DoA Estimation of GNSS Jamming Signal using Tripole Vector Antenna and Random Forest Regression

This paper presents an effective strategy for the Global Navigation Satellite System receivers (GNSS) to estimate the direction of arrival (DoA) of jamming signals. A tripole vector antenna is used which consists of three orthogonal dipole antennas. This configuration enables DoA estimation for all directions. A Random Forest Regression (RFR) based model is designed for DoA estimation using the signal received from the tripole vector antenna. Using a baseband receiver processing, the training and testing datasets are prepared. The performance accuracy of the proposed methodology is validated and verified using performance metrics such as $R^{2}$ score and Mean Squared Error. The model hyperparameter number of estimators is tuned on the basis of the said performance metrics. The proposed signal model and regression model are able to estimate DoA of a jamming signal coming from a spinning in both azimuth and elevation. The accuracy of the proposed model in terms of RMSE ranges from 0° to 0.122° for varying jamming to signal ratio (J/S).