Nonlinear Digital Self-Interference Cancellation in In-Band Full-Duplex Systems with Complex-Valued Temporal...

Area of Science:

• Wireless communication engineering
• Signal processing
• Machine learning for communications

Background:

• Full-duplex (FD) technology enables simultaneous transmission and reception on the same frequency, boosting spectral efficiency.
• Integrated sensing and communication (ISAC) architectures leverage FD for enhanced capacity and reduced latency.
• Nonlinear self-interference (SI) from hardware imperfections hinders practical FD system deployment.

Purpose of the Study:

• To propose an efficient method for modeling and canceling nonlinear self-interference (SI) in full-duplex (FD) systems.
• To address the limitations of conventional polynomial models in characterizing complex nonlinear SI.
• To improve the performance and reduce the computational complexity of SI cancellation (SIC).

Main Methods:

• Development of a complex-valued temporal convolutional network (CV-TCN) utilizing a wavelet activation function for nonlinear SI modeling.
• Implementation of the CV-TCN as an SI canceller within an FD-ISAC architecture.
• Performance evaluation through simulations comparing CV-TCN against traditional polynomial-based and other SI cancellation techniques.

Main Results:

• The proposed CV-TCN model demonstrates superior nonlinear SI cancellation (SIC) performance compared to conventional methods.
• The CV-TCN canceller exhibits lower inference complexity and higher parameter efficiency.
• The CV-TCN effectively suppresses residual SI power by approximately 7.95 dB after linear SIC, maintaining it 2.23 dB above the noise floor.

Conclusions:

• The CV-TCN offers a powerful and efficient solution for nonlinear SI modeling and cancellation in FD systems.
• This approach significantly enhances the feasibility and performance of FD-based ISAC architectures.
• The proposed method represents a notable advancement in overcoming hardware limitations for next-generation wireless communication systems.