5 The source-channel coding theorem.pptVIP

5. The source-channel coding theorem * 第五章 信源-信道编码定理 Any real communication system can be subsumed under the model of Figure. The encoder block represents all the data processing performed on the source output before transmission. The channel is assumed to accept channel input symbols at discrete intervals, and response to emit channel to emit channel output symbols. The “decoder” block represents all the data processing performed on the channel output before delivery to the destination. We are given the channel’s input alphabet and output alphabet. Also, we are given the channel statistics. For each positive integer n and ? ? 0, define Cn(? ) by 给定信道输入字母表和输出字母表，同时给定信道统计分布。n和? ? 0，定义Cn(? )为 The capacity-cost function C(? ) is defined by 容量-代价函数 C(? )定义为 (a) each codeword satisfies b (xi) ? n? (b) (c) , i = 1, 2, …, M. Channel Coding Theorem: Fix ?0 ? ?min. Then for any triple (?, C , ? ) with ? ?0, C C(?0 ), ? 0, there exist for all sufficiently large n, a code {x1, x2, …, xM} of length n and a decoding rule such as We are given a source alphabet, a destination alphabet and the “distortion” d (u, v). Also, we are given the source statistics. For each positive integer k and ? ? ? min define Rk (? ) by 给定信源字母表、信宿字母表和失真d (u, v)，同时给定信源统计分布。对于所有的正整数k和? ? ? min，定义Rk (? )为 The rate-distortion function of the source is defined by 信源的率失真函数定义为 Source coding Theorem：Fix ? ?min. Then for any pair (? ‘, R ’) with ? ‘ ? and R ’ R (?), for all sufficiently large k there exists a source code C = (v1, v2, …, vM) of length k such that: （a）M ? 2 ?k R ?； （b）d (C) ? 。 The average cost The average distortion The rate of transmission Theorem 5.1 (the source-channel coding theorem) For a given source and channel: (a) The parameters must satisfy (b) Conversely, given numbers ?0 ? ?min, ? ?min, and r C(? )/ R (? ), it is possible to design a system of the type depicted in Figure such that