| In 1993, a new class of error-correcting code named Turbo codes was proposed by C.Berrou etal in France, whose performances are close to the Shannon limit when the data block length is long enough. Turbo codes are parallel concatenated recursive systematic convolutional codes. Their encoders consist of two constituent recursive systematic convolutional sub-encoders linked by an interleaver. While the decoding is performed by two concatenated component decoders with an iterative decoding algorithm.A systematic analysis of Turbo codes, especially the interleaving technique of Turbo codes has been given in this paper. First, the encoder and decoder basic structure of Turbo codes are introduced. Then, the fundamental knowledge of interleaver technique is introduced, and the block interleaver is compared with the random interleaver. The comparison shows that the latter may be the better one. However, with increased interleaver length, the performance of traditional S-random interleaver cannot be suitable for Turbo codes. So we advance a new random interleaver, which is the improvement of the traditional S-random interleaver, to be used in high speed mobile communication systems for the next generation. The simulation results show that the new one is better than the traditional one on generating time and performance. Finally, from the analysis of distance spectrum, we advance two ways to improve on the "error floor" of Turbo codes. From the relative analysis of the extrinsic information and the input information, we find a new way to design interleaver, which from the computer simulation results has better performance than others as a whole.Although there are still many problems unsolved on interleaver design. It may take effect on the whole performance of Turbo codes and it is a key technique for Turbo codes design. So interleaver design is a research focus on the coding community. Turbo-coding is a break through in error-control coding theory and is expected to be widely used.
|
PDF Full Text Request