On APP Decoding

Carlos R.P. Hartmann, Syracuse University
Luther D. Rudolph, Syracuse University
Kishan Mehrotra, Syracuse University, Department of Electrical Engineering and Computer Science
Guy J. Snedeker, Syracuse University



In this paper we show that APP decoding for a linear code C is optimum not for C, but for a minimum-distance-2 code which contains C as a subcode when the codewords of are transmitted with equal probability. However, APP decoding is shown to be a symptotically optimum for C for high SNR when C is a binary one-step orthogonolizable code with equiprobable codewords transmitted over the AWGN channel.