Deep Learning-Assisted Robust Joint Estimation of Array Amplitude–Phase Errors and Direction of Arrival Using Auxiliary Elements
The International Journal of Applied Sciences · 2026
DOI: 10.64440/ijas/ijas10011 ↗Kurz gefasst
Fachartikel in der Zeitschrift The International Journal of Applied Sciences, 2026. Verfasst von AHMAT WAHU ALI. Die Arbeit nennt 17 Literaturangaben.
Abstract (Originalsprache)
Array imperfections are unavoidable in practical sensing systems and can significantly deteriorate the accuracy of direction-of-arrival (DOA) estimation. Among the different forms of array perturbation, sensor gain and phase errors are particularly common because of manufacturing tolerances, receiver-chain mismatch, aging, temperature fluctuations, and imperfect calibration. To address the degradation of DOA estimation caused by gain–phase errors, this study develops a deep learning-assisted robust joint estimation framework for simultaneously estimating sensor gain–phase errors and the DOA of incident signals. A small number of accurately calibrated auxiliary array elements are introduced at one side of the array to impose a structured constraint on the gain–phase error matrix. Based on this structure, a transformation matrix is constructed and integrated with a subspace-based estimation framework to jointly recover the sensor errors and source directions. To further improve robustness under unequal source powers and low signal-to-noise ratio (SNR) conditions, a deep learning module is incorporated to learn the nonlinear relationship between distorted array observations, calibration parameters, and DOA information. The proposed framework therefore combines the interpretability of model-based subspace estimation with the adaptability of data-driven learning. In particular, the auxiliary elements improve parameter identifiability, while the learning-assisted refinement reduces the sensitivity of the estimator to inaccurate covariance estimation and large power disparities among incident sources. The proposed method is evaluated through numerical simulations under different SNR levels, source-power ratios, numbers of snapshots, gain–phase error levels, and numbers of auxiliary elements. The results demonstrate that the proposed strategy can achieve accurate simultaneous estimation of array errors and DOAs while maintaining improved robustness in nonideal sensing environments. The method provides a practical approach for array calibration and intelligent spatial sensing in radar, wireless communication, localization, and other applied sensing systems.
Zitieren
APA
ALI, A. W. (2026). Deep Learning-Assisted Robust Joint Estimation of Array Amplitude–Phase Errors and Direction of Arrival Using Auxiliary Elements. The International Journal of Applied Sciences. https://doi.org/10.64440/ijas/ijas10011
Vancouver
ALI AW. Deep Learning-Assisted Robust Joint Estimation of Array Amplitude–Phase Errors and Direction of Arrival Using Auxiliary Elements. The International Journal of Applied Sciences. 2026. doi:10.64440/ijas/ijas10011
BibTeX
@article{ALI2026,
author = {ALI, AHMAT WAHU},
title = {Deep Learning-Assisted Robust Joint Estimation of Array Amplitude–Phase Errors and Direction of Arrival Using Auxiliary Elements},
journal = {The International Journal of Applied Sciences},
year = {2026},
doi = {10.64440/ijas/ijas10011}
}
EUVIMED