Neural Network Self-Interference Cancellation for Full-Duplex Wireless Systems

Video Demonstration

Simulation Images and Output Snapshots

Project Overview

A neural-network digital canceller that learns nonlinear transmitter leakage and suppresses residual self-interference in an in-band full-duplex link.

The model is structured around nonlinear self-interference modeling, ANN regression, residual cancellation and full-duplex link performance. It is suitable for scholars who need a clear implementation path, measurable outputs and a page that connects the video demonstration with the underlying engineering method.

System Architecture and Main Components

• Full-duplex transmitter and receiver chains
• Self-interference coupling channel
• Power-amplifier nonlinearity
• Reference-signal feature generator
• ANN regression canceller
• Residual interference and BER analyzer

MATLAB / Simulation Methodology

1. Generate desired and self-interference baseband signals.
2. Model linear multipath coupling and nonlinear transmitter distortion.
3. Create training features from delayed in-phase and quadrature samples.
4. Train the neural network to estimate received self-interference.
5. Subtract the estimate and evaluate cancellation, EVM and BER.

Control and Analysis Strategy

The central technical emphasis is nonlinear self-interference modeling, ANN regression, residual cancellation and full-duplex link performance. Measurements are converted into controller or analysis variables, limits are applied to maintain realistic operation, and disturbances are introduced to evaluate stability, tracking quality, efficiency and transient performance.

The implementation can be extended with parameter optimization, artificial-intelligence control, comparative algorithms, hardware-in-the-loop preparation or publication-style performance indices, depending on the research objective.

Expected Simulation Outputs

• Received and cancelled signal spectra
• Self-interference cancellation in dB
• Training loss and regression performance
• Residual error waveform
• BER or EVM before and after cancellation

Video Summary and Simulation Transcript

The video begins with the complete Neural Network Self-Interference Cancellation for Full-Duplex Wireless Systems model and identifies the principal subsystems: Full-duplex transmitter and receiver chains, Self-interference coupling channel, Power-amplifier nonlinearity, Reference-signal feature generator.

It then explains the signal flow and demonstrates nonlinear self-interference modeling, ANN regression, residual cancellation and full-duplex link performance. Reference commands and operating conditions are applied so that the controller, converter or physical model can be observed during steady-state and transient operation.

The final scopes focus on received and cancelled signal spectra, self-interference cancellation in db, training loss and regression performance, residual error waveform. These plots support result discussion, controller comparison, report preparation and further PhD or FYP development.

Research Applications and Possible Extensions

• In-band full-duplex radio research
• Nonlinear interference mitigation
• 6G physical-layer studies
• ANN-based communication projects
• Controller or algorithm comparison using identical operating scenarios
• Parameter sensitivity, optimization and publication-style result analysis

Related Simulation Projects