Enhanced Cross-Layer Communication Through Modified XGBoost-DNN Algorithm for Cognitive Radio Networks: Performance Analysis and Evaluation

Cognitive Radio Networks (CRNs) are very promising communication systems that urgently need effective spectrum-sharing techniques. The overall functional integration across layers is effective in managing CRNs, enhancing performance, hence stability, under changing conditions. This study presents a modified XGBoost-DNN algorithm that enhances cross-layer information exchange in CRNs by selecting features using both XGBoost and DNNs for prediction. A simulation test with 1,000 CR nodes that lasted 100 seconds showed that the suggested algorithm is 15% better than current methods in terms of accuracy, throughput, and latency, achieving 92.5 percent spectrum sensing accuracy. Furthermore, it improves throughput by 10% and reduces latency by 20%, which sets it apart from previous methods. We have compared the Spectrum Utilization Ratio (SUR) and Bit Error Rate (BER) results, demonstrating the effectiveness of the proposed algorithm in cross-layer communication, as well as in spectrum control and management. This research addresses the integration issues to enhance cognitive radio, paving the way for the creation of the new generation of wireless communications systems.