Iterative Eualization and Source Decod´ing for Vector Quantized Sources

In this contribution an iterative (turbo) channel equalization and source decoding scheme is considered. In our investigations the source is modelled as a Gaussian-Markov source, which is compressed with the aid of vector quantization. The communications channel is modelled as a time-invariant channel contaminated by intersymbol interference (ISI). Since the ISI channel can be viewed as a rate-1 encoder and since the redundancy of the source cannot be perfectly removed by source encoding, a joint channel equalization and source decoding scheme may be employed for enhancing the achievable performance. In our study the channel equalization and the source decoding are operated iteratively on a bit-by-bit basis under the maximum a-posteriori (MAP) criterion. The channel equalizer accepts the a priori information provided by the source decoding and also extracts extrinsic information, which in turn acts as a priori information for improving the source decoding performance. Simulation results are presented for characterizing the achievable performance of the iterative channel equalization and source decoding scheme. Our results show that iterative channel equalization and source decoding is capable of achieving an improved performance by efficiently exploiting the residual redundancy of the vector quantization assisted source coding