Efficient Spectrum Sensing with RNN-GRU in Cognitive Radio Networks

Resumen

Modern wireless communication systems face increasing challenges in efficiently managing radio frequency spectrum in highly dynamic and congested environments. Cognitive radios play a pivotal role by utilizing advanced spectrum sensing techniques to identify available frequency bands and avoid interference. This study introduces a novel model that integrates Recurrent Neural Networks (RNN) and Gated Recurrent Units (GRU), addressing the limitations of traditional spectrum sensing methods. RNNs excel at capturing temporal patterns in signal data, while GRUs enhance learning efficiency and adaptability to rapidly changing signal characteristics. Unlike previous approaches, this hybrid model demonstrates superior performance in complex and noisy environments. Evaluated using the RadioML 2016.10a dataset and key metrics such as F1 score, MCC, and CKC, the proposed technique outperforms both traditional and recent methods in accuracy and efficiency. These findings highlight the potential of this innovative approach to significantly enhance spectrum utilization and reliability in wireless sensor networks (WSNs).