ISB 与ISPB

ISB 与ISPB

1.ISB

1.1 估计法（code phase）

接收机 $r$ 和 卫星 $s$ 在频率 $j$ 上的载波相位观测方程可
$$\begin{gathered} \begin{array}{rl}{\varphi }_{r,j}^s=& {\rho }_r^s+d{t}_r-d{t}^s+{\alpha }_{r,j}^s\\ & +{\lambda }_j\left({\delta }_{r,j}^s+{\phi }_{r,j}-{\delta }_j^s-{\phi }_j^s+N_{r,j}^s\right)+{ϵ}_{r,j}^s\end{array} \\ {p}_{r,j}^s={\rho }_r^s+d{t}_r-d{t}^s+a_{r,j}^s+d_{r,j}^G-d_j^s+e_{r,j}^s \end{gathered}$$
${\varphi }_{r,j}^s$为载波相位观测值，${\rho }_r^s$ 为接收机到卫星的距离；$d{t}_r、d{t}^s$ 分别为接收机端和卫星端钟差，${\alpha }_{r,j}^s$ 为与频率相关的大气层延迟； ${\lambda }_j$ 为波长，${\delta }_{r,j}^s$ 表示接收机相位硬件延迟， ${\phi }_{r,j}$ 接收机初始相位，${\delta }_j^s$ 卫星端相位硬件延迟， ${\phi }_j^s$ 卫星端初始相位； $d_{r,j}^s$接收端伪距的硬件延迟, $d_j^s$卫星端伪距硬件延迟；$N_{r,j}^s$ 为整数模糊度；${ϵ}_{r,j}^s$为包含噪声在内的其他误差。

同一卫星，流动站（下标2）与参考站（下标1）的观测值单差，得：
$$\begin{array}{rl}{\varphi }_{12,j}^s=& {\varphi }_{2,j}^s-{\varphi }_{1,j}^s={\rho }_{12}^s+d{t}_{12}+{\alpha }_{12,j}^s\\ & +{\lambda }_j\left({\delta }_{12,j}^{B2}+{\phi }_{12,j}+N_{12,j}^s\right)+{ϵ}_{12,j}^s\\ p_{12,j}^s=& p_{2,j}^s-p_{1,j}^s={\rho }_{12}^s+d{t}_{12}+a_{12,j}^s+d_{12,j}^{B2}+e_{12,j}^s\end{array}$$
$j=1,\dots ,f$ ，表示信号频率标号， $s=1_{B2},\dots ,m_{B2},$表示追踪到的BDS-2的卫星标号，。 同理对BDS-3有：
$$\begin{array}{rl}{\varphi }_{12,j}^q=& {\varphi }_{2,j}^q-{\varphi }_{1,j}^q={\rho }_{12}^q+d{t}_{12}+{\alpha }_{12,j}^q\\ & +{\lambda }_j\left({\delta }_{12,j}^{B3}+{\phi }_{12,j}+N_{12,j}^q\right)+{ϵ}_{12,j}^q\\ p_{12,j}^q=& p_{2,j}^q-p_{1,j}^q={\rho }_{12}^q+d{t}_{12}+a_{12,j}^q+d_{12,j}^{B3}+e_{12,j}^q\end{array}$$
$q=1_{B3},\dots ,m_{B3}$表示观测到的BDS-3的卫星标号。对于零基线或者短基线而言，我们可以认为两个站对同一卫星同一频率信号的大气层延迟是一样的。如果我们选择一个参考卫星， $s=1_{B2}$ ,$q=1_{B3}$，则有：
$$\begin{gathered} \begin{array}{l}{\varphi }_{12,j}^{1_{B2}s}={\varphi }_{12,j}^s-{\varphi }_{12,j}^{1_{B2}}={\rho }_{12}^{1_{B2}s}+{\lambda }_j{N}_{12,j}^{1_{B2}s}+{ϵ}_{12,j}^{1_{B2}s}\\ p_{12,j}^{1_{B2}s}=p_{12,j}^s-p_{12,j}^{1_{B2}}={\rho }_{12}^{1_{B2}s}+e_{12,j}^{1_{B2}s}\end{array} \\ \begin{array}{rl}{\varphi }_{12,j}^{1_{B3}q}& ={\varphi }_{12,j}^q-{\varphi }_{12,j}^{1_{B3}}={\rho }_{12}^{1_{B3}q}+{\lambda }_j{N}_{12,j}^{1_{B3}q}+{ϵ}_{12,j}^{1_{B3}q}\\ p_{12,j}^{1_{B3}q}& =p_{12,j}^q-p_{12,j}^{1_{B3}}={\rho }_{12}^{1_{B3}q}+e_{12,j}^{1_{B3}q}\end{array} \end{gathered}$$
对于BDS-2和BDS-3的重叠频率（B1I、B3I），我们可以对不同系统的单差间求差，
$$\begin{array}{l}{\varphi }_{12,j}^{1_{B2q}}={\varphi }_{12,j}^q-{\varphi }_{12,j}^{1_{B2}}={\rho }_{12}^{1_{B2}q}+{\lambda }_j\left({\delta }_{12,j}^{B2B3}+N_{12,j}^{1_{1}q}\right)+{ϵ}_{12,j}^{1_{B2q}q}\\ p_{12,j}^{1_{B2}q}=p_{12,j}^q-p_{12,j}^{1_{B2}}={\rho }_{12}^{1_{B2}q}+d_{12,j}^{B2B3}+e_{12,j}^{1_{G}q}\end{array}$$
根据系统间双差结果，$ISB={\delta }_{12,j}^{B2B3}={\delta }_{2,j}^{B2B3}-{\delta }_{1,j}^{B2B3}，ISPB=d_{12,j}^{B2B3}=d_{2,j}^{B2B3}-d_{1,j}^{B2B3}$, 分别表示BDS-2和BDS-3的相位和伪距的系统间偏差(BDS-2和BDS-3特定频率信号在同一个接收机上不同的硬件延迟)，$r=1,2$。

Characterization of between-receiver GPS-Galileo inter-system biases and their effect on mixed ambiguity resolution Dennis Odijk • Peter J. G. Teunissen

1.2 ISB的随机模型

五系统 GNSS（GPS + GLONASS + BDS + Galileo + QZSS）非差非组合模型（引入ISB）
$$\begin{array}{rl}& \{\begin{array}{l}{p}_{r,j}^{s,G}={\mathbf{u}}_r^{s,G}\cdot \mathbf{x}+cd{\bar{t}}_r^G+{\mathrm{M}\mathrm{w}}_r^{s,G}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r+{\gamma }_j^G\cdot {\bar{I}}_{r,1}^{s,G}+{\epsilon }_{r,j}^{s,G}\\ p_{r,j}^{s,R}={\mathbf{u}}_r^{s,R}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{ISB}}_r^R+c{\Theta }_{r,j}^{s,R}+{\mathrm{M}\mathrm{w}}_r^{s,R}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r+{\gamma }_j^R\cdot {\bar{I}}_{r,1}^{s,R}+{\epsilon }_{r,j}^{s,R}\\ p_{r,j}^{s,C}={\mathbf{u}}_r^{s,C}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{I}\mathrm{S}\mathrm{B}}_r^C+{\mathrm{M}\mathrm{w}}_r^{s,C}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r+{\gamma }_j^c\cdot {\bar{I}}_{r,1}^{s,C}+{\epsilon }_{r,j}^{s,C}\\ p_{r,j}^{s,E}={\mathbf{u}}_r^{s,E}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{I}\mathrm{S}\mathrm{B}}_r^E+{\mathrm{M}\mathrm{w}}_r^{s,E}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r+{\gamma }_j^E\cdot {\bar{I}}_{r,\mathrm{l}}^{s,E}+{\epsilon }_{r,j}^{s,E}\\ p_{r,j}^{s,J}={\mathbf{u}}_r^{s,J}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{I}\mathrm{S}\mathrm{B}}_r^J+{\mathrm{M}\mathrm{w}}_r^{s,J}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r+{\gamma }_j^J\cdot {\bar{I}}_{r,1}^{s,J}+{\epsilon }_{r,j}^{s,J}\end{array}\\ & \{\begin{array}{l}{l}_{r,j}^{s,G}={\mathbf{u}}_r^{s,G}\cdot \mathbf{x}+cd{\bar{t}}_r^G+{\mathrm{M}\mathrm{w}}_r^{s,G}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r-{\gamma }_j^G\cdot {\bar{I}}_{r,1}^{s,G}+{\lambda }_j^{s,G}\cdot {\bar{N}}_{r,j}^{s,G}+{\xi }_{r,j}^{s,G}\\ l_{r,j}^{s,R}={\mathbf{u}}_r^{s,R}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{I}\mathrm{S}\mathrm{B}}_r^R+{\mathrm{M}\mathrm{w}}_r^{s,R}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r-{\gamma }_j^R\cdot {\bar{I}}_{r,1}^{s,R}+{\lambda }_j^{s,R}\cdot {\bar{N}}_{r,j}^{s,R}+{\xi }_{r,j}^{s,R}\\ l_{r,j}^{s,C}={\mathbf{u}}_r^{s,C}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{I}\mathrm{S}\mathrm{B}}_r^C+{\mathrm{M}\mathrm{w}}_r^{s,C}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r-{\gamma }_j^c\cdot {\bar{I}}_{r,1}^{s,C}+{\lambda }_j^{s,C}\cdot {\bar{N}}_{r,j}^{s,C}+{\xi }_{r,j}^{s,C}(4.4)\\ l_{r,j}^{s,E}={\mathbf{u}}_r^{s,E}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{I}\mathrm{S}\mathrm{B}}_r^E+\mathrm{M}{\mathrm{w}}_r^{s,E}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r-{\gamma }_j^E\cdot {\bar{I}}_{r,1}^{s,E}+{\lambda }_j^{s,E}\cdot {\bar{N}}_{r,j}^{s,E}+{\xi }_{r,j}^{s,E}\\ l_{r,j}^{s,J}={\mathbf{u}}_r^{s,J}\cdot \mathbf{x}+cd{\bar{t}}_r^G+c{\mathrm{I}\mathrm{S}\mathrm{B}}_r^J+{\mathrm{M}\mathrm{w}}_r^{s,J}\cdot {\mathrm{Z}\mathrm{W}\mathrm{D}}_r-{\gamma }_j^J\cdot {\bar{I}}_{r,1}^{s,J}+{\lambda }_j^{s,J}\cdot {\bar{N}}_{r,j}^{s,J}+{\xi }_{r,j}^{s,J}\end{array}\end{array}$$
其中，
$${\mathrm{ISB}}_r^Q=\left(d_{r,{\mathrm{I}\mathrm{F}}_{12}}^Q-d_{r,{\mathrm{I}\mathrm{F}}_2}^G\right)+\left(d{D}^Q-d{D}^G\right)(Q\ne G)$$
式中, ${\mathrm{I}\mathrm{S}\mathrm{B}}_r^Q$ 表示 $Q$ 系统的 ISB 参数。可以看出，ISB 不仅源于不同 GNSS 系统
对应的接收机硬件延迟差异 $(d_{r,1{\mathrm{F}}_2}^Q-d_{r,{\mathrm{I}\mathrm{F}}_2}^G,$ 与接收机有关 $),$ 而且还源于不同
GNSS 钟差产品相应的不同钟差基准约束引入的时间差异 $(d{D}^Q-d{D}^G$ 与接收机
无关 $）$ 。该模型的待估参数向量 X 为:
$$\mathbf{X}={\left[\mathbf{x},cd{\bar{t}}_r^G,c{\mathrm{ISB}}_r^Q,{\mathbf{Z}\mathbf{W}\mathbf{D}}_r,{\overline{\mathbf{I}}}_{r,1}^G,{\overline{\mathbf{I}}}_{r,1}^Q,{\overline{\mathbf{N}}}_{r,j}^G,{\overline{\mathbf{N}}}_{r,j}^Q\right]}^T$$
模型化 ISB 参数的三个动态模型为即时间常数、随机游走过程和白噪声过程。

周锋. 多系统GNSS非差非组合精密单点定位相关理论和方法研究 [D]; 华东师范大学, 2018.

1.2.1常数估计

ISB 参数可作为时间常数来估计:
$${\mathrm{ISB}}_{r,0}^s(k)={\mathrm{ISB}}_r^s(k-1)$$
其中，
$${\sigma }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{,.0}^Q(k)}^2={\sigma }_{\mathrm{I}\mathrm{S}\mathrm{B}}^2$$
式中， $k$ 表示历元号 $;{\mathrm{ISB}}_{r,0}^s(k)$ 表示第 $k$ 历元 ISB 的初值; ${\mathrm{ISB}}_r^s(k-1)$ 是第 $k-1$
历元 ISB 估计值 $;{\sigma }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{,0}^2(k)}^2$ 是第 $k$ 历元 ISB 参数的先验方差，而 ${\sigma }_{{\mathrm{I}\mathrm{S}\mathrm{B}}^s(k-1)}^2$ 是第 $k-1$
历元 ISB 参数更新后的方差

1.2.2随机游走过程

ISB 参数可看作随机游走过程来估计:
$${\mathrm{ISB}}_{r,0}^s(k)={\mathrm{ISB}}_r^Q(k-1)+{\omega }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,0}^e(k)},{\omega }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,0}^s(k)}\sim N\left(0,{\sigma }_{{\omega }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,0}^2(k)}^2}\right)$$
其中，
$${\sigma }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,0}^s(k)}^2={\sigma }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_r^s(k-1)}^2+{\sigma }_{{\omega }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,0}^s(k)}^2}$$
式中，ISB 参数的变化部分 $\left({\omega }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,0}^e(k)}\right)$ 对应的方差为 ${\omega }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,0}^s(k)}^2$

1.3.3.白噪声过程

ISB 参数可看作白噪声过程来估计:
$${\mathrm{ISB}}_{r,0}^Q(k)\sim N\left({\mathrm{ISB}}_{r,\mathrm{s}\mathrm{p}\mathrm{p}}^Q(k),{\sigma }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,\mathrm{p}\mathrm{r}\mathrm{i}}^e(k)}^2\right)$$
式中， ${\mathrm{ISB}}_{r,\text{ spp }}^Q(k)$ 表示伪距单点定位 $(\mathrm{S}\mathrm{P}\mathrm{P})$ 计算的 $\mathrm{I}\mathrm{S}\mathrm{B}$ 值 $;{\sigma }_{{\mathrm{I}\mathrm{S}\mathrm{B}}_{r,\mathrm{p}\mathrm{r}\mathrm{i}}^s(k)}^2$ 是 ISB 参数
的先验方差。

2.ISPB

It has been known that it is the ISPB originating in stations that can inhibit inter-GNSS double-difference AR (e.g. Odijk and Teunissen 2013 ). Regarding PPP, the hurdle is that station UPDs can no longer be removed through single differencing between satellites belonging to different GNSS. In contrast to Eq. 5, we have inter-GNSS single-difference ambiguities
$$\begin{array}{l}{\hat{N}}_{i,\mathrm{m}}^{kq}=\frac{L_{i,\mathrm{m}}^k}{{\lambda }_{\mathrm{G},\mathrm{m}}}-\frac{L_{i,\mathrm{m}}^q}{{\lambda }_{\mathrm{C},\mathrm{m}}}={\breve{N}}_{i,\mathrm{m}}^{kq}+{\bar{b}}_{i,\mathrm{m}}^{\mathrm{G}\mathrm{C}}-{\bar{b}}_{\mathrm{m}}^{kq}\\ {\lambda }_{\mathrm{G},\mathrm{n}}{\hat{N}}_{i,1}^k-{\lambda }_{\mathrm{C},\mathrm{n}}{\hat{N}}_{i,1}^q={\lambda }_{\mathrm{G},1}{\hat{N}}_{i,3}^k-{\lambda }_{\mathrm{C},1}{\hat{N}}_{i,3}^q\\ -\frac{\lambda \mathrm{G},\mathrm{m}}{g_{\mathrm{G}}+1}{\tilde{N}}_{i,\mathrm{m}}^k+\frac{\lambda \mathrm{C},\mathrm{m}}{g\mathrm{C}+1}{\tilde{N}}_{i,\mathrm{m}}^q\end{array}$$
where satellite $k$ is from GPS and $q$ is from BeiDou. Equation $8\mathrm{b}$ can be further developed into
$$\begin{array}{r}{\lambda }_{\mathrm{G},\mathrm{n}}{\hat{N}}_{i,1}^{kq}+\left({\lambda }_{\mathrm{G},\mathrm{n}}-{\lambda }_{\mathrm{C},\mathrm{n}}\right){\hat{N}}_{i,1}^q={\lambda }_{\mathrm{G},1}{\hat{N}}_{i,3}^k-{\lambda }_{\mathrm{C},1}{\hat{N}}_{i,3}^q\\ -\frac{\lambda \mathrm{G},\mathrm{m}}{g_{\mathrm{G}}+1}{\tilde{N}}_{i,\mathrm{m}}^{kq}-\left(\frac{\lambda \mathrm{G},\mathrm{m}}{g_{\mathrm{G}}+1}-\frac{\lambda \mathrm{C},\mathrm{m}}{g_{\mathrm{C}}+1}\right){\tilde{N}}_{i,\mathrm{m}}^q\end{array}$$
and afterward
$$\begin{array}{rl}{\lambda }_{\mathrm{G},\mathrm{n}}{\hat{N}}_{i,1}^{kq}=& {\lambda }_{\mathrm{G},1}{\hat{N}}_{i,3}^k-{\lambda }_{\mathrm{C},1}{\hat{N}}_{i,3}^q-\left({\lambda }_{\mathrm{G},\mathrm{n}}-\lambda \mathrm{C},\mathrm{n}\right){\hat{N}}_{i,1}^q\\ & -\frac{\lambda \mathrm{G},\mathrm{m}}{g_{\mathrm{G}}+1}{\tilde{N}}_{i,\mathrm{m}}^{kq}-\left(\frac{\lambda \mathrm{G},\mathrm{m}}{g_{\mathrm{G}}+1}-\frac{\lambda \mathrm{C},\mathrm{m}}{g\mathrm{C}+1}\right){\tilde{N}}_{i,\mathrm{m}}^q\\ =& {\lambda }_{\mathrm{G},\mathrm{n}}\left({\check{N}}_{i,1}^{kq}+{\bar{b}}_{i,\mathrm{n}}^{\mathrm{G}\mathrm{C}}-{\bar{b}}_{\mathrm{n}}^{kq}\right)\end{array}$$
We note that
$$\{\begin{array}{l}{\bar{b}}_{i,\mathrm{m}}^{\mathrm{G}\mathrm{C}}={\bar{b}}_{i,\mathrm{m}}^{\mathrm{G}}-{\bar{b}}_{i,\mathrm{m}}^{\mathrm{C}}\ne 0\\ {\bar{b}}_{i,\mathrm{n}}^{\mathrm{G}\mathrm{C}}={\bar{b}}_{i,\mathrm{n}}^{\mathrm{G}}-{\bar{b}}_{i,\mathrm{n}}^{\mathrm{C}}\ne 0\end{array}$$
which are the wide-lane and narrow-lane ISPB for station $i$ between GPS and BeiDou, respectively. Equation 9 is used to compute inter-GPS/BeiDou single-difference narrow-lane ambiguities $\left({\hat{N}}_{i,1}^{kq}\right).$ On the right-hand side of this equation, we then need to know ${\hat{N}}_{i,3}^k$ and ${\hat{N}}_{i,3}^q$ which are directly PPP output, and also ${\tilde{N}}_{i,\mathrm{m}}^{kq}$ which is provided by resolving Eq. $8\mathrm{a}$. However, ${\tilde{N}}_{i,\mathrm{m}}^q$ cannot be derived since we are unable to resolve undifferenced wide-lane ambiguities with Eq. 8 a; ${\hat{N}}_{i,1}^q$ cannot be obtained either due to its dependence on ${\tilde{N}}_{i,\mathrm{m}}^q$ (see Eq. 3b).

Despite this difficulty, we note that the absolute values of the coefficients for ${\tilde{N}}_{i,\mathrm{m}}^q$ and ${\hat{N}}_{i,1}^q$ in Eq. 9 satisfy that
$$\{\begin{array}{l}\left|\frac{{\lambda }_{\mathrm{G},\mathrm{m}}}{g_{\mathrm{G}}+1}-\frac{{\lambda }_{\mathrm{C},\mathrm{m}}}{g_{\mathrm{C}}+1}\right|<0.82\mathrm{c}\mathrm{m}\\ {\lambda }_{\mathrm{G},\mathrm{n}}-{\lambda }_{\mathrm{C},\mathrm{n}}\mid <0.14\mathrm{c}\mathrm{m}\end{array}$$
Both are far smaller than ${\lambda }_{\mathrm{G},\mathrm{n}}=10.695\mathrm{c}\mathrm{m},$ hence implying that minor errors within ${\tilde{N}}_{i,\mathrm{m}}^q$ and ${\hat{N}}_{i,1}^q,$ if present, will have limited impact on ${\hat{N}}_{i,1}^{kq}.$ We therefore introduce the following approximations
$$\{\begin{array}{rl}{\tilde{N}}_{i,\mathrm{m}}^q& \approx {\hat{N}}_{i,\mathrm{m}}^q=\frac{L_{i,\mathrm{m}}^q}{{\lambda }_{\mathrm{C},\mathrm{m}}}\\ {\hat{N}}_{i,1}^q& \approx \frac{g\mathrm{C}+1}{g_{\mathrm{C}}}{\hat{N}}_{i,3}^q-\frac{1}{g_{\mathrm{C}}-1}{\hat{N}}_{i,\mathrm{m}}^q\end{array}$$
which refer to Eq. 3 . Contrasting Equations $3\mathrm{a}$ and $4,$ we find that the approximation errors of Eq. 12 consist in the
difference between $\left({\bar{b}}_{i,\mathrm{m}}^{\mathrm{C}}-{\bar{b}}_{\mathrm{m}}^q\right)$ and $\left({\hat{\bar{b}}}_{i,\mathrm{m}}^{\mathrm{C}}-{\hat{\bar{b}}}_{\mathrm{m}}^q\right)$ where the
former contributes to the estimation of the latter. The impact of this approximation on ${\hat{N}}_{i,1}^{kq}$ will be inspected in Sect. 2.4 . With Eqs. $8\mathrm{a}$ and 9 along with the substitution of Eq. 12 , we can calculate ${\hat{N}}_{i,\mathrm{m}}^{kq}$ and ${\hat{N}}_{i,1}^{kq},$ similar to Eq. $5,$ aiming at estimating station ISPBs.

We first presume that ISPBs ${\bar{b}}_{i,\mathrm{m}}^{\mathrm{G}\mathrm{C}}$ and ${\bar{b}}_{i,\mathrm{n}}^{\mathrm{G}\mathrm{C}}$ should remain the same for all pairs of GPS/BeiDou satellites observed at station $i$. From Eqs. 8 a and $9,$ we can form double-difference ambiguities between stations $i$ and $j$ such that satellite FCBs ${\bar{b}}_{\mathrm{m}}^{kq}$ and ${\bar{b}}_{\mathrm{n}}^{\mathrm{k}q}$ cancel completely, leading to
$$\{\begin{array}{l}{\hat{N}}_{ij,\mathrm{m}}^{kq}={\check{N}}_{ij,\mathrm{m}}^{kq}+{\bar{b}}_{ij,\mathrm{m}}^{\mathrm{G}\mathrm{C}}\\ {\hat{N}}_{ij,1}^{kq}={\check{N}}_{ij,1}^{kq}+{\bar{b}}_{ij,\mathrm{n}}^{\mathrm{G}\mathrm{C}}\end{array}$$
where
$$\{\begin{array}{l}{\tilde{N}}_{ij,\ast }^{kq}={\tilde{N}}_{i,\ast }^{kq}-{\tilde{N}}_{j,\ast }^{kq}\\ {\hat{N}}_{ij,\ast }^{kq}={\hat{N}}_{i,\ast }^{kq}-{\hat{N}}_{j,\ast }^{kq}\\ {\bar{b}}_{ij,\ast }^{GC}={\bar{b}}_{i,\ast }^{\mathrm{G}\mathrm{C}}-{\bar{b}}_{j,\ast }^{\mathrm{G}\mathrm{C}}\end{array}$$
Note that for intra-system double-difference ambiguities, the ISPB terms ${\bar{b}}_{ij,\mathrm{m}}^{\mathrm{G}\mathrm{C}}$ and ${\bar{b}}_{ij,\mathrm{n}}^{\mathrm{G}\mathrm{C}}$ will be absent, and ${\hat{N}}_{ij,\mathrm{m}}^{kq}$ and ${\hat{N}}_{ij,1}^{kq}$
have the integer property naturally. In case of inter-system AR, however, we have to correct for ${\bar{b}}_{ij,\mathrm{m}}^{\mathrm{G}\mathrm{C}}$ and ${\bar{b}}_{ij,\mathrm{n}}^{\mathrm{G}\mathrm{C}}$ before being able to resolve ${\hat{N}}_{ij,\mathrm{m}}^{kq}$ and ${\hat{N}}_{ij,1}^{kq}$

Similar to Eq. $7,$ for all stations collecting GPS and BeiDou data simultaneously, we can calculate differential ISPBs

between stations $i$ and $j$ using
$$\{\begin{array}{l}{\hat{\bar{b}}}_{ij,\mathrm{m}}^{\mathrm{G}\mathrm{C}}={\hat{N}}_{ij,\mathrm{m}}^{kq}-\left[{\hat{N}}_{ij,\mathrm{m}}^{kq}\right]\\ {\hat{b}}_{ij,\mathrm{n}}^{\mathrm{G}\mathrm{C}}={\hat{N}}_{ij,1}^{kq}-\left[{\hat{N}}_{ij,1}^{kq}\right]\end{array}$$
Here, we have presumed that ${\bar{b}}_{ij,\mathrm{m}}^{\mathrm{G}\mathrm{C}}$ and ${\bar{b}}_{ij,\mathrm{n}}^{\mathrm{G}\mathrm{C}}$ are both merely the fractional parts of ISPB and their integer parts have been assimilated into ${\breve{N}}_{ij,\mathrm{m}}^{kq}$ and ${\breve{N}}_{ij,1}^{kq},$ respectively. For any given baseline, we will have a good number of interGPS/BeiDou double-difference ambiguities, and differential ISPB between the two relevant stations can thus be computed by averaging all estimates of ${\bar{b}}_{ij,\mathrm{m}}^{\mathrm{G}\mathrm{C}}$ and ${\bar{b}}_{ij,\mathrm{n}}^{\mathrm{G}\mathrm{C}}$. The resultant ISPBs can afterwards be converted into undifferenced values (e.g., ${\bar{b}}_{i,\mathrm{m}}^{\mathrm{G}\mathrm{C}}$ and ${\bar{b}}_{i,\mathrm{n}}^{\mathrm{G}\mathrm{C}}$ ) by choosing a reference station ISPB.