EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Robust and Lightweight Deep Learning Framework for Cooperative Spectrum Sensing in Cognitive Radio Ad Hoc Networks Under Dynamic Channel Conditions

Majid Shaikh, Dr. Priti Mishra

Computer and Decision Making: An International Journal · 03.11.2026 · 3 (6S) · 1305-1328

DOI: 10.1922/comdem.v3i6s.18912 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 03.11.2026. Verfasst von Majid Shaikh und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Cognitive Radio Ad Hoc Networks (CRAHNs) provide a flexible approach to improving spectrum utilization by enabling secondary users (SUs) to access licensed frequency bands whenever primary users (PUs) are inactive. The effectiveness of this approach depends on accurate spectrum occupancy detection, as missed detections can cause harmful interference to licensed communications, whereas false alarms unnecessarily restrict spectrum access for secondary users. Although deep learning has demonstrated potential for processing complex radio-frequency signals, modulation recognition and binary spectrum occupancy detection represent distinct tasks, and results obtained under limited simulation conditions may not adequately reflect real-world network variability. To address these challenges, this study proposes an evaluation framework for cooperative spectrum sensing based on a Convolutional Neural Network–Bidirectional Gated Recurrent Unit with an attention mechanism (CNN–BiGRU–Attention). The convolutional component extracts discriminative local features from in-phase and quadrature (I/Q) signal samples, while the bidirectional recurrent layer captures temporal dependencies and the attention mechanism emphasizes informative representations relevant to primary-user detection. To improve cooperative decision-making, the framework incorporates both majority-voting and confidence-weighted fusion strategies for combining sensing information from multiple secondary users. The proposed approach will be evaluated against conventional energy detection and representative deep learning architectures, including 1D-CNN, LSTM, GRU, Temporal Convolutional Network (TCN), and a lightweight Transformer. The experimental design considers variations in signal-to-noise ratio (SNR), noise uncertainty, fading environments, the number of cooperating secondary users, reporting-channel impairments, and previously unseen channel conditions. Evaluation will focus on probability of detection at a specified false-alarm rate, missed-detection probability, receiver operating characteristic area under the curve (ROC-AUC), computational latency, model size, and cooperative reporting overhead. A controlled binary occupancy dataset with explicitly defined signal-presence labels and scenario-level data separation will be used to support reproducible experimentation and reduce data leakage between training and evaluation sets. The study is intended to assess not only detection performance but also the trade-offs between sensing reliability, computational cost, and cooperative communication requirements. Experimental results and comparative conclusions will be reported only after implementation and systematic testing; therefore, this abstract makes no claims of measured performance improvement or established superiority over existing methods.

Zitieren

APA

Shaikh, M., & Mishra, D. P. (2026). A Robust and Lightweight Deep Learning Framework for Cooperative Spectrum Sensing in Cognitive Radio Ad Hoc Networks Under Dynamic Channel Conditions. Computer and Decision Making: An International Journal, 3(6S), 1305-1328. https://doi.org/10.1922/comdem.v3i6s.18912

Vancouver

Shaikh M, Mishra DP. A Robust and Lightweight Deep Learning Framework for Cooperative Spectrum Sensing in Cognitive Radio Ad Hoc Networks Under Dynamic Channel Conditions. Computer and Decision Making: An International Journal. 2026;3(6S):1305-1328. doi:10.1922/comdem.v3i6s.18912

BibTeX
@article{Shaikh2026,
  author = {Shaikh, Majid and Mishra, Dr. Priti},
  title = {A Robust and Lightweight Deep Learning Framework for Cooperative Spectrum Sensing in Cognitive Radio Ad Hoc Networks Under Dynamic Channel Conditions},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {1305--1328},
  doi = {10.1922/comdem.v3i6s.18912}
}
Kostenlos starten →