Digital Self-Interference Cancellation Strategies for In-Band Full-Duplex: Methods and Comparisons

Area of Science:

• Wireless Communication
• Signal Processing

Background:

• In-band full-duplex (IBFD) systems enhance spectral efficiency by allowing simultaneous transmission and reception.
• Self-interference (SI) is a critical challenge hindering IBFD system deployment.
• Digital self-interference cancellation (SIC) is crucial for mitigating SI in multiple-input multiple-output (MIMO) systems.

Purpose of the Study:

• To evaluate digital SIC methodologies for MIMO wireless transceivers with digital beamforming.
• To model nonlinearities and memory effects from hardware impairments using advanced polynomial and neural network approaches.
• To assess the robustness of SIC under different training and adaptation schemes.

Main Methods:

• Two digital SIC approaches were evaluated: a Generalized Memory Polynomial (GMP) model and a Phase-Normalized Neural Network (PNN).
• Itô-Hermite polynomial basis functions were used to augment the GMP model.
• Estimation techniques including Least Squares (LS), Least Mean Squares (LMS), and Fast Kalman (FK) were employed for model coefficient estimation.
• Simulations incorporated realistic power amplifier (PA) characteristics, channel conditions, and high-order modulation schemes.

Main Results:

• For time-variant scenarios, a low-complexity adaptive SIC using GMP with FK estimation proved effective.
• In time-invariant scenarios, an open-loop SIC approach based on PNN demonstrated superior performance.
• The PNN approach maintained robustness across various modulation schemes.

Conclusions:

• Digital SIC is essential for practical IBFD MIMO systems.
• The choice between GMP and PNN models depends on channel time-variance.
• PNN-based SIC offers high performance and robustness in static channel conditions.