Polar-like Codes and Asymptotic Tradeoff among Block Length, Code Rate, and Error Probability

TL;DR

This paper introduces a comprehensive framework for analyzing the fundamental tradeoffs among block length, code rate, and error probability in polar-like codes over arbitrary channels, unifying and extending previous results.

Contribution

It develops a generalized asymptotic tradeoff analysis for polar-like codes, connecting error exponents and scaling exponents to provide a unified tradeoff characterization.

Findings

Interpolating tradeoff among block length, rate, and error probability established.

Grafted polar coding variant nearly matches random code performance with manageable complexity.

All points within a specific tradeoff triangle are achievable with the proposed framework.

Abstract

A general framework is proposed that includes polar codes over arbitrary channels with arbitrary kernels. The asymptotic tradeoff among block length $N$, code rate $R$, and error probability $P$ is analyzed. Given a tradeoff between $N,P$ and a tradeoff between $N,R$, we return an interpolating tradeoff among $N,R,P$ (Theorem 5). $\def\Capacity{\text{Capacity}}$Quantitatively, if $P=\exp(-N^{\beta^*})$ is possible for some $\beta^*$ and if $R=\Capacity-N^{1/\mu^*}$ is possible for some $1/\mu^*$, then $(P,R)=(\exp(-N^{\beta'}),\Capacity-N^{-1/\mu'})$ is possible for some pair $(\beta',1/\mu')$ determined by $\beta^*$, $1/\mu^*$, and auxiliary information. In fancy words, an error exponent regime tradeoff plus a scaling exponent regime tradeoff implies a moderate deviations regime tradeoff. The current world records are: [arXiv:1304.4321][arXiv:1501.02444][arXiv:1806.02405] analyzing Ar{\i}kan's codes over BEC; [arXiv:1706.02458] analyzing Ar{\i}kan's codes over AWGN; and [arXiv:1802.02718][arXiv:1810.04298] analyzing general codes over general channels. An attempt is made to generalize all at once (Section IX). As a corollary, a grafted variant of polar coding almost catches up the code rate and error probability of random codes with complexity slightly larger than $N\log N$ over BEC. In particular, $(P,R)=(\exp(-N^{.33}),\Capacity-N^{-.33})$ is possible (Corollary 10). In fact, all points in this triangle are possible $(\beta',1/\mu')$-pairs. $$ \require{enclose} \def\r{\phantom{\Rule{4em}{1em}{1em}}} \enclose{}\r^\llap{(0,1/2)}_\llap{(0,0)} \enclose{left,bottom,downdiagonalstrike}\r_\rlap{(1,0)} \enclose{}\r $$