Novel irregular LDPC codes and their application to iterative detection of MIMO systems

Low-density parity-check (LDPC) codes are among the best performing error correction codes currently known.

For higher performing irregular LDPC codes, degree distributions have been found which produce codes with optimum performance in the infinite block length case. Significant performance degradation is seen at more practical short block lengths. A significant focus in the search for practical LDPC codes is to find a construction method which minimises this reduction in performance as codes approach short lengths.

In this work, a novel irregular LDPC code is proposed which makes use of the SPA decoder at the design stage in order to make the best choice of edge placement with respect to iterative decoding performance in the presence of noise. This method, a modification of the progressive edge growth (PEG) algorithm for edge placement in parity-check matrix (PCM) construction is named the DOPEG algorithm. The DOPEG design algorithm is highly flexible in that the decoder optimisation stage may be applied to any modification or extension of the original PEG algorithm with relative ease. To illustrate this fact, the decoder optimisation step was applied to the IPEG modification to the PEG algorithm, which produces codes with comparatively excellent performance. This extension to the DOPEG is called the DOIPEG.

A spatially multiplexed coded iteratively detected and decoded multiple-input multiple-output (MIMO) system is then considered. The MIMO system to be investigated is developed through theory and a number of results are presented which illustrate its performance characteristics. The novel DOPEG code is tested for the MIMO system under consideration and a significant performance gain is achieved.