Reduction of encoding delay in compression of binary sources using turbo codes: A two-stage algorithm

An iterative algorithm is presented in the literature, to losslessly compress binary sources by using turbo codes. This algorithm searches for the smallest codeword length to guarantee zero distortion; and offers promising compression rates at the expense of large encoding delays. In this work we propose an improved encoding algorithm that works in two stages. At the first stage we estimate the codeword length. The estimated length does not guarantee zero distortion; however, we set a criterion to make sure that the distortion is below a defined threshold. At the second stage we improve our estimation to remove the remaining distortion and achieve lossless compression. Our simulations show that by careful choice of parameters: (i) our proposed algorithm achieves the same average compression rate as the algorithm appearing in the literature, and (ii) the delay introduced by our proposed algorithm, which is the sum of delays introduced by first and second stages, is less than the delay introduced by the algorithm appearing in the literature.