Analog Least Mean Square Loop for Self-Interference Cancellation in In-band Full-duplex Systems

Analog Least Mean Square (ALMS) loop is a promising method to cancel self-interference (SI) in in-band full-duplex (IBFD) systems. In this talk, the steady state analyses of the residual SI powers in both analog and digital domains are derived in frequency domain. It is proved that the ALMS loop amplifies the frequency components of the residual SI at the edges of the signal spectrum in the analog domain, but the matched filter in the receiver chain reduces this effect. This results in a significant improvement of the interference suppression ratio in the digital domain before information data detection. The lower bounds of the interference suppression ratio given by the ALMS loop in both analog and digital domains are then addressed. These lower bounds are proved to be joint effects of the loop gain, tap delay, number of taps, and transmitted signal properties. The discovered relationship among these parameters allows the flexibility in choosing appropriate parameters when designing the IBFD systems under given constraints. Finally, the effects of I/Q imbalance in the ALMS loop and the upper bound of the degradation on SI cancellation performance are then investigated. The degradation is proved to be insignificant even under severe conditions of I/Q imbalance. The upper bound of the degradation provides an essential reference for the system design.