SCIENCE@home

Enhancing IRS Localization via Deep Learning-Based AOA and Distance Estimation

Nikhila Kotlapara Surendran,Syed Samiul Alam,Haolin Tang,T. Tahi,Yanxiao Zhao

Published 2025 in Journal of AI-Driven Communication Engineering

ABSTRACT

In Intelligent Reflecting Surface (IRS)-assisted communication systems, accurate user localization, particularly Angle-Of-Arrival (AoA) and range estimation are challenging due to the computational complexity and limited resolution of traditional Multiple Signal Classification (MUSIC) algorithms. This paper introduces a hybrid IRS framework that combines machine learning with a modified MUSIC algorithm to achieve high-precision localization and enhanced security. The system integrates two Convolutional Neural Networks (CNNs): RefineNet, which refines AoA and range estimates from MUSIC pseudo-spectra, and ElementNet, which optimizes the number and placement of active IRS elements to balance accuracy with resource efficiency. Notably, ElementNet shows that only eight active elements are sufficient to obtain 90% of the best achievable localization accuracy, highlighting the efficiency of the proposed design. Validation on the DeepMIMO dataset demonstrates superior range accuracy and AoA precision. This work sheds light on the secure and high-precision localization for diverse wireless applications.  

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-25 of 25 references · Page 1 of 1

CITED BY

  • No citing papers are available for this paper.

Showing 0-0 of 0 citing papers · Page 1 of 1