English

Log-logarithmic Time Pruned Polar Coding

Information Theory 2020-12-14 v1 math.IT

Abstract

A pruned variant of polar coding is proposed for binary erasure channels. For sufficiently small ε>0\varepsilon>0, we construct a series of capacity achieving codes with block length N=ε5N=\varepsilon^{-5}, code rate R=CapacityεR=\text{Capacity}-\varepsilon, error probability P=εP=\varepsilon, and encoding and decoding time complexity bC=O(loglogε)\text{bC}=O(\log\left|\log\varepsilon\right|) per information bit. The given per-bit complexity bC\text{bC} is log-logarithmic in NN, in CapacityR\text{Capacity}-R, and in PP; no known family of codes possesses this property. It is also the second lowest bC\text{bC} after repeat-accumulate codes and their variants. While random codes and classical polar codes are the only two families of capacity-achieving codes whose NN, RR, PP, and bC\text{bC} were written down as explicit functions, our construction gives the third family. Then we generalize the result to: Fix a prime qq and fix a qq-ary-input discrete symmetric memoryless channel. For sufficiently small ε>0\varepsilon>0, we construct a series of capacity achieving codes with block length N=εO(1)N=\varepsilon^{-O(1)}, code rate R=CapacityεR=\text{Capacity}-\varepsilon, error probability P=εP=\varepsilon, and encoding and decoding time complexity bC=O(loglogε)\text{bC}=O(\log\left|\log\varepsilon\right|) per information bit. The later construction gives the fastest family of capacity-achieving codes to date on those channels.

Keywords

Cite

@article{arxiv.1905.13340,
  title  = {Log-logarithmic Time Pruned Polar Coding},
  author = {Hsin-Po Wang and Iwan Duursma},
  journal= {arXiv preprint arXiv:1905.13340},
  year   = {2020}
}

Comments

13 pages, 13 figures; we extend arXiv:1812.08106 and remove "BEC" from title