vix.ing · top · new · best · stats

Channel Polarization: A Method for Constructing Capacity-Achieving Codes for Symmetric Binary-Input Memoryless Channels

2008/07/24 by Erdal Arikan, Erdal Arıkan · 1 voice · 4,464 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Engineering · Mathematics · #Algorithm #Arithmetic #Binary code #Binary number #Binary symmetric channel #Cellular Automata and Applications #Channel (broadcasting) #Channel capacity #Channel code #Computer science #DNA and Biological Computing #Decoding methods #Electronic engineering #Engineering #Error Correcting Code Techniques #Mathematics #Polarization (electrochemistry) #Telecommunications #Theoretical computer science #cs.IT #math.IT

paper · pdf · doi:10.1109/tit.2009.2021379

published in IEEE Transactions on Information Theory 55(7), 3051-3073 (Institute of Electrical and Electronics Engineers) · The version which appears in the IEEE Transactions on Information Theory, July 2009

arxiv published 2008/07/24 · openalex publication_date 2009/06/16 · arxiv created 2009/07/20 · arxiv updated 2016/11/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/05

Abstract

A method is proposed, called channel polarization, to construct code sequences that achieve the symmetric capacityI(W)of any given binary-input discrete memoryless channel (B-DMC)W. The symmetric capacity is the highest rate achievable subject to using the input letters of the channel with equal probability. Channel polarization refers to the fact that it is possible to synthesize, out ofNindependent copies of a given B-DMCW, a second set ofNbinary-input channels\WN(i):1≤ i≤ N\such that, asNbecomes large, the fraction of indicesifor whichI(WN(i))is near1approachesI(W)and the fraction for whichI(WN(i))is near0approaches1-I(W). The polarized channels\WN(i)\are well-conditioned for channel coding: one need only send data at rate1through those with capacity near1and at rate0through the remaining. Codes constructed on the basis of this idea are called polar codes. The paper proves that, given any B-DMCWwithI(W)> 0and any target rateR ≪ I(W), there exists a sequence of polar codes\\Fraktur Cn;n≥ 1\such that\Fraktur Cnhas block-lengthN=2n, rate≥ R, and probability of block error under successive cancellation decoding bounded asPe(N,R) ≤ O(N^-1\over 4)independently of the code rate. This performance is achievable by encoders and decoders with complexityO(Nlog N)for each.

Cited by

Discussions

Related