Approximate Linear Time ML Decoding on Tail-Biting Trellises in Two Rounds

TL;DR

This paper introduces a fast, approximate maximum likelihood decoding algorithm for tail-biting trellises that operates in linear time with only two rounds, significantly improving efficiency over previous methods.

Contribution

It presents a novel two-round linear time decoding algorithm for tail-biting trellises, reducing complexity from O(mlogm) to O(m).

Findings

Algorithm achieves linear time complexity.

Simulation results show near-ML performance.

Reduces decoding time significantly.

Abstract

A linear time approximate maximum likelihood decoding algorithm on tail-biting trellises is prsented, that requires exactly two rounds on the trellis. This is an adaptation of an algorithm proposed earlier with the advantage that it reduces the time complexity from O(mlogm) to O(m) where m is the number of nodes in the tail-biting trellis. A necessary condition for the output of the algorithm to differ from the output of the ideal ML decoder is reduced and simulation results on an AWGN channel using tail-biting rrellises for two rate 1/2 convoluational codes with memory 4 and 6 respectively are reported