On the Analog Self-Interference Cancellation for Full-Duplex Communications With Imperfect Channel State Information

This paper studies the performance of the analog multi-tap (MT) canceller, where the tap coefficients are calculated based on the estimated self-interference (SI) channel state information (CSI). In this paper, both the time-invariant and the time-varying SI channels scenarios are investigated, considering the dynamic range of the analog-to-digital converter (ADC) and the linearity of the receiver chain. Closed-form expressions are first developed to calculate the residual SI power after the MT cancellation, characterizing the joint effects of the imperfect SI CSI, reconstruction errors on the estimated SI CSI, and the variation of the SI channels. Then, the achievable rate of the full-duplex transceivers is derived as a function of the residual SI power after the MT cancellation, dynamic range of the ADC, and the linearity of the receiver chain. Theoretical and simulated results show that with imperfect SI CSI, deploying more taps may harm the amount of analog SI cancellation. The sensitivity of the canceller to the doppler frequency shift reduces the amount of analog SI cancellation, and thus brings rate gain loss even with a doppler negligible in conventional communications.