Communication and Radar Signal Separation in Overlapping Frequency Bands Using Conv-TasNet

In modern communication systems, the coexistence of communication and radar signals within the same frequency band presents a significant challenge due to potential interference. This study investigates the application of Conv-TasNet, a deep learning model originally developed for speech separation, to the problem of separating overlapping communication and radar signals. By simulating scenarios where these signals overlap, we demonstrate that Conv-TasNet can effectively distinguish between the two, even under challenging interference conditions. The model processes the In-phase and Quadrature-phase (I/Q) components of the mixed signals, enabling accurate separation and subsequent demodulation of the communication signal. The experimental results indicate a tendency for the Bit Error Rate (BER) to decrease across all tested modulation schemes as the Signal-to-Interference Ratio (SIR) increases. For SIR levels above -26 dB, Binary Phase Shift Keying (BPSK) and Quadrature Phase Shift Keying (QPSK) modulation schemes achieve a BER below 10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">–2</sup>, while 16-Quadrature Amplitude Modulation (16-QAM) maintains a BER below 10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup>. These findings highlight the potential of Conv-TasNet to enhance the reliability of communication systems in environments with significant radar interference.