Student: Muhammed Ahmed Leghari
Advisor: Prof. Dr.-Ing. Werner Henkel
Time frame: Spring 2019
Equalization and decoding are either seen as two independent blocks. In case of a decision feedback equalizer, replacing the detector by a decoder is known to be impossible due to the decoding delay. This lead to Tomlinson-Harashima precoding as one of the solutions, thereby moving the feedback filter to the transmitter. Instead of an equalizer structure, one could use maximum-likelihood sequence estimation in the form of a Viterbi or BCJR algorithm and further link those to the decoder in a Turbo fashion, exchanging LLRs between the BCJR and the message passing algorithm of the LDPC decoder. Those schemes are known as Turbo equalization.
In this project, we like to investigate, if the equalizer can be more directly included into the LDPC decoding process. This follows a recent structural idea developed in the group.
Integrated solutions are otherwise known from the decoding of convolutional codes, where an equalizer is computed inside the paths of, e.g., a Viterbi decoder. This solves the delay issue as well, but leads to a significantly higher complexity, since the number of equalizers that have to processed is equal to the number of trellis states. In our integrated equlization-LDPC decoding proposal, we hope to be able to avoid a complexity increase, but may still require an increased number of iterations.