Is it possible to use Huffman coding on information sources with memory? If yes, then how? Do you know some references on this topic ?
Try Chapter 13 of the book by Cover and Thomas. It discusses optimal coding in the context of k'th order Markov processes
You can use HC on any source, even if in the presence of memory effect it is not optimal. I Agree with Eric_Fabre: you can exploit redundancy of the information source and encode only unpredictable information.
I'd store state of information and information string in subsequent tags and encode progressively to simulate memory towards adaptive compression with Huffman encoding.
Huffman coding can be used as long as there is a first order probability distribution available for the source, but it does not mean the encoding process will be optimal for sources with memory.
The simplest solution: you should code the sufficiently long blocks of symbols.
- Badges
- Science topic