가중 비이진 RA 부호의 새로운 중단 기준 제안

Abstract

In this thesis, we develope a new stopping criterion for the weighted nonbinaryrepeat-accumulate (WNRA) codes over GF(q). The decoding complexityof WNRA codes increases in proportion to the modulation order. Similar tobinary repeat-accumulate (RA) codes, a stopping criterion can be a very effectivemethod in order to reduce the decoding complexity of WNRA decoders withsome performance degradation.The message vectors in nonbinary belief-propagation (BP) decoder is updatedin every iteration. In addition, one of the elements in message vectors becomesmore dominant whenever the iteration proceeds. In this reason, observing thebehaviors of message vectors in every iterations can provide the information ofdecoding progress indicating the possiblity of successful decoding.We propose a new stopping criterion which requires negligible additionalcomplexity with little performance degradation. By choosing the thresholdvalue adequately, the proposed stopping criterion have the additional complexityonly in bit-wise comparison operations. The simulation results show thatthe decoding algorithm by using the proposed stopping criterion reduces theaverage number of iterations of BP decoders for WNRA codes over GF(q) withq-ary frequency-shift keying (FSK) by less than one half compared to the algorithmby using the fixed number of iterations within 0.05 dB frame error rate(FER) performance degradation at FER 10¡1.