Polarization-Adjusted Convolutional (PAC) Codes as a Concatenation of Inner Cyclic and Outer Polar- and Reed-Muller-like Codes

Mohsen Moradi

2023-08-03

Abstract:Polarization-adjusted convolutional (PAC) codes are a new family of linear block codes that can perform close to the theoretical bounds in the short block-length regime. These codes combine polar coding and convolutional coding. In this study, we show that PAC codes are equivalent to a new class of codes consisting of inner cyclic codes and outer polar- and Reed-Muller-like codes. We leverage the properties of cyclic codes to establish that PAC codes outperform polar- and Reed-Muller-like codes in terms of minimum distance.

Information Theory

What problem does this paper attempt to address?

The problem this paper attempts to address is:

**How to demonstrate that Polarization-Adjusted Convolutional (PAC) codes are superior to Polar codes and Reed-Muller (RM) codes in terms of minimum distance, and to show that PAC codes can be viewed as concatenated codes with an inner convolutional code and outer Polar and RM-like codes.**

Specifically, the main contributions of the paper include:
1. **Proposing a new coding scheme**: PAC codes are equivalent to a concatenated code composed of an inner convolutional code and outer Polar and RM-like codes.
2. **Proving non-decreasing row weight**: Any odd number of clockwise cyclic shifts of any row of the generator matrix of Polar and RM-like codes, plus some next row, does not decrease the weight of that row.
3. **Utilizing the algebraic structure of convolutional codes**: Demonstrating that PAC codes are superior to Polar and RM-like codes in terms of minimum distance.
4. **Proving matrix properties**: Proving that the sum of any row of the generator matrix of Polar and RM-like codes with the next row equals some clockwise cyclic shifts of that row.

Through these contributions, the paper not only theoretically proves the superiority of PAC codes but also provides performance improvements in practical applications. For example, by choosing appropriate convolutional codes, the weight enumeration of PAC codes can be significantly improved, thereby outperforming RM codes in error correction performance.