A fast block-sorting algorithm for lossless data compression

Abstract

Summary form only given. Introduces a new transformation for block-sorting data compression methods. The transformation is similar to the one presented by Burrows and Wheeler, but avoids the drawbacks of uncertain runtime and low performance with large blocks. The cost is a small compression loss and a slower back transformation. In addition to that it is well suited for hardware implementation. Typical applications include real-time data recording, fast communication lines, on the fly compression and any other task requiring high throughput. The difference between this transformation and the original block-sort transformation is that the original transformation sorts on unlimited context, whereas this transformation sorts on limited context (typically a few bytes) and uses the position in the input block to determine the sort order in the case of equal contexts.