Analysis of the iterative decoding of product codes using the Gaussian approximation.pdfVIP

Analysis of the Iterative Decoding of Product Codes Using the Gaussian Approximation F. Lehmann and G. M. Maggio Center for Wireless Communications (CWC), UC San Diego, 9500 Gilman Drive La Jolla, CA 92093-0407 Abstract— In this paper we introduce a model of the iter- available. The validity of this assumption was first recog- ative decoding of product codes. For each iteration, an ap- nized for LDPC and turbo codes by Wiberg [9] and used proximation of the bit error rate (BER) at the output of the in [10] to perform an analytical analysis of LDPC codes. decoders is used to evaluate the density of the extrinsic in- formation. The decoding thresholds obtained for the additive The Gaussian approximation in conjunction with Monte- white Gaussian noise (AWGN) and the Rayleigh fading chan- Carlo simulations has also been proposed to analyze the nel are consistent with the simulation results available in the performances of the turbo decoding algorithm [11], [12]. literature. Previous work concerning Gaussian approximations has relied on different parameters in order to obtain a one- dimensional model, namely the mean value in [10], the I . IN TRODU CTION SNR (signal-to-noise ratio) in [11], [12], the mutual in- Product codes where introduced by Elias [1] as a class formation in [13], [14] and the BER [14], [15]. Another of powerful concatenated block codes with high minimum method based on matching mean and covariance