Improving block sorting compression using Simulated Annealing

Abstract

Data compression becomes a common technique with large data processing these days. Burrows-Wheeler trans formation is one of the last algorithms developed for text compression. This algorithm uses lexicographic sorting of the block of data and produces sequences of sequences of same symbols. These sequences are long, but may be longer when we modify the sorting to group symbols which has the same predecessors. The problem is that the number of possible permutations is huge and cannot be simply searched by the exhaustive search. Therefore, we test Simulated Annealing as a searching algorithm.