Estimation strategies for channel side information applied to the decoding of LDPC codes over impulsive noise channels

ARNONE, L. J.; LIBERATORI, M. C.; RABIOGLIO, L.; CASTIÑEIRA MOREIRA, J.

Congreso; Information Processing and Control Workshop; 2019

UNS

In this paper the Bit Error Rate (BER) performanceof a Low-Density Parity-Check (LDPC) code operating over aMiddleton´s Class A is improved by modifying the initializationstep of the classic Sum-Product (SP) decoding algorithm forLDPC codes, implemented in its logarithmic form. The improvementis obtained by initializing the corresponding decoderwith initialization values calculated using the pdf function ofthe Middleton?s Class impulsive noise channel, evaluated overchannel samples for the two possible values ?0? and ?1?. Thisprocedure requires from the knowledge of channel parameters,which can be estimated by different methods. The best BERperformance obtained corresponds to the initialization done withperfect knowledge of channel parameters.Estimation methods considered in this paper are theKanemoto, Zabin - Poor and Expectation-Maximization (EM)methods. We have evaluated these estimation methods by usingthe estimated values of channel parameters needed to calculatethe values of the Middleton?s Class A impulsive noise channelpdf, for the two possible values ?0? and ?1?. We have foundthat none of the estimation methods utilized in its original formprovided good enough channel parameters estimations, to makethe corresponding decoder perform as in the case of perfectknowledge of the channel parameters.Averaging procedures have to be applied over the initialestimates obtained with the different methods to make thecorresponding decoding algorithm obtain the best BER performance.Among the different methods, averaging over estimationsprovided by the Zabin-Poor method appears to be the best optionif implementation complexity and performance are both takeninto account.