Demo: Real-Time Implementation of Optimal Nonlinear Self-Interference Cancellation for Full-Duplex Radio

Abstract

The full-duplex radio can potentially double the spectral efficiency with perfect self-interference cancellation. Traditionally, nonlinear digital self-interference cancellation (SIC) uses least mean squares (LMS) algorithms using Volterra series and Hammerstein basis expansions. However, this traditional approach slows down the convergence speed and degrades the cancellation performance due to the correlation among the nonlinear basis functions. In this demo, we develop the optimal nonlinear digital SIC for the IEEE 802.11a Wi-Fi full-duplex systems. Our approach harnesses the LMS algorithm built upon Ito-Hermite polynomials that form a set of the orthogonal basis for the complex Gaussian input process. We develop a software-defined radio full-duplex testbed compliant to the IEEE 802.11a Wi-Fi standards. Using this testbed, we show experimental results of the proposed optimal SIC algorithm and verify the superiority by comparing it with the existing SIC algorithms.