Moderate Deviation Asymptotics for Variable-Length Codes With Feedback

DOI10.1109/TIT.2019.2902146zbMATH Open1432.94086arXiv1707.04850OpenAlexW2962693394MaRDI QIDQ5224079FDOQ5224079

Authors: Lan V. Truong, Vincent Y. F. Tan 

Publication date: 19 July 2019

Published in: IEEE Transactions on Information Theory (Search for Journal in Brave)

Abstract: We consider data transmission across discrete memoryless channels (DMCs) using variable-length codes with feedback. We consider the family of such codes whose rates are $h{o}_N$ below the channel capacity $C$, where $h{o}_N$ is a positive sequence that tends to zero slower than the reciprocal of the square root of the expectation of the (random) blocklength $N$. This is known as the moderate deviations regime and we establish the optimal moderate deviations constant. We show that in this scenario, the error probability decays sub-exponentially with speed $exp(-(B/C)Nh{o}_N)$, where $B$ is the maximum relative entropy between output distributions of the DMC.

Full work available at URL: https://arxiv.org/abs/1707.04850

zbMATH Keywords

data transmission across discrete memoryless channels (DMCs)moderate deviations regimevariable-length codes with feedback

Mathematics Subject Classification ID

Large deviations (60F10) Prefix, length-variable, comma-free codes (94A45) Source coding (94A29)

Cited In (3)

• Finite blocklength analysis of energy harvesting channels
• Moderate Deviations in Channel Coding
• The Error Exponent of Variable-Length Codes Over Markov Channels With Feedback

This page was built for publication: Moderate Deviation Asymptotics for Variable-Length Codes With Feedback

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5224079)