EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Hybrid Model for Data-Driven Channel Estimation in mmWave Massive MIMO Systems with Compressed Sensing and Deep Learning

Yang Haoxuan, Faizan Qamar, Syed Hussain Ali Kazmi, Khairul Akram Zainol Ariffin, Kashif Nisar

Mehran University Research Journal of Engineering and Technology · 01.04.2026 · 45 (2) · 204-216

DOI: 10.22581/0712 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Mehran University Research Journal of Engineering and Technology, online erschienen am 01.04.2026. Verfasst von Yang Haoxuan und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Millimeter-wave (mmWave) massive Multiple-Input Multiple-Output (MIMO) technology promises to provide gigabit-level transmission rates for 5G/6G networks. However, this technology suffers from high channel dimensionality and sparse signals, requiring many pilot signals for estimation using traditional methods, which are also time-consuming. This paper proposes a two-stage hybrid Compressed Sensing-Deep Learning (CS-DL) framework that combines the model advantages of the Orthogonal Matching Pursuit (OMP) algorithm with the data processing capabilities of a lightweight Convolutional Neural Network (CNN). The first stage exploits the signal's sparsity in the angular domain, requiring only 5% of the pilot signal to obtain preliminary channel estimates. In the second stage, signal gain is optimized using a residual CNN, which reduces error interference. MATLAB simulation results on the DeepMIMO dataset indicate that, compared to the traditional OMP algorithm, the proposed framework reduces the Normalized Mean Square Error (NMSE) by up to 1.8 dB, increases Spectral Efficiency (SE) by approximately 9%, and decreases the total estimation time to about 0.7 milliseconds, thereby satisfying the latency requirements of 5G systems. Additionally, the framework offers remarkable flexibility, enabling integration with low-resolution hardware and hybrid beamforming architectures, making it suitable for real-time deployment in applications such as automotive communication systems and mmWave access points.

Zitieren

APA

Haoxuan, Y., Qamar, F., Kazmi, S. H. A., Ariffin, K. A. Z., & Nisar, K. (2026). Hybrid Model for Data-Driven Channel Estimation in mmWave Massive MIMO Systems with Compressed Sensing and Deep Learning. Mehran University Research Journal of Engineering and Technology, 45(2), 204-216. https://doi.org/10.22581/0712

Vancouver

Haoxuan Y, Qamar F, Kazmi SHA, Ariffin KAZ, Nisar K. Hybrid Model for Data-Driven Channel Estimation in mmWave Massive MIMO Systems with Compressed Sensing and Deep Learning. Mehran University Research Journal of Engineering and Technology. 2026;45(2):204-216. doi:10.22581/0712

BibTeX
@article{Haoxuan2026,
  author = {Haoxuan, Yang and Qamar, Faizan and Kazmi, Syed Hussain Ali and Ariffin, Khairul Akram Zainol and Nisar, Kashif},
  title = {Hybrid Model for Data-Driven Channel Estimation in mmWave Massive MIMO Systems with Compressed Sensing and Deep Learning},
  journal = {Mehran University Research Journal of Engineering and Technology},
  year = {2026},
  volume = {45},
  number = {2},
  pages = {204--216},
  doi = {10.22581/0712}
}
Kostenlos starten →