Attention and Augmented Deep Temporal Convolutional Networks (AA-DTCN) for Enhanced Hybrid Spectrum Sensing in Cognitive Radio Networks
Year: 2026
Type: conference-paper
Abstract: Spectrum sensing in Cognitive Radio Networks (CRNs) is challenging due to channel conditions, low SNR, noise uncertainty, diverse features, and dynamic spectrum allocation. However, existing models such as RNN/CNN and LSTM have poor performance in modeling, feature fusion, and accuracy, as well as limitations in fusion and precise attention. To overcome these shortcomings, the attention-Augmented Deep Temporal Convolution Network (AA-DTCN) uses temporal attention and convolution, along with additive fusion, and works under dynamic media and low SNR. The proposed AA-DTCN outperforms all CNN Models by 90.89 and opens the path for next-generation CRNs and other temporal-based sensing applications.
Authors: R. Koteswara Rao, Madona B. Sahaai
Institution: Vels University
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W7224967880