卫星通信中一种快速RPEM 算法

记忆多项式预失真能够有效地补偿卫星信道的非线性失真，但存在收敛速度慢的缺点．将RPEM算法应用于卫星记忆多项式预失真中，提出了一种快速收缩递归预测误差算法．该算法通过利用Shrinkage方法和L₁范数与L₂范数差值最小化公式，获得了噪声自由先验误差与噪声自由后验误差的关系，推导了一种最佳遗忘因子，有效地提高了收敛速度．仿真结果表明，收缩递归预测误差算法能够有效地抑制信号的星座扭曲和频谱再生，解决了非线性失真的问题，同时在几乎不增加计算复杂性的基础上，能够获得较快的收敛速度和较好的稳定性．

Fast RPEM algorithm in satellite communication

The memory polynomial pre-distortion is an effective technique to compensate the nonlinearity for the satellite channel. However, its adaptive algorithm has the defect of the slow convergence speed. To counteract this problem, a fast shrinkage Recursive Prediction Error Method (SH-RPEM) algorithm is devised for the memory polynomial pre-distortion over the nonlinear satellite channel. By using a known Shrinkage method and L₁-L₂ minimization formulation, the relationship between the noise-free posteriori and priori error signal is obtained. Furthermore, an optimal forgetting factor which can significantly enhance the convergence speed is proposed by minimizing the mean square of the noise-free posteriori error signal. Simulation results show that the SH-RPEM algorithm can effectively suppress constellation wrapping and spectrum regrowth of the signal and solve the nonlinear distortion. Moreover, the SH-RPEM algorithm can achieve a faster convergence speed and better stability without increasing the computational complexity.