Generation of Corrections to the Time Scale of the Regional Point

GENERATION OF CORRECTIONS TO THE RPKNP TIME SCALE

Key Parameters

Notation Description
\( S^{[10]} \)Vector of weighted observed pseudoranges (signal delay measurements corrected for relativistic and gravitational effects).
\( R \)Vector of weighted computed geometric ranges (true distance between receiver and satellite).
\( A \)Diagonal matrix of weighting coefficients (\( \alpha_{i,j,k} \)), reflecting the contribution of each PNS station.
\( \Delta \tau_1, \dots, \Delta \tau_L \)Vector of clock offset estimates for \( L \) regional PNS stations.
\( \Delta \tau_{\text{GLONASS}} \)Estimate of the "GLONASS-GPS" time scale divergence.
\( K_{\text{GLONASS}} \)Vector including the coefficient \( k_{\text{GLONASS}} \) for applying the GLONASS-GPS time scale correction.
\( S_{i,j,k,3} \)Observed signal delay (pseudorange) from satellite \( j \) of system \( k \) (GPS/GLONASS) to receiver \( i \).
\( \alpha_{i,j,k} \)Weighting coefficient used to account for relativistic and gravitational effects in signal delay measurements.
\( \zeta \)Small correction factor in the formula for \( \alpha_{i,j,k} \).
\( R_{i,j,k} \)Geometric distance between receiver \( i \) and satellite \( j \) of system \( k \).
\( (X_i, Y_i, Z_i) \)Known coordinates of receiver \( i \).
\( (x_{j,i,k}, y_{j,i,k}, z_{j,i,k}) \)Coordinates of satellite \( j \) of system \( k \) relative to receiver \( i \).
\( k_{\text{GLONASS}} \)Coefficient equal to 1 for GLONASS and 0 for GPS, determined by the NSA type.
\( c \)Speed of light in vacuum (299,792,458 m/s).
\( V \)Normal (variance-covariance) matrix for the least squares method, used to determine parameters.
\( \mathbf{W} \)Diagonal matrix of weighting coefficients, with elements (\( 1/\sigma_k^2 \)) depending on measurement precision \( \sigma_k \).
\( \sigma_k \)Standard deviation of measurements for satellites of system \( k \) (GPS or GLONASS).
\( I_4 \)Identity matrix of dimension \( 4 \times 4 \).
\( V_{0,0}, V_{0,1}, V_{1,0}, V_{1,1} \)Blocks of matrix \( V \) used for inversion via the Frobenius formula.
\( \mathbf{H} \)Schur complement, a key component in the Frobenius formula for inverting block matrices.
\( \Delta \dot{\tau}_1, \dots, \Delta \dot{\tau}_L \)Vector of frequency divergence estimates for \( L \) regional PNS stations.
\( \Delta \dot{\tau}_{\text{GLONASS}} \)Estimate of the "GLONASS-GPS" generator frequency divergence.
\( \dot{\mathbf{S}}^{[10]} \)Vector of weighted rates of change of observed pseudoranges (pseudorange rates or Doppler measurements).
\( \dot{\mathbf{R}} \)Vector of weighted rates of change of computed geometric ranges.
\( \dot{R}_{i,j} \)Rate of change of geometric distance between receiver \( i \) and satellite \( j \).
\( (\dot{x}_{j,i}, \dot{y}_{j,i}, \dot{z}_{j,i}) \)Components of the relative velocity of satellite \( j \) with respect to receiver \( i \).

GENERATION OF CORRECTIONS TO THE RPKNP TIME SCALE

1. Generation of frequency-time corrections is performed taking into account known coordinates and velocity vector components of the CCS.

Corrections accounting for the divergence between PNS and GPS time scales \(\left| \begin{matrix} \Delta \tau_1 & \dots & \Delta \tau_L \end{matrix} \right|^T\), as well as estimates of the "GLONASS-GPS" time scale divergence (\(\Delta \tau_{\text{GLONASS}}\)), are determined by solving the following system of equations:

\[ S^{[10]} = R + A \cdot \left| \begin{matrix} \Delta \tau_1 \\ \vdots \\ \Delta \tau_L \end{matrix} \right| + K_{\text{GLONASS}} \cdot \Delta \tau_{\text{GLONASS}}. \]

----------------------------

Where

\[ S^{[10]} = \left| \begin{matrix} \dots \\ S_{1,j,k,3} \cdot \alpha_{1,j,k} \\ \dots \\ S_{2,j,k,3} \cdot \alpha_{2,j,k} \\ \dots \\ S_{L,j,k,3} \cdot \alpha_{L,j,k} \\ \dots \end{matrix} \right|; \quad R = \left| \begin{matrix} \dots \\ R_{1,j,k} \cdot \alpha_{1,j,k} \\ \dots \\ R_{2,j,k} \cdot \alpha_{2,j,k} \\ \dots \\ R_{L,j,k} \cdot \alpha_{L,j,k} \\ \dots \end{matrix} \right|; \quad A = \left| \begin{matrix} \alpha_{1,j,k} & 0 & \cdots & 0 \\ 0 & \alpha_{2,j,k} & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \cdots & \alpha_{L,j,k} \end{matrix} \right|; \]

\( S^{[10]} \) represents weighted observed signal delays, adjusted by \( \alpha_{i,j,k} \) to account for relativistic and atmospheric effects. \( R \) is the geometric distance between stations and satellites, also weighted by \( \alpha_{i,j,k} \). \( A \) is a diagonal matrix where each element \( \alpha_{i,j,k} \) scales the time offset for the \( i \)-th station, ensuring the system reflects individual station contributions.

\[ \alpha_{i,j,k} = \left( 1 - \zeta S_{i,j,k} \right); \]

This coefficient \( \alpha_{i,j,k} \) adjusts signal delays for relativistic and gravitational effects, where \( \zeta \) is a small correction factor and \( S_{i,j,k} \) is the raw signal delay. It ensures accuracy by compensating for the curvature of spacetime near Earth.

\[ R_{i,j,k} = \sqrt{(x_{j,i,k} - X_i)^2 + (y_{j,i,k} - Y_i)^2 + (z_{j,i,k} - Z_i)^2}; \]

\( R_{i,j,k} \) calculates the Euclidean distance between the satellite position \( (x_{j,i,k}, y_{j,i,k}, z_{j,i,k}) \) and the receiver station \( (X_i, Y_i, Z_i) \), serving as the baseline geometric delay in the system.

\[ \mathbf{K}_{\text{GLONASS}} = \left| \begin{array}{c} \dots \\ k_{\text{GLONASS}} \cdot \alpha_{1,j,k} \\ \dots \\ k_{\text{GLONASS}} \cdot \alpha_{2,j,k} \\ \dots \\ k_{\text{GLONASS}} \cdot \alpha_{L,j,k} \\ \dots \end{array} \right|; \]

\( \mathbf{K}_{\text{GLONASS}} \) is a vector that applies the GLONASS-GPS time offset correction, scaled by \( k_{\text{GLONASS}} \) (1 for GLONASS, 0 for GPS) and \( \alpha_{i,j,k} \). This allows the system to selectively adjust for GLONASS time divergences.

\( k_{\text{GLONASS}} \) — coefficient equal to one for GLONASS and zero for GPS, determined by the NSA type \((k)\).

The system of equations is solved in the form:

\[ \begin{vmatrix} \Delta \tau_1 \\ \Delta \tau_2 \\ \vdots \\ \Delta \tau_L \\ \Delta \tau_{\text{GLONASS}} \end{vmatrix} = \frac{1}{c} \cdot V^{-1} \cdot \left( \begin{vmatrix} A \quad \mathbf{K}_{\text{GLONASS}} \end{vmatrix}^T \cdot \mathbf{W} \cdot \left( \mathbf{S}^{[10]} - \mathbf{R} \right) \right). \]

where

\[ c = 299792458 \text{ m/s} \text{ — speed of light}; \]

\[ V = \left| A \quad \mathbf{K}_{\text{GLONASS}} \right|^T \cdot \mathbf{W} \cdot \left| A \quad \mathbf{K}_{\text{GLONASS}} \right| + 10^{-10} \cdot I_4 = \]

\[ = \left| \begin{array}{ccccc} 10^{-10} + \sum\limits_{j,k} \frac{\alpha_{1,j,k}^2}{\sigma_k^2} & 0 & \cdots & 0 & \sum\limits_{j,k} \frac{k_{\text{GLONASS}} \cdot \alpha_{1,j,k}^2}{\sigma_k^2} \\ 0 & 10^{-10} + \sum\limits_{j,k} \frac{\alpha_{2,j,k}^2}{\sigma_k^2} & \cdots & 0 & \sum\limits_{j,k} \frac{k_{\text{GLONASS}} \cdot \alpha_{2,j,k}^2}{\sigma_k^2} \\ \vdots & \vdots & \ddots & \vdots & \vdots \\ 0 & 0 & \cdots & 10^{-10} + \sum\limits_{j,k} \frac{\alpha_{L,j,k}^2}{\sigma_k^2} & \sum\limits_{j,k} \frac{k_{\text{GLONASS}} \cdot \alpha_{L,j,k}^2}{\sigma_k^2} \\ \sum\limits_{j,k} \frac{k_{\text{GLONASS}} \cdot \alpha_{1,j,k}^2}{\sigma_k^2} & \sum\limits_{j,k} \frac{k_{\text{GLONASS}} \cdot \alpha_{2,j,k}^2}{\sigma_k^2} & \cdots & \sum\limits_{j,k} \frac{k_{\text{GLONASS}} \cdot \alpha_{L,j,k}^2}{\sigma_k^2} & 10^{-10} + \sum\limits_{i=1}^{L} \sum\limits_{j,k} \frac{k_{\text{GLONASS}} \cdot \alpha_{i,j,k}^2}{\sigma_k^2} \end{array} \right| ; \]

Each row corresponds to a regional PNS station, while the last row and column correspond to the GLONASS time scale correction. Diagonal elements combine measurement variance and weighting by \( \alpha_{i,j,k} \); off-diagonal terms reflect GLONASS coupling. The regularization term \( 10^{-10} \) on the diagonal prevents singularity of the matrix.
\( V \) is the variance-covariance matrix, combining weighted contributions from \( A \) and \( \mathbf{K}_{\text{GLONASS}} \), with \( \mathbf{W} \) reflecting measurement uncertainties (\( 1/\sigma_k^2 \)). The \( 10^{-10} \cdot I_4 \) term adds a small regularization to ensure matrix invertibility, stabilizing the solution against noise.

\( W \) — diagonal matrix of weighting coefficients, whose elements \( \left( \frac{1}{\sigma_k^2} \right) \) depending on which NSA
(\( k = 0 \) or \( k = 1 \)) they correspond to, equal \( \sigma_{GPS}^{-2} \) or \( \sigma_{\text{GLONASS}}^{-2} \).

\( I_4 \) — identity matrix of dimension \( 4 \times 4 \).

Inversion of matrix \( V \) is performed using the Frobenius formula (often referred to as the block matrix inversion formula using the Schur complement; see MathWorld or Wikipedia).

\[ V^{-1} = \left| \begin{array}{cc} V_{0,0}^{-1} + V_{0,0}^{-1} \cdot V_{0,1} \cdot H^{-1} \cdot V_{1,0} \cdot V_{0,0}^{-1} & -V_{0,0}^{-1} \cdot V_{0,1} \cdot H^{-1} \\ H^{-1} \cdot V_{1,0} \cdot V_{0,0}^{-1} & H^{-1} \end{array} \right|, \]


\[ \text{where} \quad \textbf{H} = \textbf{V}_{1,1} - \textbf{V}_{1,0} \cdot \textbf{V}_{0,0}^{-1} \cdot \textbf{V}_{0,1}; \]

\( \mathbf{H} \) is the Schur complement of block \( \mathbf{V}_{0,0} \) in the matrix \( \mathbf{V} \). It encapsulates the effect of eliminating the regional station parameters from the system, allowing stable inversion by reducing dimensionality when solving for GLONASS-related terms. For more on the Schur complement, see Wikipedia.

\[ \textbf{V}_{0,0}, \textbf{V}_{0,1}, \textbf{V}_{1,0}, \textbf{V}_{1,1} \text{ — blocks of matrix } \textbf{V}: \]

The matrix \( \mathbf{V} \) is partitioned into four blocks: \( \mathbf{V}_{0,0} \) handles intra-station (PNS) covariances, \( \mathbf{V}_{1,1} \) handles the GLONASS variance, and \( \mathbf{V}_{0,1}, \mathbf{V}_{1,0} \) model the cross-correlation between PNS and GLONASS components. This structure is essential for applying block-matrix inversion.

\[ \mathbf{V} \;=\; \left|\, \begin{array}{c|cc} & L & 1 \\ \hline L & \mathbf{V}_{0,0} & \mathbf{V}_{0,1} \\ 1 & \mathbf{V}_{1,0} & \mathbf{V}_{1,1} \end{array} \right|. \]

This matrix layout shows how \( \mathbf{V} \) is divided into submatrices: \( \mathbf{V}_{0,0} \) contains terms for PNS stations; \( \mathbf{V}_{1,1} \) is for GLONASS; cross-terms \( \mathbf{V}_{0,1} \) and \( \mathbf{V}_{1,0} \) describe their interaction.

If any diagonal element of matrix \( \mathbf{V}^{-1} \) exceeds \( 10^8 \), then the corresponding parameter \( (\tau_i, \dot{\tau}_i \text{ or } \tau_{\text{GLONASS}}, \dot{\tau}_{\text{GLONASS}}) \) cannot be determined, and its previously formed value remains unchanged.

This threshold \( 10^8 \) indicates ill-conditioning of the matrix, suggesting that the data is insufficient to resolve certain time offsets or frequency divergences. Retaining prior values prevents propagation of errors.

The values of divergence between PNS and GPS generator frequencies \( (|\Delta \dot{\tau}_1 \dots \Delta \dot{\tau}_L|^T) \), as well as the estimate of "GLONASS-GPS" generator frequency divergence \( (\Delta \dot{\tau}_{\text{GLONASS}}) \), are determined by the formula:

\[ \begin{vmatrix} \Delta \dot{\tau}_1 \\ \Delta \dot{\tau}_2 \\ \vdots \\ \Delta \dot{\tau}_L \\ \Delta \dot{\tau}_{\text{GLONASS}} \end{vmatrix} = \frac{1}{c} \cdot V^{-1} \cdot \left( \begin{vmatrix} A \quad \mathbf{K}_{\text{GLONASS}} \end{vmatrix}^T \cdot \mathbf{W} \cdot \left( \dot{\mathbf{S}}^{[10]} - \dot{\mathbf{R}} \right) \right). \]

where

\[ \dot{S}^{[10]} = \left| \begin{matrix} \dots \\ \dot{S}_{1,j,k,3} \cdot \alpha_{1,j,k} \\ \dots \\ \dot{S}_{2,j,k,3} \cdot \alpha_{2,j,k} \\ \dots \\ \dot{S}_{L,j,k,3} \cdot \alpha_{L,j,k} \\ \dots \end{matrix} \right| \quad \quad \dot{R} = \left| \begin{matrix} \dots \\ \dot{R}_{1,j} \cdot \alpha_{1,j} \\ \dots \\ \dot{R}_{2,j} \cdot \alpha_{2,j} \\ \dots \\ \dot{R}_{m,j} \cdot \alpha_{m,j} \\ \dots \end{matrix} \right|; \]

\( \dot{S}^{[10]} \) and \( \dot{R} \) are vectors of Doppler-derived rates of pseudoranges and geometric distances. Each element is scaled by \( \alpha_{i,j,k} \), maintaining consistency with time-delay weighting. These vectors are essential for determining frequency divergences.

\[ \dot{R}_{i,j} = \frac{ \dot{x}_{j,i} \cdot (x_{j,i} - X_{i}) + \dot{y}_{j,i} \cdot (y_{j,i} - Y_{i}) + \dot{z}_{j,i} \cdot (z_{j,i} - Z_{i}) }{R_{i,j}}. \]

\( \dot{R}_{i,j} \) computes the rate of change of distance based on relative velocities \( (\dot{x}_{j,i}, \dot{y}_{j,i}, \dot{z}_{j,i}) \) between satellite and station, normalized by the distance \( R_{i,j} \). This reflects the Doppler effect contribution to frequency divergences.

Implementation Result: The described algorithm generates time corrections \( \Delta \tau_i \) and frequency divergences \( \Delta \dot{\tau}_i \) for the RPKNP regional positioning system. This implementation applies weighted matrix operations and block inversion techniques to ensure synchronization between regional PNS stations and GPS/GLONASS time scales.