Machine learning in GNSS receivers for improved velocity outputs
Machine learning techniques are used to compute predicted range rate errors in a GNSS receiver. In one embodiment, training data is computed to provide true range rate error data for a set of received GNSS signals. A system extracts features from the set of received GNSS signals and uses the extracted features and the true range rate error data to train a model (e.g., a set of one or more neural networks) that can produce predicted range rate errors for use in correcting measurements. The trained set of one or more neural networks can be deployed in GNSS receivers and used in the GNSS receivers to correct Doppler measurements using the predicted range rate errors provided by the trained set of one or more neural networks.
1. A global navigation satellite system (GNSS) receiver comprising:
an antenna to receive GNSS signals from GNSS satellites (SVs);
a radiofrequency (RF) front end coupled to the antenna to amplify the GNSS signals;
an analog to digital converter (ADC) coupled to the RF front end to generate a digital representation of received GNSS signals;
a baseband memory coupled to the ADC to store the digital representation;
a GNSS processing system coupled to the baseband memory to process the received GNSS signals, the GNSS processing system including a set of correlators that provide outputs that include correlation outputs;
wherein the GNSS processing system includes processing logic to extract a set of features from the received GNSS signals, the set of features being predetermined based on a trained model in the GNSS receiver and being related to range rate errors;
wherein the GNSS processing system includes processing logic to apply the set of features as an input to the trained model; and
wherein the GNSS processing system includes processing logic to generate, by the trained model, a set of one or more predicted range rate errors to correct one or more measurements made by the GNSS receiver.
2. The GNSS receiver as in claim 1 , wherein the GNSS processing system sets a pseudorange rate uncertainty based on the set of one or more predicted range rate errors.
3. The GNSS receiver as in claim 1 , wherein the set of one or more predicted range rate errors is used to adjust for errors in Doppler measurements made by the GNSS receiver.
4. The GNSS receiver as in claim 1 , wherein the set of features comprises one or more of: (1) a statistical computation of carrier phase measurements; (2) a statistical computation of pseudorange measurements; (3) a carrier to noise density ratio (C/NO); (4) an elevation of a GNSS satellite; (5) a signal to noise ratio of GNSS signals; (6) a statistical computation of frequency of received GNSS signals; (7) a statistical computation of Doppler measurements; or (8) a statistical computation of pseudorandom code phase measurements.
5. The GNSS receiver as in claim 4 , wherein the statistical computation is either: (a) a standard deviation; or (b) a mean.
6. The GNSS receiver as in claim 4 , wherein the trained model comprises one or more deep neural networks that comprises at least two hidden layers.
7. The GNSS receiver as in claim 6 , wherein the processing logic prunes data in the set of features, balances data in the set of features, and scales data in the set of features.