Convergence behavior of iterative source-channel decoding

Whenever digital communication is subject to transmission errors, the utilization of reliability information at the receiver is henefi-cial. Soft-channel decoding techniques improve the ermr correcr-ing capabilities and in addition, sofrbir source decoding performs error concealmenr. While the reliability gain of channel decoding is based on artificial diversities explicitly introduced at the transmitter , sofrbit sauce decoding exploits implicit redundancies remaining in the hitstream after state-of-the-art source encoding. In an iterative source-channel decoding scheme the reliability gains of both are exchanged iteratively in order to enhance the common error resistance. However, first investigations have shown that the number of profitable iterations is limited to small values. In this paper, the convergence behavior is analyzed using a mod-em tool called EXIT cham [l]. EXIT charts permit to predict the convereence behavior bv studvine the individual comoonents. As I ~ I-Figure 2: Receiver for iterative source-channel decoding; To sim-a novelty, EXIT characteristics are specified for sofrbit source decoding. The convergence analysis is confirmed by simulation. plify matters notation regards the deinterleaved domain 1. ITERATIVE SOURCE-CHANNEL DECODING At the receiver reliahilitv information about the transmission of At time instant r. a source encoder determines a set & , of M source codec parameters u ~ , ~ with K = 1 ,. .. A4 denoting the position within the set. The single components u-, ~ in the set t~, are assumed to be statistically independent among each other. Each value U=,,, which is continuous in magnitude but discrete in time, is individually quantized by ZK" reproduction levels +It). The reproduction levels are invariant with respect to T and the whale set i s given by 0,. To each quantizer reproduction level iit' specified at time instant r a unique bit pattem x ~ , ~ is assigned. Therein, K , denotes the length of x % , ~. The complete set of hit pattems specified at time instant r is given by x, (see Figure 1). Figure 1: Transmitter for iterative source-channel decoding scrambles the incoming data bit stream in a deterministic manner. Thereby, the interleavermight be sized such that several consecutive sets s7 are rearranged in common. If so, the interleaver introduces a delay, which might be unacceptable in practical applications. However, the interleaver has to be designed such that the diversities resulting either from source or from channel encoding allow independent reliability gains at the receiver. As diversity of source encoding is …