抄録
In this paper, a fast model identification algorithm based on Givens rotation is presented for the digital predistortion (DPD) application. Because most of the computation work is operated in the hardware, the new algorithm has much faster identification speed than traditional recursive least square (RLS) method or least square (LS) method. Compared with common fast identification algorithms, the new algorithms exhibits the higher identification precision. And final experiment shows that the adjacent channel power ratio (ACPR) after the DPD with the new algorithm is -50dB, 8dB better than that with the conjugate gradient (CG) method, which verifies that the new algorithm has very good performance on both identification precision and identification speed.
