Blind identification and real-time calibration of memory nonlinearity based on RLS algorithm

Regular broadband software radio receivers could not simultaneously satisfy the requirements for high power efficiency and high dynamic acquiring. This paper develops a real-time digital calibration system based on recursive least square algorithm. A blind identification criterion for the minimizing of the total energy of the nonlinear distortions in the compensated output is designed as the characteristic of the input signal can hardly be obtained by the receiver in advance. The coefficients of the nonlinear model are measured and adaptively updated with the nonstationary input signals. Results on both multi-tone and 16-QAM modulating signals show that with the proposed system, the Spurs-Free-Dynamic-Range (SFDR) of the effective receiver's front-end working in nonlinear region would achieve 20 dB's improvement.