Context-tree-based CSI compression
What is it?
We propose a novel context-tree-based approach to compressing time-varying CSI for wireless communications. The proposed scheme combines lossy vector quantisation, by means of data-adapted companders, with lossless compression, based on symbol probabilities estimated by a context-tree model.
How does it work?
Our scheme combines two steps.
- Vector quantisation. The new vector quantisation technique is based on a class of parametrised companders applied on each component of the normalised vectors. In particular, we propose the β-law compander, inspired by the beta distribution. Our algorithm optimises the compander parameters whenever empirical data are available.
- Data compression. Then, we compress the sequence of quantisation indices using a context-tree-based approach. Essentially, we build and regularly update the context-tree maximising (CTM) model, and encode the current symbol at each instant with the estimated distribution.
Resources
- Implementation codes are available on GitHub.
Journal paper
H.K. Miyamoto and S. Yang, “Context-Tree-Based Lossy Compression and Its Application to CSI Representation”, IEEE Transactions on Communications, early access, 2022. [IEEEXplore] [arXiv] [BibTeX]
Conference paper
H.K. Miyamoto and S. Yang, “A CSI Compression Scheme Using Context Trees”, International Zurich Seminar on Information and Communication (IZS), Zurich, 2022, pp. 24-28. doi: 10.3929/ethz-b-000535273. [Link] [BibTeX]