Research On Digital Predistortion Technique In Broadband Wireless MIMO-OFDM System

This paper focuses on the digital predistortion technique in the transmitter of multiple input multiple output(MIMO)- orthogonal frequency division multiplexing(OFDM) systems. In order to solve the key problems and challenges of existing digital predistortion(DPD) technology in broadband wireless MIMO-OFDM communication systems, two categories of adaptive baseband DPD methods applied for wireless MIMO systems are studied including the DPD method of compensating for crosstalk, and the DPD method of compensating for IQ imbalance.First of all, the evolution and difficult problems of MIMO digital predistortion are briefly summarized. Then the MIMO transmitter's predistortion architecture and corresponding theoretical foundation are illustrated, followed by the introduction of high power amplifier(HPA) nonlinear model, identification structure and evaluation criteria of DPD performance. Furthermore, two primary interference effects of MIMO predistortion system are introduced, namely, crosstalk interference and IQ imbalance. Both of them are analyzed in terms of their generation cause, model description and influence on system performance. Subsequently, several existing typical MIMO predistortion methods are analyzed and compared, including the power spectral density(PSD), the error vector magnitude(EVM) and the normalized mean square error(NMSE).Since most of the existing MIMO digital predistortion methods only compensate for the crosstalk interference, but the IQ imbalance is ignored, an improved Crossover Digital Predistorter(CO-DPD) method combined with IQ compensator is proposed. The idea of this method is that an IQ compensator is added to the feedback link of CO-DPD, so that both the IQ imbalance and crosstalk can be compensated simultaneously. Furthermore, the relevant power spectrum density(PSD), adjacent channel power ratio(ACPR), error vector magnitude(EVM) and normalized mean square error(NMSE) are evaluated through computer simulation. Comparing with traditional CO-DPD method, it is demonstrated that the out-of-band spectral regrowth of signal can be reduced effectively, meanwhile, EVM can be reduced more than 10 d B, and NMSE be reduced 15 d B when the crosstalk is-20 d B, gain imbalance 3% and phase imbalance 3~0.Moreover, the implementation complexity issue is also considered. Generally, most of the existing MIMO digital predistortion methods estimate the coefficients of predistorter and crosstalk based on Least Squares(LS) algorithm, resulting in high computational complexity. To tackle with this issue, a joint MIMO adaptive predistortion method to compensate for both crosstalk interference and IQ imbalance is proposed in this paper. In this method, the coefficient of crosstalk and predistorter can be extracted by using an efficient adaptive learning approach, such as LMS and RLS algorithm. The effectiveness of the proposed DPD method is well confirmed through a comprehensive theoretical analysis and sufficient numerical simulation results.