Architecture to store path history in a trellis and its application to Viterbi decoding
Sadiq M. Sait · International Journal of Electronics · 1992
In applications of dynamic programming, a certain number of alternatives (paths) are explored and, as more and more information is gathered, some paths may merge. In Viterbi decoding, the number of new paths created is equal to the number of paths discarded. In this paper we present an architecture that can be used to store and update the paths dynamically. It consists of a mapping of the trellis of the Viterbi decoding into a 2D array of simple cells. We show how the path storage can be efficiently implemented in VLSI.