English

Local Decoding in Distributed Compression

Information Theory 2022-09-28 v2 math.IT

Abstract

It was recently shown that the lossless compression of a single source XnX^n is achievable with a notion of strong locality; any XiX_i can be decoded from a constant number of compressed bits, with a vanishing in nn probability of error. By contrast, we show that for two separately encoded sources (Xn,Yn)(X^n,Y^n), lossless compression and strong locality is generally not possible. Specifically, we show that for the class of ``confusable'' sources, strong locality cannot be achieved whenever one of the sources is compressed below its entropy. Irrespective of nn, for some index ii the probability of error of decoding (Xi,Yi)(X_i,Y_i) is lower bounded by 2O(d)2^{-O(d)}, where dd denotes the number of compressed bits accessed by the local decoder. Conversely, if the source is not confusable, strong locality is possible even if one of the sources is compressed below its entropy. Results extend to an arbitrary number of sources.

Keywords

Cite

@article{arxiv.2204.07518,
  title  = {Local Decoding in Distributed Compression},
  author = {Shashank Vatedka and Venkat Chandar and Aslan Tchamkerten},
  journal= {arXiv preprint arXiv:2204.07518},
  year   = {2022}
}

Comments

11 pages