多电平PAM系统的最佳均衡理论

详细论述了多电平ＰＡＭ（脉冲幅度调制）系统的最佳均衡理论。首先建立了多电平ＰＡＭ系统的一种二进制码间干扰等效模型，将之视为二进制多路系统，从而说明其均稀可以用多个独立的二进制均衡器联事完成。在将均衡问题视为分类判决问题的基础上，根据贝叶斯准则导出了多电平ＰＡＭ系统的最佳均衡有达式和所能取得的最小符号概率的计算公式和近似计算公式。从最佳均衡解可以看到，无论是二进制通信系统还是多电平ＰＡＭ系统，其最佳

Qptimun Equalization Theory of Multi-level Pulse Amplitude Modulation Systems

The optimum equalization theory for overcoming the intersymbol interferences (ISI) in multi-level pulse amplitude modulation (PAM) systems is establishead. A kind of equivalent binary ISI model of PAM systems is put forward. By viewing the PAM system as binary multiplexed systems, it is shown that its equalization may be implemented jointly by serval independent binary equalizers, based on the viewpoint that equalizations belong to classfication decision problems, the optimum equalization solutions are obtained according to the bayes decision rule. Both the exact and approximate formula to compute theminimum error-symbol -probability, which the PAM system can reach by usingthe ideal optimum equalizer, are also obtained. According to the optimum equalization solution expressions, it is explained that, the optimum equalization solutions are highly nonlinear; and the trditional linear structur equalizers are incapable of implemnting such optimum equalization, therefore the equalizers must be incorporeted into nonlinearity to do so. The neural network,which is a nonlinear dynamic system, has been successfully applied in the channel equalization of binary digital communication systems. Obviously, the results obtained in this paper also suggest an effective way, to apply neural network inmulti-level PAM systems to implement optimum channel equalizations.