Self-Interference Channel Training for Full-Duplex Massive MIMO Systems.

Area of Science:

• Wireless Communication Systems
• Signal Processing
• Information Theory

Background:

• Full-duplex (FD) technology promises higher spectral efficiency for next-generation wireless systems.
• A key challenge in FD is mitigating self-interference (SI) caused by simultaneous transmission and reception.
• Effective SI cancellation relies heavily on accurate estimation of the SI channel.

Purpose of the Study:

• To propose a novel partial SI channel training method for FD massive MIMO systems.
• To minimize residual SI power under limited training overhead.
• To optimize pilot allocation for improved system performance.

Main Methods:

• Developed a partial SI channel training framework where only a subset of SI channel vectors are trained.
• Implemented a strategy to reuse previous estimates for untrained SI channel vectors.
• Optimized the partial training strategy to minimize expected residual SI power, considering Rician fading channels.

Main Results:

• The proposed partial training scheme significantly reduces residual SI power compared to full training.
• It outperforms other simple partial training schemes in FD massive MIMO systems.
• Achieved improved sum-rate performance under limited training overhead.

Conclusions:

• The novel partial SI channel training method is effective for FD massive MIMO systems.
• This approach enhances spectral efficiency by minimizing self-interference.
• The method offers a practical solution for real-world FD network deployment.