Efficient Decoders for Turbo Product Codes

Abstract

Turbo codes and low density parity check (LDPC) codes are a class of high performance error correcting codes that closely approach the channel capacity set by Shannon. Turbo product codes (TPC) are one of the turbo codes categories constructed using two or more simple linear block codes. The main reason behind concatenation is to achieve better bit error rate (BER) performance with manageable complexity. TPCs are particularly powerful error correcting codes at high code rates. LDPC codes are a special subclass of linear block codes that contains a small density of 1's in the parity check matrix that result in large minimum distance which positively a¤ect the BER performance. Like turbo codes, LDPC codes have an error correcting capability near Shannon's limit. In this thesis, various linear block codes and their characteristics are studied and simulated. Moreover, extensive Monte Carlo simulations are conducted to evaluate the BER performance of non-sequential (NSQ) decoding algorithms over additive white Gaussian noise (AWGN) and fading channels. NSQ is a new decoding method developed to improve the BER performance of relatively high code rates by decoding the received codeword iteratively but not sequentially. Additionally, the BER performance for coded orthogonal frequency division multiplexing (OFDM) systems over non-linear channel has been considered. It has been demonstrated that the BER performance degradation due to simple peak-to-average power ratio (PAPR) reduction, such as clipping, does not have major impact on coded systems. In this thesis, a technique named failure mode (FM) has been devised for better BER performance with negligible complexity for LDPC codes. In this technique, the received decoded signal is not maintained if it does not satisfy a pre-de…ned criterion. Another novel decoding technique, named turbo LDPC (TLDPC) is proposed. This algorithm applies double iterative decoding by using the turbo concept with LDPC decoding algorithm. It was demonstrated that this technique has a good BER performance when FM is applied. Furthermore, the complexities of particular TLDPC codes were estimated.

Date of Award	Dec 2012
Original language	American English
Supervisor	Arafat Aldweik (Supervisor)