A serially concatenated BCH-Turbo code scheme over an Additive White Gaussion Noise channel

Abstract

This work deals with the implementation of a serially concatenated BCH-turbo code scheme for faster decoding over an Additive White Gaussian Noise (AWGN) channel. Turbo codes, first presented in 1993, represent the most innovative and breakthrough work in error control coding in recent times. These codes offer nearcapacity performance on deep space and satellite channels. Turbo codes are essentially parallel concatenated convolutional codes. Research in turbo codes involved combining some dormant concepts with novel techniques for decoding, to offer the kind of performance that these codes do. Turbo codes use an iterative decoder which could take up a considerable amount of processing time at the receiver. Using a sole turbo code for deep space applications requires several iterations between the multiple decoders at the receiver. Typically no less than 10 iterations would be required for the decoding algorithm to converge so that no significant loss in performance is observed. This work attempts to look at one way to reduce the decoding delay by using a concatenated coding scheme. The outer BCH code is essentially used to clear out any residual errors that may remain after the decoding of the inner turbo code. The proposed coding scheme employs an inner turbo code and an outer BCH code which would permit the decoder to operate at fewer iterations so that the turbo code decoding process speeds up and at the same time provides enough protection (error correcting capability) within the BCH code, so that no loss in performance is observed due to the reduced iterations within the turbo code. BCH codes are relatively simple to decode using the Berlekamp-Massey algorithm and have enough algebraic structure within them to use where multiple error-correcting capability is required. It is also likely that the concatenated coding scheme would be less complicated than a fully iterative turbo decoder because of the intensive computations required at each iteration of the turbo decoder.