Chase-Pyndiah Et. Al. : How Do You Could Be Used?

1228 Words5 Pages
SOVA-based decoding is generally inferior to the MAP-based decoding in terms of BER performance. The authors show that the original SOVA introduces two types of distortions in its output. They show that the performance of the SOVA can be improved by normalizing its output and eliminating the correlation between intrinsic and the extrinsic information derived in each decoding iteration. A modified updating rule can also be applied to improve performance. It was discovered that the original SOVA proposed in [48] omits some updates for ease of implementation, leading to overestimation of the reliability [51]. It was proved that the modified SOVA of [48] is equivalent to Max-log-MAP algorithm [52]. Application of SOVA-based SISO decoding for…show more content…
either +1 or –1; where +1 represents a bit 1 and –1 represents a bit 0); and it is possible that one or both of c+ and c– may not exist due to the fact that the Chase algorithm operates using a subset of the 2n possible codewords. In the event that there are not two competing codewords, the Pyndiah algorithm uses a scaling factor, βm, which approximates the value of the extrinsic information for the mth decoding. These values, which are a function of m, are specified in [47]. The values for βm in [44] are approximated using computer simulation of a (64,57)2 BTC (the superscript “x” indicates x dimensional BTC). A simple, yet effective alternative for approximating βm, which does not require computer simulation, is presented in [53]. A soft-input value, λi,j, is computed by the decoder for each entry in C. These soft-values are used to select the codewords in each subset, the competing codewords, as well as, in computing the extrinsic information. The soft-input is computed using: (10) where αm is a scaling factor, which is either pre-defined for increasing values of m [44], or computed after every decoding by assuming that the extrinsic information is governed by Gaussian statistics [53]; a method that is very similar to the one used in computing the scaling factor in [50]. The above equations demonstrate how to compute the soft-input and extrinsic likelihood values for a row decoder. For the

    More about Chase-Pyndiah Et. Al. : How Do You Could Be Used?

      Get Access