Hybrid AI-Based Cognitive Radio for Spectrum Sensing and Robust Multimedia Transmission

The efficiency in the use of the frequency spectrum and the ability to provide reliable data transmission represent two important challenges of current wireless communications, especially in cognitive radio based on dynamic or unpredictable channels. The present work develops an integrated architecture for hybrid cognitive radio, where energy detection, fuzzy logic and deep learning can be used for improving the performance of the spectrum sensing and ensuring a good level of reliability in the transmission of multimedia content. Thus, it will be possible to combine a classical threshold-based energy detection with fuzzy reasoning to manage uncertainty, whereas a CNN could be used for detecting complex signal patterns and improving classification accuracy. Additionally, neural error-correcting capabilities and CNN-based denoising functionalities have been added to improve the reliability of data transmission and the quality of reconstructed content. The architecture developed in this work has been designed for the transmission of multimedia content such as text, images, audio and video through a noisy channel represented by AWGN and QPSK modulation. The experimental results obtained show that the hybrid sensing method allows improving the consistency of detection and reducing the number of misclassifications compared to traditional methods. Moreover, an inverse relation between BER and PSNR was found. Therefore, the proposed pipeline for transmitting multimedia content ensures high-quality image/video reconstructions even when operating in different channel conditions. Finally, despite some limitations due to model bias and simple assumptions about channel behavior, the proposed framework presents several advantages such as scalability and adaptability to solve problems related to intelligent spectrum management.