In-band full duplex (IBFD) radio represents one of the key technologies for future wireless communication and radar applications. A major challenge of this technology is to mitigate the strong self-interfer-ence (SI) so that the residual SI level falls below the receiver's noise floor. Radio frequency (RF) self-inter-ference cancellation (SIC) is essential for preventing an IBFD receiver from becoming saturated by the SI. We commence with an in-depth review of the promising analog least mean square (ALMS) adaptive filtering architecture, conceived for RF SIC in the IBFD radio RF front-end. The cancellation circuits employing this architecture can be implemented purely by analog components without any involvement of more power-thirsty digital signal processing. The behaviors, performance, and implementation of the ALMS loop are presented. Finally, their applications in various IBFD radios are discussed, and future research directions are provided.