Hardware architecture for lowering the error floor of LTE turbo codes

Turbo codes are well known error-correcting codes used in many communication standards. However, they suffer from error floors. Recently, a method - denoted as the flip and check algorithm - that lowers the error floor of turbo codes was proposed. This method relies on the identification of the least reliable bits during the turbo decoding process. Gains of about one order of magnitude were reached in terms of error rate performance. In this article, the first hardware implementation of the method is presented. The feasibility and hardware complexity are addressed by studying the impact of the algorithmic parameters of the technique. Synthesis results for FPGA implementations are reported and compared to turbo decoders implementations.