卫星通信中一种快速RPEM算法

Fast RPEM algorithm in satellite communication

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.

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.