The Problem

For the past century or so, wireless transceivers have typically operated in a half-duplex fashion, orthogonalizing transmission and reception in the time domain and/or the frequency domain. This is to prevent self-interference, which manifests between the transmitter and receiver when a device attempts to receive while transmitting over the same spectrum. Self-interference combines with a desired receive signal (i.e., the signal of interest) over the air before making its way through the receive chain of the full-duplex device. Successful reception of a desired receive signal is often virtually impossible since self-interference is typically many orders of magnitude stronger than a desired receive signal (which has typically propagated a far distance). This has motivated devices to operate in half-duplex fashion to circumvent self-interference.

A transceiver operating in a full-duplex fashion subjects itself to self-interference.
A transceiver operating in a full-duplex fashion subjects itself to self-interference.

The Solution

To successfully equip a transceiver with full-duplex capability, a means to mitigate self-interference is necessary. The approach to mitigating self-interference is perhaps most naturally summarized by the following question.

If a device knows what signal it transmitted, why can’t it subtract that transmitted signal from the signal it receives to subtract self-interference and leave only the desired signal?

The simple fact that a full-duplex transceiver knows its own transmit signal offers it the exciting potential to cancel self-interference. While knowing its transmit signal is powerful, it is not trivial to cancel self-interference since a filtered version of the transmitted signal reaches the receiver, not the transmitted signal itself. Put simply, self-interference is not a clean copy of the transmitted signal but rather a filtered one. Nonetheless, a full-duplex transceiver can attempt to synthesize self-interference by estimating the self-interference channel that the transmit signal propogates through. This synthesis of self-interference can be executed in the analog domain (i.e., in hardware) or digitally (i.e., in software/logic).

Analog self-interference cancellation aims to reconstruct and cancel self-interference in hardware.
Analog self-interference cancellation aims to reconstruct and cancel self-interference in hardware.

The figure above depicts analog self-interference cancellation, where a (physical) tunable filter is placed between the transmitter and receiver. This filter is driven by a portion of the transmit signal and is injected at the receiver. The full-duplex transceiver digitally estimates the self-interference channel during a quiet period, for instance, by transmitting a known signal and comparing it to the received signal. The analog self-interference cancellation filter is then configured to replicate this estimate of the self-interference channel to synthesize self-interference that closely matches the true self-interference. Then, the synthesized self-interference is inverted before injecting it at the receiver where it will cancel the true self-interference to leave the desired receive signal nearly interference free (if done properly).

A two-staged approach to self-interference cancellation, combining analog and digital stages.
A two-staged approach to self-interference cancellation, combining analog and digital stages.

Self-interference cancellation can analogously be executed digitally, where filtering and subsequent injection with the received signal can take place in software/logic, which may be more sophisticated and flexible than analog self-interference cancellation. There are practical factors that motivate the need for analog self-interference cancellation, meaning not all cancellation can take place digitally. As such, most practical routes to full-duplex use a two-staged approach: first cancelling a portion of self-interference in analog, then the majority of residual self-interference digitally.

Self-interference cancellation and full-duplex capability have been researched extensively since around 2010. There are a number of sophisticated approaches to cancel self-interference, including and beyond digital and analog self-interference cancellation, many of which tackle artifacts of practical transceivers such as hardware imperfections.

Below is a list of recommended readings on a variety of topics related to full-duplex and self-interference cancellation.

  • D. Bharadia et al., “Full Duplex Radios,” 2013.
  • D. Korpi et al., “Full-Duplex Transceiver System Calculations: Analysis of ADC and Linearity Challenges,” 2014.