Blind LMS algorithm with multiple error cost function for self-interference compensation in full-duplex relays

Full-duplex relays will play a key role in future networks since they increase the system diversity, coverage, and throughput; ideally doubling the spectral efficiency. However, the full duplex operation produces coupling between the transmitter and receiver front ends that drastically reduces the relay performance and it can even turn it unstable. Then, to allow a full duplex operation it is necessary to compensate for the coupling. The autocorrelation pattern of the baseband received signal is perturbed by the relay self-interference. We propose a blind adaptive least mean squared (LMS) algorithm to compensate for the coupling that employs the known second-order statistics of the source signal as a reference. In order to improve the canceler performance, we propose a novel multiple error cost function that includes the information of several autocorrelation lags to adapt the canceler coefficients. Additionally, we present a study of the proposed criterion and analyze the stationary points of the adaptive algorithm. Finally, we show the canceler performance using simulations.