measurement models. Grounded in the distributional theory for the DIA method, our positioning safety analysis utilizes the conditional probability density functions (PDFs) of the combined EKF and DIA position error, which are generally nonnormal. We compute the probability of vehicle positioning failure in two cases 1) when the dependence is considered and 2) when it is not, to quantify the over-optimism introduced by ignoring this dependence. Finally, we present our conclusions and recommendations.

Positioning technologies are widely used in automotive, aviation, rail, and maritime safety-critical applications. Therefore, the computation of the probability of positioning failure for vehicles, which is the probability that the position estimator is outside a safety region, is of interest for positioning safety analyses. Since parameter estimation and statistical hypothesis testing for model misspecifications are commonly employed in positioning algorithms, the resulting position estimator is conditioned on the statistical hypothesis testing outcome. Hence, the probability density function (PDF) of the vehicle position estimator that accounts for the dependence between the two inference concepts should be used in the computations. In this contribution, we propose a method to compute the probability of positioning failure using the PDF of the vehicle position estimator, which accounts for the aforementioned dependence and is based on rare event simulation techniques, specifically Importance Sampling and the Cross-Entropy method. We apply the proposed method to a satellite-based positioning scenario, in decimeter precision, of an automated vehicle. The results show that the proposed method enables extensive positioning safety-analyses giving insights that can be used in the development of positioning algorithms and deciding whether safety targets and/or requirements are met. Finally, we discuss some limitations of the method and propose several further improvements.

Navigation with radio-signals using GNSS or a terrestrial positioning system in urban environments is susceptible to multipath propagation, which can severely degrade positioning accuracy. In a Line-of-Sight (LOS) multipath channel, the received signal is composed of a direct path component and a sum of time-shifted and attenuated replicas of the transmitted signal. When these multipath components are not accounted for in the time-delay estimation (TDE) model, they may introduce substantial estimation bias. For positioning, only the first arriving path is of interest. Therefore, it is crucial to focus on estimating the reflections that most significantly affect the TDE of this primary path, while ignoring others with negligible impact. To reduce the impact of close-in multipath in TDE, we propose Maximum Likelihood estimators that account for the strongest reflections, with models considering either one or two multi-path components. The Maximum Likelihood Estimation (MLE) problem is optimized using the Space Alternating Generalized Expectation-Maximization (SAGE) method. To reduce computational load, the delay search space for each path is constrained based on the maximum bias observed in the multipath error envelope (MPEE). To assess the ranging accuracy for the various MLE estimators that account for multiple paths, we utilize a synthetically generated channel based on the Saleh-Valenzuela model. Additionally, we benchmark the positioning performance of these estimators using channel impulse responses recorded with a terrestrial positioning prototype system tested at The Green Village on the TU Delft campus.

Teunissen (J Geod 98(83):1–16, 2024) proposed the ambiguity-resolved (AR) detection theory for GNSS mixed-integer model validation. In this contribution, we study the performance of the AR detector through analysis and simulation experiments and compare it with the ambiguity-float (AF) and ambiguity-known (AK) detectors. We describe how the detectors can be implemented and how to evaluate their performance by computing the power as functions of the model misspecifications’ size. We present two simulation experiments with single- and dual-frequency GPS models and demonstrate that the AR detector can provide a larger detection power than the AF detector, even if the success rate is not close to one. Then, we obtain power functions over 25 user locations with five observation models and 72 satellite geometries per location per model. We find that the AR detector increases the detection probability of ionosphere and troposphere delays by 47% and 60% on average when the success rate is larger than 97.5% and the level of significance is 0.01. We also find the AR detection power to be larger than that of the AF detector in case of multi-dimensional misspecifications.

Ambiguity resolution plays a critical role in fast and high-precision applications of the Global Navigation Satellite System (GNSS). The parameter estimation of high-precision GNSS can benefit from ambiguity resolution when its success rate is very close to 1 (e.g., larger than 0.995); otherwise it is better, in order to avoid a substantial probability of incorrect resolution, to ignore the integer property of the ambiguity, and use the float solution. Nonetheless, model validation and fault detection can still benefit from the integer property of the ambiguity with relatively low ambiguity resolution success rates (e.g. between 0.8 and 0.995) by applying the ambiguity-resolved (AR) detector test statistic based on the ambiguity-resolved residual. Due to the integer property of the resolved (so-called fixed) ambiguity estimator, the distribution of the ambiguity-resolved residual cannot be evaluated analytically. Consequently, the critical value of the AR detector has to be obtained numerically via Monte Carlo simulation of the quantile. Due to the inherent uncertainty in the Monte Carlo simulation process, the implementation of the AR detector also needs to evaluate the uncertainty associated with the simulated critical value. If the simulation uncertainty is large, the actual significance level of the detector may deviate significantly from the target value. In this study, we first describe the process of simulating the samples of the AR test statistic and obtaining the AR critical value for a given significance level through Monte Carlo simulation of the quantile. A histogram of the AR test statistic samples will be shown as an example to illustrate the irregular shape of the distribution of this test statistic. Furthermore, we introduce three methods that can be used to evaluate the uncertainty of the simulated critical value: 1) variance based on the asymptotic normality of the Monte Carlo quantile estimator, 2) confidence interval based on a distribution-free approach, and 3) variance obtained numerically by repeating the simulation. We conduct experiments to compare the above three methods in terms of the consistency between the simulation uncertainties reported by these methods. It will also be shown how the uncertainty of the critical value simulation is affected by the specified significance level. Moreover, we provide the uncertainties of the critical value simulations for nine observation models with various numbers of simulation samples and significance levels, offering insights into the number of samples that should be used for simulating the ARD critical value with the desired uncertainty when applying the AR detector.

GNSS-based positioning plays an important role in safety-critical applications (e.g., automotive, aviation, shipping, rail) where positioning safety is paramount. Safety analyses typically include a probability-based formulation, such as calculating the probability that the position estimator falls outside a safety region (probability of positioning failure). Once this probability is being computed, it can be compared to an application specific requirement to decide whether or not this requirement is met. In this context, we carry out a component-wise analysis of the probability of positioning failure for the Detection, Identification, and Adaptation (DIA) estimator. The probability of positioning failure is formulated based on the DIA estimator’s probability density function (PDF) which accounts for the dependence between parameter estimation and statistical hypothesis testing. The probability of positioning failure can be further expressed in terms of its conditional components via the law of total probability which enables the assessment of the Most Impactful Components (MICs). Knowledge about the MICs can be useful to determine the main contributors to the probability of positioning failure. Using a dual GNSS (GPS and Galileo) positioning model for an automated vehicle, we compute the MICs based on the conditional PDFs components of the DIA estimator. Furthermore, we determine the worst-case scenario by computing the maximum total probability of positioning failure over a range of magnitudes for the outliers in the observables and over different orientations of the vehicle. These types of analyses can be useful during the design stage of DIA estimators to verify whether the safety requirements for positioning are met. Lastly, we summarize our contributions and provide an outlook on future work.

MUDE stands for Modelling, Uncertainty and Data for Engineers, a required module in the MSc programs from the faculty of Civil Engineering and Geosciences at Delft University of Technology in the Netherlands.

The current version of the MUDE Textbook can be found at mude.citg.tudelft.nl/book and the most recent "complete" version is mude.citg.tudelft.nl/book/2024. Additional information about the book and its contents can be found on the Credits Page from 2024; technical information about the book and its source code can be found in the README of the GitHub repository TUDelft-MUDE/book. General information about MUDE can be found at mude.citg.tudelft.nl.

This Zenodo record archives the HTML files and provides a DOI for the MUDE Textbook. In general, the GitHub repository github.com/TUDelft-MUDE/book and book URL mude.citg.tudelft.nl/book should be used as primary links for the book, whereas Zenodo is used as an archive and DOI publisher, providing a "permanent" URL. The book is registrered in TU Delft's Research Portal PURE too.

The recommended citation for the MUDE Textbook is provided on the Credits page of the book (link above); the Zenodo recommendation on the side of this page should not be used (neither should the citation in the source code record).

Inspired by the attractive features of least-squares theory in many practical applications, this contribution introduces least-squares-based deep learning (LSBDL). Least-squares theory connects explanatory variables to predicted variables, called observations, through a linear(ized) model in which the unknown parameters of this relation are estimated using the principle of least-squares. Conversely, deep learning (DL) methods establish nonlinear relationships for applications where predicted variables are unknown (nonlinear) functions of explanatory variables. This contribution presents the DL formulation based on least-squares theory in linear models. As a data-driven method, a network is trained to construct an appropriate design matrix of which its entries are estimated using two descent optimization methods: steepest descent and Gauss–Newton. In conjunction with interpretable and explainable artificial intelligence, LSBDL leverages the well-established least-squares theory for DL applications through the following three-fold objectives: (i) Quality control measures such as covariance matrix of predicted outcome can directly be determined. (ii) Available least-squares reliability theory and hypothesis testing can be established to identify mis-specification and outlying observations. (iii) Observations’ covariance matrix can be exploited to train a network with inconsistent, heterogeneous and statistically correlated data. Three examples are presented to demonstrate the theory. The first example uses LSBDL to train coordinate basis functions for a surface fitting problem. The second example applies LSBDL to time series forecasting. The third example showcases a real-world application of LSBDL to downscale groundwater storage anomaly data. LSBDL offers opportunities in many fields of geoscience, aviation, time series analysis, data assimilation and data fusion of multiple sensors.

This paper presents a terrestrial networked positioning system that obtains a reliable time reference from a national time scale realization and distributes it in a prototype to six roadside base stations through a fiber-optic Gigabit Ethernet network. Wireless wideband signals are transmitted by the base stations, thereby enabling positioning by a mobile receiver with an accuracy of one decimeter in a multipath urban environment. The scalability and compatibility of this system with existing telecommunication-network technologies paves the way for wide-area global navigation satellite system-independent back-up systems for timing and positioning with improved coverage and performance. The results presented in this paper are based on research carried out within the scope of a project funded by the Dutch Research Council (NWO, project 13970).

Global navigation satellite systems (GNSS) are widely used for navigation and time distribution^1–3, features that are indispensable for critical infrastructure such as mobile communication networks, as well as emerging technologies such as automated driving and sustainable energy grids^3,4. Although GNSS can provide centimetre-level precision, GNSS receivers are prone to many-metre errors owing to multipath propagation and an obstructed view of the sky, which occur particularly in urban areas where accurate positioning is most needed^1,5,6. Moreover, the vulnerabilities of GNSS, combined with the lack of a back-up system, pose a severe risk to GNSS-dependent technologies⁷. Here we demonstrate a terrestrial positioning system that is independent of GNSS and offers superior performance through a constellation of radio transmitters, connected and time-synchronized at the subnanosecond level through a fibre-optic Ethernet network⁸. Using optical and wireless transmission schemes similar to those encountered in mobile communication networks, and exploiting spectrally efficient virtual wideband signals, the detrimental effects of multipath propagation are mitigated⁹, thus enabling robust decimetre-level positioning and subnanosecond timing in a multipath-prone outdoor environment. This work provides a glimpse of a future in which telecommunication networks provide not only connectivity but also GNSS-independent timing and positioning services with unprecedented accuracy and reliability.

This book provides an introduction, at academic level, into the field of surveying and mapping. The book has been compiled based on hand-outs and readers written for the third-year course Surveying and Mapping, in the bachelor program Civil Engineering at Delft University of Technology. This book covers a wide range of measurement techniques, from land surveying, GPS/GNSS and remote sensing to the associated data processing, the underlying coordinate reference systems, as well as the analysis and visualization of the acquired geospatial information.

GPS is undoubtedly the most popular system providing positioning, navigation and timing (PNT) services to a host of applications, industries and infrastructures. GPS is mass-adopted, has worldwide coverage, has an impressive up-time and can be used with a wide range of receiver devices, featuring low to high cost and low to high precision.

This paper presents a methodology to design a sparse multiband ranging signal with a large virtual bandwidth, from which time delay and carrier phase are estimated by a low complexity multivariate maximum likelihood (ML) method. In the estimation model for a multipath channel, not all reflected paths are considered, and time delay and carrier phase are estimated in a step-wise manner to further reduce the computational load. By introducing a measure of dependence and a measure of bias for a multipath reflection, we analyse the bias, precision and accuracy of time delay and carrier phase estimation. Since these two indicators are determined by the signal spectrum pattern, they are used to formulate an optimization for signal design. By solving the optimization problem, only a few bands from the available signal spectrum are selected for ranging. Consequently, the designed signal only occupies a small amount of signal spectrum but has a large virtual bandwidth and can thereby still offer a high ranging precision with only a small bias, based on the low-complexity simplified ML method. Numerical and laboratory experiments are carried out to evaluate the ranging performance of the proposed estimation method based on sparsely selected signal bands. Relative positioning, in which we only measure a change in position, based on either the time delay estimates or the carrier phase estimates, is presented as a proof-of-concept for precise positioning. The results show that positioning based on only 7 out of 16 signal bands, sparsely placed in the available spectrum, achieves a decimeter level accuracy when using time delay estimates, and a millimeter level accuracy when using carrier phase estimates. Compared with the case of using all available bands, and without largely decreasing the positioning performance, the computational complexity when using the sparse multiband signal can be reduced by about 80%.

For validation and demonstration of high accuracy ranging and positioning algorithms and systems, a wideband radio signal generation and acquisition testbed, tightly synchronized in time and frequency, is needed. The development of such a testbed requires solutions to several challenges. Tight time and frequency synchronization, derived from a centrally distributed time-frequency reference signal, needs to be maintained in the hardware of the transmitter and receiver nodes, and wideband signal acquisition requires sustainable data throughput between the receiver and host PC as well as data storage at GB level. This article presents a testbed for wideband radio signal acquisition, for validation and demonstration of high accuracy ranging and positioning. It consists of multiple Ettus X310 universal software radio peripherals (USRPs) and supports high accuracy (<100 ps) time-deterministic, sustainable signal transmission and acquisition, with a bandwidth up to 320 MHz (in dual channel mode) and frequencies up to 6 GHz. Generation and processing of wideband arbitrary signal waveforms is done offline. To realize these features, radio frequency on chip (RFNoC) compatible HDL units were developed for integration in the X310 SDR platform. Wideband transmission and signal acquisition at a lower duty cycle is applied to reduce the data offloading throughput to the host's personal computer (PC). Benchmarking of the platform was performed to demonstrate sustainable long duration dual channel acquisition. Indoor range measurements with the synchronous operation of the testbed show a decimeter-level accuracy.

Terrestrial positioning systems are being investigated as the complement to the global navigation satellite systems (GNSS), to provide precise and reliable positioning services in a GNSS-challenged environment. In this paper, we present the positioning performance of a ground-based positioning system, in which a multiband OFDM burst is used as a ranging signal to estimate carrier phase, and all transmitters are tightly synchronized by optically distributed time and frequency reference signals. The receiver, like in GNSS, runs on its own clock. An experiment has been carried out in an outdoor living lab environment to demonstrate the flexibility of precise positioning using carrier phase with the proposed ground-based system. During the experiment, the receiver was moved over a trajectory of 17 m forth and back, and acquired the ranging signal for 71 seconds. Without calibrating the different initial phase offsets among the transmitters, we keep the carrier phase cycle ambiguities as float numbers and compute the so called float position solutions. The root mean-squared error (RMSE) of the position solution in East and North direction are 4.22 cm and 4.63 cm, respectively, demonstrating the high-accuracy potential of the proposed burst oriented hybrid optical-wireless terrestrial positioning system.

Smartphone accelerometers and low-cost Global Navigation Satellite System (GNSS) equipment have faced rapid and important advancement, opening a new door to deformation monitoring applications such as landslide, plate tectonics and structural health monitoring (SHM). The precision potential and operational feasibility of the equipment play an important role in the decision making of campaigning for affordable solutions. This paper focuses on the evaluation of the empirical precision, including (auto)time correlation, of a common smartphone accelerometer (Bosch BMI160) and a low-cost dual frequency GNSS reference-rover pair (u-blox ZED-F9P) set to operate at high rates (50 and 5 Hz, respectively). Additionally, a high-rate (5 Hz) GPS-only baseline-based multipath (MP) correction is proposed for effectively removing a large part of this error and allowing to correctly determine the instrumental noise of the GNSS sensor. Furthermore, the benefit of smartphone-based validation for the tracking of dynamic displacements is addressed. The estimated East-North-Up (ENU) precision values (σ^) of ±7.7, 8.1 and 9.6 mms2 are comparable with the declared precision potential (σ) of the smartphone accelerometer of ±8.8mms2. Furthermore, the acceleration noise shows only mild traces of (auto)correlation. The MP-corrected 3D (ENU) empirical precision values of ±2.6, 3.6 and 6.7 mm were found to be better by 30–40% than the straight-out-of box precision of the GNSS sensor, attesting the usefulness of the MP correction. The GNSS sensors output position information with time correlation of typically tens of seconds. The results indicate exceptional precision potential of these low-power-consuming, small-scale, affordable sensors set to operate at a high-rate over small regions. The smartphone-based dynamic displacement validation shows that GNSS data of a low-cost sensor at a 5 Hz sampling rate can be successfully used for tracking dynamic processes.

Time-based ranging accuracy is inversely proportional to the signal bandwidth. A larger the signal bandwidth leads to a higher accuracy of time delay estimation, but more complex hardware is needed. Alternatively, we explore the idea of using multiple narrow signal bands (e.g., 10 MHz of each) to create a large virtual signal bandwidth, which maintains the spectral efficiency but largely improves the ranging accuracy. Considering the impact of multipath, the propagation delay of the LoS path is computed from the estimated channel impulse response (CIR). In this paper, we propose an approach to sparsely select signal bands for ranging and positioning based on convex optimization. The Cramér-Rao lower bound (CRLB) for the propagation delay and gain estimators, as a performance criterion, is employed in the constraint of the optimization. The CRLB is derived in a two-path channel, so that the accuracy and the correlation between the LoS path and the reflection are taken into account. Experiments are conducted in a laboratory environment to illustrate the proposed signal design methodology dedicated for ranging with a sub-decimeter accuracy.

In this contribution, we propose a method for statistically evaluating the risk in a deformation monitoring system. When the structure under monitoring moves beyond tolerance, the monitor system should issue an alert. Only a very small probability is acceptable of the system telling us that no change beyond a critical threshold has taken place, while in reality it has. This probability is referred to as integrity risk. We provide a formulation of integrity risk where the interaction between estimation and testing is taken into account, implying the use of conditional probabilities. In doing so, we assumed different scenarios with the alerts being dependent on both the identified hypothesis and the threat that the estimated size of deformations entails. It is hereby highlighted that a correct risk evaluation requires estimation and testing being considered together, as they are typically intimately linked. In practice, one may, however, find it simpler computation-wise to neglect the estimation–testing link. For this case, we provide an approximation of the integrity risk. This approximation may provide a too optimistic or pessimistic description of the integrity risk depending on the testing procedure and tolerances of the structure at hand. Monitoring systems, besides issuing timely alerts, are also required to provide threat estimates together with their corresponding probabilistic properties. As the testing outcome determines how the threat gets estimated, the threat estimator will then inherit the statistical properties of both estimation and testing. We derive the threat estimator b¯ _j and its probability density function, taking the contributions from combined estimation and testing into account. It is highlighted that although the threat estimator under the identified hypothesis H_j, i.e., b^ _j, is normally distributed, the estimator b¯ _j is not. It is explained that working with b^ _j instead of b¯ _j, thus ignoring the estimation–testing link, may provide a too optimistic description of the threat estimator’s quality. The presented method is illustrated by means of two simple deformation examples.