TY - BOOK

T1 - Improved GNSS Navigation with Chip-scale Atomic Clocks

AU - Krawinkel, Thomas

N1 - Doctoral thesis

PY - 2018

Y1 - 2018

N2 - The determination of position and time as well as navigation by means of a Global Navigation Satellite System (GNSS) is always based on one-way range measurements between satellites and receiver.Synchronization of their time scales is carried out by introducing so-called clock biases with respect to GNSS time.Satellite clock corrections are either provided by the system operator via the navigation message or calculated using clock products of the International GNSS Service. On the receiver side, the clock bias has to be corrected by the user. Due to the poor accuracy and limited long-term stability of the built-in quartz oscillator of a GNSS receiver, the clock bias must be estimated together with the coordinates at each measurement epoch.That is why GNSS-based three-dimensional position determination always requires at least four satellites in view.In addition, high mathematical correlations arise between the up-coordinate and clock bias of the receiver as well as other elevation-dependent error sources. As a consequence, the vertical coordinate can only be determined approximately two to three times less precisely than the horizontal coordinates.If the internal oscillator of the receiver is replaced by a more stable external one, the behavior of the latter can be modeled thanks to its higher frequency stability. An epochwise estimation is not necessary anymore. The physically meaningful prediction of the clock behavior is possible over intervals in which the integrated clock noise is smaller than that of the GNSS observations in use. This approach is referred to as receiver clock modeling (RCM).The development of so-called chip-scale atomic clocks (CSACs) makes the use of a highly stable oscillator in kinematic GNSS applications possible. Thus, connecting a CSAC to a GNSS receiver enables physically meaningful RCM in GNSS navigation based on code observations.In this thesis, the requirements regarding the frequency stability of an oscillator for RCM are investigated by means of two different CSACs.The results of the characterization of their individual frequency stability are discussed in detail, and model parameters for RCM in a Kalman filter are derived thereof.Furthermore, a new real-time applicable modeling approach in a sequential least-squares adjustment is proposed, where the receiver clock biases are modeled by means of a piecewise linear polynomial.The evaluation and validation of this approach are carried out based on a kinematic automotive experiment. In data analysis, GPS and GLONASS L1 CA-code and Doppler observations are used. The impact of receiver clock modeling is assessed by means of typical GNSS performance parameters such as precision, reliability, integrity, continuity and availability of the navigation solution.Improvements of the precision of the vertical coordinate and velocity estimates of ca. 80% and 30%, respectively, are achieved.Internal and external reliability as well as integrity are significantly enhanced, which leads to an overall more robust parameter estimation, especially in regard of statistically reliable outlier detection.It is also shown that using a CSAC in combination with clock modeling enables positioning and navigation with only three satellites in view. This is especially beneficial in areas with high shadowing effects and/or signal attenuation, for example in urban canyons.By implementing the RCM algorithm into a software receiver and using it in a pedestrian experiment, the real-time applicability of the proposed approach is successfully proven.Finally, the benefits of using a CSAC for spoofing detection are demonstrated by means of static experiment.

AB - The determination of position and time as well as navigation by means of a Global Navigation Satellite System (GNSS) is always based on one-way range measurements between satellites and receiver.Synchronization of their time scales is carried out by introducing so-called clock biases with respect to GNSS time.Satellite clock corrections are either provided by the system operator via the navigation message or calculated using clock products of the International GNSS Service. On the receiver side, the clock bias has to be corrected by the user. Due to the poor accuracy and limited long-term stability of the built-in quartz oscillator of a GNSS receiver, the clock bias must be estimated together with the coordinates at each measurement epoch.That is why GNSS-based three-dimensional position determination always requires at least four satellites in view.In addition, high mathematical correlations arise between the up-coordinate and clock bias of the receiver as well as other elevation-dependent error sources. As a consequence, the vertical coordinate can only be determined approximately two to three times less precisely than the horizontal coordinates.If the internal oscillator of the receiver is replaced by a more stable external one, the behavior of the latter can be modeled thanks to its higher frequency stability. An epochwise estimation is not necessary anymore. The physically meaningful prediction of the clock behavior is possible over intervals in which the integrated clock noise is smaller than that of the GNSS observations in use. This approach is referred to as receiver clock modeling (RCM).The development of so-called chip-scale atomic clocks (CSACs) makes the use of a highly stable oscillator in kinematic GNSS applications possible. Thus, connecting a CSAC to a GNSS receiver enables physically meaningful RCM in GNSS navigation based on code observations.In this thesis, the requirements regarding the frequency stability of an oscillator for RCM are investigated by means of two different CSACs.The results of the characterization of their individual frequency stability are discussed in detail, and model parameters for RCM in a Kalman filter are derived thereof.Furthermore, a new real-time applicable modeling approach in a sequential least-squares adjustment is proposed, where the receiver clock biases are modeled by means of a piecewise linear polynomial.The evaluation and validation of this approach are carried out based on a kinematic automotive experiment. In data analysis, GPS and GLONASS L1 CA-code and Doppler observations are used. The impact of receiver clock modeling is assessed by means of typical GNSS performance parameters such as precision, reliability, integrity, continuity and availability of the navigation solution.Improvements of the precision of the vertical coordinate and velocity estimates of ca. 80% and 30%, respectively, are achieved.Internal and external reliability as well as integrity are significantly enhanced, which leads to an overall more robust parameter estimation, especially in regard of statistically reliable outlier detection.It is also shown that using a CSAC in combination with clock modeling enables positioning and navigation with only three satellites in view. This is especially beneficial in areas with high shadowing effects and/or signal attenuation, for example in urban canyons.By implementing the RCM algorithm into a software receiver and using it in a pedestrian experiment, the real-time applicability of the proposed approach is successfully proven.Finally, the benefits of using a CSAC for spoofing detection are demonstrated by means of static experiment.

KW - GNSS

KW - Allan variance

KW - Clock modeling

KW - Software receiver

KW - Spoofing

U2 - 10.15488/4684

DO - 10.15488/4684

M3 - Doctoral thesis

CY - München

ER -