An Efficient Approach for Machine Learning-based DOA Estimation of Targets in Passive RADAR

DOI: https://doi.org/10.54985/peeref.2401w7871726

---

This research addresses challenges in machine learning-based direction of arrival (DOA) estimation. The study proposes a novel single-channel approach to overcome limitations observed in the existing two-channel approach.The proposed approach treats DOA estimation as a function approximation problem, ensuring the uniqueness of the mapping between feature and target variables. To validate this approach, radial basis function neural network and support vector regression are used as benchmark models. Simulation results show that the proposed approach outperforms the existing one in reliability and accuracy, even with fewer array elements, snapshots, and in low signal-to-noise ratio conditions.The proposed approach also enhances efficiency by reducing computational complexity in training and testing, leading to shorter execution times. This solution not only improves DOA estimation but also offers a streamlined alternative for machine learning applications in the passive RADAR system.