Spectrum sharing for radar-communication integration in high-dynamic scenarios: a CNN-enhanced beamforming approach

With the escalating deployment of wireless communication and radar systems, spectrum scarcity in high-speed dynamic environments (aviation, drones, vehicles) has become critical. Coexistence interference severely restricts coordinated development of wireless sensing and information transmission. This paper proposes a novel spectrum sharing framework: (1) utilizing Null Space Projection (NSP) and Small Singular Value Space Projection (SSVSP) for interference suppression; (2) designing an integrated beamforming strategy adaptive to high mobility; (3) integrating Convolutional Neural Network (CNN) for real-time waveform optimization. Results show the proposed method reduces SNR fluctuation by 50% and stabilizes Bit Error Rate (BER) below 10-4, outperforming traditional approaches in dynamic scenarios.