为进一步提高超奈奎斯特(FTN)码元速率传输系统中的信号检测性能、减轻符号间干扰并降低复杂度,在FTN系统中引入预编码算法,主要包括:奇异值分解(SVD),几何均值分解(GMD),统一信道分解(UCD)。将三种不同预编码方法、未预编码直接硬判决及部分判决反馈均衡(PDFE)仿真并比较得到:UCD预编码在任何信噪比条件下均优于GMD预编码,SVD预编码和PDFE性能在高信噪比时比GMD预编码性能要差,SVD预编码性能和直接硬判决相比略差。 In order to further improve the signal detection performance, reduce intersymbol interference and reduce the complexity in the ultra-Nyquist (FTN) symbol rate transmission system, a pre-coding algorithm is introduced in the FTN system, including: singular value decomposition (SVD) Geometric Mean Decomposition (GMD), Unified Channel Decomposition (UCD). Three different precoding methods, un-precoded direct hard decision and partial decision feedback equalization (PDFE) are simulated and compared: UCD precoding is better than GMD precoding, SVD precoding and PDFE performance under any signal-to-noise ratio The performance of the SVD pre-coding is slightly worse than that of the GMD pre-coding performance at high SNR.