Kalman filter ionospheric delay estimator

• US 5,867,411 A
• Filed: 12/19/1996
• Issued: 02/02/1999
• Est. Priority Date: 12/19/1996
• Status: Expired due to Term

• Current Assignee: Aerospace Corporation
• Original Assignee: Aerospace Corporation
• Inventors: Kumar, Rajendra
• Primary Examiner(s): Sheikh, Ayaz R.
• Assistant Examiner(s): Mai, Rijue
• Application Number:
  US08/770,079
• Time in Patent Office: 775 Days
• Field of Search: 375/232, 375/346, 367/135, 342/63, 364/717.04, 364/724.01, 364/724.014, 364/724.02, 364/737, 364/724.19, 364/745.04, 364/449, 702/104, 702/196
• US Class Current: 708/300
• CPC Class Codes: G01S 19/04 providing carrier phase data
• CPC Class Codes: G01S 19/071 DGPS corrections
• CPC Class Codes: G01S 19/072 Ionosphere corrections
• CPC Class Codes: H03H 17/0257 KALMAN filters
• CPC Class Codes: H03J 7/02 Automatic frequency control H03J7/18 takes precedence

Abstract

An estimator structure include a nonlinear modified Kalman filter for initial estimate and plurality of bias estimators and a parallel bank of Kalman filters for ambiguity resolution and providing final estimates. The estimator structure is useful for real-time computation of ionospheric propagation delay introduced in the navigation signals by dual frequency GPS receivers.

Claims

1. A method of generating initial parameter estimates of parameters estimated from an input having a bias and being a series of inputs at a respective plurality of discrete times, the method comprising the steps of,

bias estimating bias estimates from the inputs and from the initial parameter estimates, and

initial parameter estimating the initial parameter estimates from the input and the bias estimates, the initial parameter estimates have an estimation error dependent upon the amount of the bias, the bias estimating step and the initial parameter estimating step recursively estimate in synchronism at each of the plurality of discrete times to suppress the bias from the input to reduce the estimation error, the initial parameter estimating step comprising the steps of:

bias suppressing the input by removing the bias estimates from the inputs for providing a suppressed bias input;

providing a prediction error from the suppressed bias input and from an input prediction, the prediction error indicating suppressed bias;

providing a modified prediction error by modifying the prediction error with nonlinear bias rejection;

providing the initial parameter estimates from the modified prediction error; and

providing the input prediction error from the modified prediction error and the initial parameter estimates.

2. The method of claim 1 wherein the parameters comprise ambiguity parameters and the initial parameter estimates comprise ambiguity parameter estimates, the method further comprising the steps of

input delaying the inputs to provide delayed inputs delayed by one of the plurality of discrete times, and

resolving in synchronism at the plurality of discrete times the ambiguity parameters to provide final parameter estimates from the delayed inputs, the bias estimates and the initial parameter estimates.

3. The method of claim 2 wherein the ambiguity resolving step is a filtering step using a parallel bank of Kalman filters.

4. The method of claim 1 wherein all of the providing steps is a filtering step using a modified Kalman filter modified with nonlinear bias rejection.

5. A method of generating final parameter estimates of parameters estimated from an input having a bias and being a series of signals at a respective plurality of discrete times, the parameters comprise integer ambiguity parameters and the final parameter estimates comprise final ambiguity parameter estimates, the method comprising the steps of,

bias estimating bias estimates from the inputs and from initial parameter estimates,

bias suppressing the input by removing the bias estimates from the inputs for providing a suppressed bias input,

providing a prediction error from the suppressed bias input and from an input prediction,

providing a modified prediction error by nonlinear bias rejection of the prediction error for suppressing the bias,

providing the initial parameter estimates from the modified prediction error,

providing the input prediction error from the modified prediction error, the initial parameter estimates have an estimation error dependant upon the amount of the bias, the bias estimating step and all of the providing steps function to recursively estimate in synchronism at each of the plurality of discrete times to suppress the bias from the input to reduce the estimation error,

input delaying the inputs to provide delayed inputs delayed by one of the plurality of discrete times, and

resolving in synchronism at the plurality of discrete times the integer ambiguity parameters to provide the final parameter estimates from the delayed inputs, the bias estimates and the initial parameter estimates.

6. The method of claim 5 wherein

the resolving step is a filtering step using a parallel bank of Kalman filters, and

all of the providing steps is a filtering step using a modified Kalman filter with nonlinear bias rejection.

7. The method of claim 5 wherein

the input is a plurality of inputs each of which is a respective series of signals at the respective discrete times,

the bias estimating step comprises a plurality of bias estimating steps respectively for said plurality of inputs.

8. The method of claim 5 wherein the series of inputs are computed from modulating signals having a carrier phase uncertainty creating an ambiguity parameter in the series of inputs.

9. A method of generating Iρ

N₂ N₁ final parameter estimates of parameters estimated from an input comprising four PR₁ PR₂ ADR₁ and ADR₂ pseudorange z (k+1) inputs each having a bias and each being a series of signals at a respective plurality of discrete times (k), thee parameters comprise range ρ and

ionospheric delay I parameters and N₁ and N₂ ambiguity parameters and the final parameter estimates comprise final range and ionospheric delay parameters estimates and N₁ and N₂ ambiguity parameter estimates, the method comprising the steps of,

bias estimating b₁ (k) b₂ (k) b₃ (k) and b₄ (k) bias estimates in parallel from the respective four inputs PR₁ PR₂ ADR₁ and ADR₂ and from initial parameter estimates,

bias suppressing the inputs by respectively removing the bias estimates from the inputs for providing respectively suppressed bias inputs, z_mod (k+1),

providing respective prediction error ε

(k+1) the from suppressed bias inputs and from an input prediction,

providing a modified prediction error ε

_mod (k+1) f0r suppressing the bias,

providing the initial parameter estimates z (k/k) from the modified prediction error,

providing the input prediction error from the modified prediction error, the initial parameter estimates have an estimation error dependant upon the amount of the bias, the bias estimating step and all of the generating and providing steps function to recursively estimate in synchronism at each of the plurality of discrete times to suppress the bias from the input to reduce the estimation error,

input delaying the four PR₁, PR₂, ADR₁ and ADR₂ pseudorange inputs to provide delayed inputs z (k) delayed by one of the plurality of discrete times, and

resolving in synchronism at the plurality of discrete times the N₁ and N₂ ambiguity parameters to provide the final parameter estimates from the delayed inputs, the bias estimates and the initial parameter estimates.

10. The method of claim 9 wherein the initial parameter estimates x(k/k) comrpises

x₂ (k/k) initial ionospheric delay estimates for generating the bias estimates, and

N₁ and N₂ initial ambiguity estimates for generating the final parameter estimates.

11. The method of claim 9 wherein all of the providing steps function as a modified Kalman filter modified by four inlier/outlier rejection nonlinearity functions providing the modified prediction error.

12. The method of claim 9 wherein the resolving step functions as a bank of parallel Kalman filters.

13. The method of claim 9 wherein the four inputs have respective biases b1, b2, b3 and b4, one of the b1, b2, b3 and b4 biases is not zero and one of b1, b2, b3 and b4 biases is zero.

14. The method of claim 9 wherein the bias estimating step functions as four parallel bias estimators, one of which provides a zero bias estimate.

15. The method of claim 14 wherein the bias estimators are each a first order exponential data weighted filter.

16. The method of claim 10 wherein the N₁ and N₂ ambiguity parameters are derived from carrier phase uncertainties of L₁ and L₂ band carrier signals modulated by modulating signals from which PR₁, PR₂, ADR₁, and ADR₂ pseudorange inputs are computed.