Efficient encoding scheme for LT codes with soft iterative decoding

Abstract

The Luby transform (LT) code is the first practical implementation of the digital fountain code, and its utilisation over a noisy channel invokes iterative decoding with soft information. For a good decoding performance, the role of the encoder degree distribution is vital. Especially, when utilising the soft iterative decoder, the bit error rate (BER) performance of the LT code can be improved by increasing the number of the degree-1 encoding nodes and reducing the number of lower-degree variable nodes. This study proposes an efficient LT-code encoding algorithm that satisfies the conditions for improving the BER performance. The proposed scheme applies different priorities for selecting the neighbouring source nodes for given degrees of the encoding nodes. The simulation results demonstrate that the proposed scheme outperforms the existing schemes.