A cost-effective nonlinear self-interference canceller in full-duplex direct-conversion transceivers

Abstract

Cancellation of self-interference (SI) arising due to hardware nonidealities is a key issue in the design of mobile-scale full-duplex devices in future full-duplex (FD) communications. To efficiently address this issue in the baseband, we propose a cost-effective adaptive digital SI cancellation scheme for FD direct-conversion transceivers (DCTs) in the presence of major circuit imperfections, such as frequency-dependent I/Q imbalance, power amplifier (PA) distortion, thermal noise and quantization noise. This is achieved by employing a recent complex dual channel estimation (CDC) framework which offers low computational complexity in the widely nonlinear model fitting task which underpins this problem, to yield a dual channel nonlinear complex least-mean-square (DC-NCLMS) based SI canceller. For rigor, a unified theoretical evaluation is further performed to illustrate the second order performance optimality and low computation requirements of DC-NCLMS in both the transient and steady state stages. Further, for enhanced convergence, the proposed SI cancellation structure is equipped with an affine projection based scheme and a prewhitening procedure. Simulations in practical FD DCT settings, compliant with orthogonal frequency division multiplexing (OFDM)-based IEEE 802.11ac standard, support the findings.