Talk:Turbo code

From Scholarpedia
Jump to: navigation, search

    In my opinion, the article is not ready for publication. Substantial revisions are needed before publication.

    1) In my opinion, the article should be written in tutorial form. However, the current version is a mixture of background information and advanced material.

    The authors very briefly start with parallel concatenated encoders, and then immediately switch to a serial concatenation (Fig. 1, Precursor, Fig. 2). Even the "genesis of turbo codes" is explained on basis of a serial concatenation (Fig. 2).

    In order to solve this problem, I would suggest to start with a simple 2D block turbo code based on SPC consistuent encoders [Lodge et al.]. Depending on whether the parity check bit of the parity bits is used or not, we may talk about serial or parallel concatenation. Given this tutorial example, the exchange of extrinsic information can nicely be explained.

    The first formula is not understandable, unless it is mentioned that the systematic part of the second encoder (not code!!!) is punctured. The concept of extrinsic information is not clear from the two LLR formulas. I would suggest to drop all formulas.

    The section on weight distributions should also be skipped, because it is not helpful for the average reader.

    The discussions on "families of SISO algorithms" is wrong and should also be dropped. (The SOVA is a reduced-complexity version of the BJCR. With a slight modification of the SOVA, both algorithms can be made identical.)

    2) Substantial previous work is missing. It is mentioned that Turbo codes have been invented at ENST Bretange in 1993. This is certainly true for parallel concatenated codes based on convolutional codes. It should be mentioned, however, that in [Lodge et al.] "block product codes with iterative decoding" AND "convolutional product codes with iterative decoding" (for short: "turbo block product codes" and "turbo convolutional product codes") have been invented in parallel, see also reference [5] therein (1992!).

    [Lodge et al.] J. Lodge, R. Young, P. Hoeher, and J. Hagenauer, "Separable MAP 'filters' for the decoding of product and concatenated codes," in Proc. IEEE ICC '93, Geneva, Switzerland, May 1993, pp. 1740-1745.

    Personal tools
    Namespaces

    Variants
    Actions
    Navigation
    Focal areas
    Activity
    Tools