An adaptive two‐dimensional orthogonalization scheme for non‐linear digital canceller in full‐duplex radios

Abstract

In this letter, an adaptive two-dimensional orthogonalization (ATDO) scheme for non-linear digital canceller is proposed to realize selfinterference (SI) suppression in full-duplex radios. The proposed scheme consists of three parts. First, an adaptive lattice predictor is adopted to orthogonalize the input data over sample delay dimension. Then, a gradient-based adaptive orthogonalizer is developed to further orthogonalize the basis functions established with predictor output over non-linear order, and a new set of two-dimension orthogonal basis functions is constructed consequently. Finally, a least mean square (LMS) model-based parameter learning approach is utilized to implement the coefficients estimation with the newly established two-dimension orthogonal basis functions. The proposed ATDO scheme provides an effective solution to track quickly time-varying interference with high accuracy for real-time communication system. Experimental results illustrate that the proposed scheme can efficiently improve the cancellation performance of non-linear digital canceller, especially for higher order non-linear distortion model compared to other algorithms in previous literatures.