SICNet—Low Complexity Sample Adaptive Neural Network-Based Self-Interference Cancellation in LTE-A/5G Mobile Transceivers

The limited transmitter-to-receiver stop-band isolation of the duplexers in long term evolution (LTE) and 5G/NR frequency division duplex transceivers induces leakage signals from transmitter(s) (Tx) into receiver(s) (Rx). These are root cause a multitude self-interference (SI) problems receiver path(s) diminishing receiver’s sensitivity. Traditionally, these effects counteracted by use various different SI cancellation (SIC) architectures which typically solely target one specific problem. In this paper, we propose two novel neural networks based that can handle variety without need for architecture each effect. We additionally show suitability proposed on occurring in-band full transceivers. Further, introduce low-cost training algorithms to enable online adaptation (as opposed offline currently literature). combination concepts is shown not only beat existing their performance, but also provide sufficiently low computational complexity allowing on-chip implementations.