[Paper Review] On Pre-transformed Polar Codes

Bin Li, Huazi Zhang+1 more|arXiv (Cornell University)|Dec 13, 2019

Error Correcting Code Techniques3 references39 citations

TL;DR

The paper shows that pre-transformations using upper-triangular matrices (e.g., CRC, PC, CC) do not reduce minimum distance of the code, and can reduce the number of minimum-distance codewords, explaining why pre-transformed polar codes can outperform Polar/RM codes.

ABSTRACT

In this paper, we prove that any pre-transformation with an upper-triangular matrix (including cyclic redundancy check (CRC), parity-check (PC) and convolution code (CC) matrix) does not reduce the code minimum distance and an properly designed pre-transformation can reduce the number of codewords with the minimum distance. This explains that the pre-transformed polar codes can perform better than the Polar/RM codes.

Motivation & Objective

• Motivate the study of pre-transformations in polar coding.
• Analyze how pre-transformations affect code distance properties.
• Demonstrate that upper-triangular pre-transformations preserve minimum distance but can reduce minimum-distance codeword multiplicity.
• Explain performance implications for pre-transformed polar codes versus Polar/RM codes.

Proposed method

• Prove that any pre-transformation with an upper-triangular matrix does not reduce the code minimum distance.
• Consider specific pre-transformations such as CRC, parity-check, and convolution code matrices.
• Provide theoretical justification for how properly designed pre-transformations influence the minimum-distance codeword count.

Experimental results

Research questions

• RQ1 Do upper-triangular pre-transformations affect the minimum distance of the code?
• RQ2 Can designed pre-transformations reduce the number of minimum-distance codewords without increasing the minimum distance?
• RQ3 Why do pre-transformed polar codes potentially outperform Polar/RM codes?
• RQ4 What role do CRC, PC, and CC matrices play in the distance properties of the transformed code?

Key findings

• Upper-triangular pre-transformations do not reduce the code minimum distance.
• Properly designed pre-transformations can reduce the number of minimum-distance codewords.
• Pre-transformed polar codes can perform better than Polar/RM codes due to reduced minimum-distance codeword multiplicity.
• The results explain the performance benefits observed for pre-transformed polar codes.