Two classes of algorithms for asymptotically optimal simplified MLSD with application to digital radio

Two classes of simplified trellis decoding algorithms are defined and analyzed. Even with a significant simplification (several orders of magnitude) these algorithms yield an error probability which for large signal-to-noise ratio (SNR) is asymptotically the same as the one for Viterbi decoding, apart from a small, insignificant factor. Even for moderate to low SNRs the simplified methods have error probabilities comparable to those of the Viterbi algorithm. This shows that there is a great freedom in designing simplified trellis decoding algorithms, where many kinds of tradeoffs (simplicity vs. performance) can be made