针对最大熵方法(MEM)存在的迭代计算量大和误差扩散的问题，以及 ESPＲIT算法存在的平坦快衰落信道预测精度受制于自相关函数估计的问题，文中提出基于扩展Prony 算法的平坦快衰落信道预测算法． 扩展 Prony 算法通过最小二乘法拟合和计算高次代数方程来求复根，无需估计自相关函数且可批量计算，能够提高预测精度和运算效率．理论和仿真结果表明，扩展 Prony 算法较 MEM 和 ESPＲIT 算法具有更高的预测精度，同预测精度下具有 3 ～5dB 的信噪比优势，且算法性能稳定，运算效率高．

In order to overcome the high iteration load and error diffusion existing in maximum entropy method (MEM) and to improve the prediction accuracy of ESPＲIT algorithm subjected to autocorrelation function estima-tion，a flat fast-fading channel prediction algorithm on the basis of extended Prony algorithm is proposed． By means of extended Prony algorithm，complex roots can be obtained via least square fitting and higher algebraic equations without estimating autocorrelation function，batch computation becomes possible，and thus both the prediction accu-racy and the computation efficiency improve． Theoretical and simulated results show that，in comparison with MEM and ESPRIT algorithms，extended Prony algorithm provides higher prediction accuracy，3 ～5 dB advantages at the same prediction accuracy，more steady performance and higher operation efficiency．