Limiting Factors in Self-Interference Cancellation

Various experimental results have stated that the self-interference (SI) in full-duplex communication can be mitigated to properly detect the intended signal. However, relatively little is known about the cancellation limits of these systems. A good SI-cancellation strategy passes first by a proper understanding of the nature of this SI. This is highly related to the physical implementations of radio-frequency (RF) transceivers that suffer from a variety of non-idealities defying the perfect knowledge of the received SI. In this chapter, we investigate the various factors that can limit the SI-cancellation performance in full-duplex wireless communications systems. We first classify the known SI-cancellation architectures based on where the reference signal is taken to cancel the SI. By combining the effects of the transceiver impairments, estimation error and SI channel, our analysis reveals that the main bottleneck SI-cancellation turns out to be either the quantization noise, the local oscillator phase noise or the estimation error, depending on the selected SI-cancellation architecture. We provide illustrative numerical results that can facilitate the SI-cancellation architecture selection to meet the specific target system requirements.