Coding for channels with localized errors: The non-binary cases
The author considers the following coding problem: Messages are sent over a ``channel such that the encoder knows in advance the locations where an error might occur. The decoder does not know these locations in advance. The problem is to find the maximum number of messages M(n,t,q) which can be encoded into words of wordlength n over an alphabet having q symbols (using the knowledge over the possible error locations) such that the decoder can decode correctly (i.e. find the message back). The author finds upper and lower bounds for M(n,t,q) which are asymptotically equal i.e. he finds the exact value of \(\lim_{n\to \infty}(1/n) \log M(n,{\mathcal T}n,q).\) In the binary case \((q=2)\) this result was already obtained by Bassalygo, Gelfand and Pinsker and presented at the 4th Swedish-Soviet International Workshop on Information Theory.
- Coding in a channel with generalized defects and random errors
- On the locality of codeword symbols in non-linear codes
- Bounds for codes in case of separately localized errors
- scientific article; zbMATH DE number 3700721 (Why is no real title available?)
- Linear-time binary codes correcting localized erasures
- scientific article; zbMATH DE number 764514 (Why is no real title available?)
This page was built for publication: Coding for channels with localized errors: The non-binary cases
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q752664)