Deep Learning Approach for Angle Estimation Based on Antenna Array Features in Dense Planar Arrays

Read full text Summarise with VIRAT
Authors: Sarmad Muneer Abdulhussein, Asaad. S. Daghal
Cited by: 0
References: 43

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper proposes a learning-based 2D direction of arrival (DoA) estimation framework for 28 GHz holographic planar arrays.It employs a synthetic multi-user orthogonal frequency division multiplexing scenario for 8x8 and 16x16 dense arrays with spatial covariance matrix regression features.A deep neural network is used to predict azimuth and elevation, which are then used to generate holographic beams.For the 8x8 array, average root mean squared error (RMSE) is 0.00311, and average mean absolute error is 0.00252.The proposed approach outperforms linear regression, decision trees, random forest, support vector regression, and a fully connected dense neural network.For the 16x16 array, average RMSE is 0.044 (azimuth) and 0.013 (elevation).Beam validation is performed on 300 tests, showing that 95% of pointing errors are within 4°.Peak loss is centered at 0 dB with most values within ±0.3 dB.

Fields of Study & Subjects

Direction-of-Arrival Estimation Techniques Antenna Design and Analysis Millimeter-Wave Propagation and Modeling

Publication details

Journal / source
International journal of intelligent engineering and systems
Volume
19
Issue
4
Pages
970–987
Publication date
23 March 2026
Language
EN