Recursive flexible convolutional encoders for parallel concatenation

In this paper, path pruning and puncturing are employed for the construction of recursive systematic convolutional encoders that can vary both their rate and decoding complexity. The basic goal is to use the resulting encoders in parallel concatenated coding schemes, leading to flexible turbo code constructions. A family of good constituent encoders is provided. Simulation results indicate that with the proposed method, in various SNR ranges, a decrease in computational complexity can result even in an increase in performance.