Blind Source Separation For Full-Duplex Systems: Potential and Challenges

Full-duplex communications systems that transmit and receive simultaneously suffer self-interference due to the mixing of the transmitted signal and the weaker received signal at the same node. The problem becomes compounded in Multi-Input Multi-Output (MIMO) systems, where considerable overhead is dedicated to training. In this article, we discuss using blind source separation techniques, namely Independent Component Analysis (ICA) to reduce training overhead in MIMO in-band full-duplex wireless communication systems. Practical limitations are discussed and experimental results that compare ICA to traditional Least Square approaches are presented, showing the superiority of ICA, especially in low SNR regimes.