Spatially Coupled Codes in Turbo Equalization

Abstract: In this thesis we investigate the application of spatial coupling in turbo equalization. We show that with spatial coupling we can have a fresh perspective on the trade-off between performance in the waterfall and error floor region often encountered in the choice of codes for turbo equalization. Describing turbo equalization in the perspective of message passing on factor graphs, we show various ways of introducing memory in the encoding process and compare the performance improvement through simulations. We also discuss an effective way of decoding such spatially coupled system by using window decoding in order to minimize the decoding latency and complexity. We manged to show that with spatial coupling we can have it both ways; that is, have a good performance in the waterfall region while having low error floors.