Symbol Based Log-MAP in Concatenated LDPC-Convolutional Codes

Abstract

In this paper we study the use of a high rate Low Density Parity Check (LDPC) codes in concatenated coding structures. Specifically, we use the LDPC code as an outer code, with a convolutional code as an inner code. We decode the convolutional code using a symbol based Log-MAP (Maximum a posteriori probability) decoder, and feed the soft outputs of this decoder into a non-binary Galois Field LDPC decoder. We compare this concatenation scheme using 16 QAM modulation with one using a bit based Log-MAP decoder over Additive White Gaussian Noise (A WGN) and Stanford University Interim (SVI-3) channel model. The new proposed decoding technique has shown improvements for both channel models.