Skip to main content


Wireless communication systems are constantly evolving to enable the delivery of new services and satisfy the user’s demands. These new applications (e.g., augmented/virtual reality, connected cars, 4k/8k video content) require a considerable increase in transmission capacity. Traditionally, improving the physical layer waveform spectrum efficiency has been the primary driver for increasing the link capacity. Nevertheless, the channel coding solutions, constellation designs, or other techniques already offer results close to the Shannon limit, and therefore, there is not much room for improvement. Consequently, the electromagnetic spectrum poses a hard limit for implementing high-capacity use cases. Therefore, the current standards and technologies need to improve the offered spectral efficiency rate to adapt to the new use cases from a new perspective. A good alternative is to enable full-duplex communications in 5G/6G, which, theoretically, can double the spectral efficiency compared with half-duplex systems. Nevertheless, the main drawback of this idea is the self-interference or loopback signal that is generated and added to the received signal. In this context, Artificial Intelligence (AI) appears as a promising candidate to downgrade the impact of the loopback signal and increase the performance of full-duplex communication systems. 

Our proposal is to analyze different AI techniques and test their performance with traditional loopback signal attenuation mechanisms. Particularly, our work focuses on enhancing two specific receiver chain modules using AI: beamforming and digital signal cancellation. First, in the case of beamforming, the directivity of the receiver array antenna is adapted based on the channel state information to minimize the loopback component and maximize the desired reception. In this case, the objective of the AI is to calculate the optimum antenna pattern. Then, the digital cancellation module uses AI techniques to improve the loopback channel estimation and enhance the loopback signal cancellation.