Geometry-Based Statistical Channel Models of Reflected-Path Self-Interference in Full-Duplex Wireless

Self-interference (SI) channel characteristics are crucial for SI cancellation (SIC) techniques of in-band full-duplex (IBFD) wireless communications. This paper develops two geometry-based statistical channel models-geometry-based concentric ellipses model (GBCEM) and geometry-based concentric circles model (GBCCM)-for full-duplex systems in micro- and macrocell environments, respectively, to analyze the characteristics of reflected-path SI (RSI) compared with the desired signal (DS). The probability density functions of incident angle and path length are derived to identify the fading characteristics of RSI and DS. The simulations based on GBCEM and GBCCM intuitively present the signal characteristics by the signal envelope in the time domain, the statistics of signal envelope, the autocorrelation properties, and the power spectral density. We find that the fading characteristics of RSI and DS can be modeled as Rayleigh when reflectors sit far away from the receiver. We also find that RSI is still likely to be Rayleigh fading, even if DS is not when reflectors scatter in a specific region restricted by the range of incident angle. Our works on the characteristics of RSI and DS are generally helpful to explore a more efficient SIC mechanism for IBFD systems, and the developed models are useful for both simulation and analysis purposes.