Abstract:Recently, a novel variation of polar codes known as polarization-adjusted convolutional (PAC) codes has been introduced by Arıkan. These codes significantly outperform conventional polar and convolutional codes, particularly for short codeword lengths, and are shown to operate very close to the optimal bounds. It has also been shown that if the rate profile of PAC codes does not adhere to certain polarized cutoff rate constraints, the computation complexity for their sequential decoding grows exponentially. In this paper, we address the converse problem, demonstrating that if the rate profile of a PAC code follows the polarized cutoff rate constraints, the required computations for its sequential decoding can be bounded with a distribution that follows a Pareto distribution. This serves as a guideline for the rate-profile design of PAC codes. For a high-rate PAC\,$(1024,899)$ code, simulation results show that the PAC code with Fano decoder, when constructed based on the polarized cutoff rate constraints, achieves a coding gain of more than $0.75$ dB at a frame error rate (FER) of $10^{-5}$ compared to the state-of-the-art 5G polar and LDPC codes.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: in Polarization - Adjusted Convolutional (PAC) coding, how to ensure that the computational complexity of its sequential decoding remains within a limited range. Specifically, when the rate configuration of PAC codes follows the polarization cut - off rate constraint, whether the decoding computation distribution of each data bit can be limited within the Pareto distribution. This provides guidance for designing PAC codes with low decoding complexity.
### Problem Background
PAC codes are a new type of polar codes. They are significantly superior to traditional polar codes and convolutional codes at short code lengths and are close to the optimal boundary. However, if the rate configuration of PAC codes does not satisfy certain polarization cut - off rate constraints, the computational complexity of its sequential decoding will increase exponentially. Therefore, how to ensure that the decoding complexity of PAC codes is controllable is an important research problem.
### Main Contributions
This paper solves the inverse problem of the above - mentioned problem by introducing the concept of metric polarization in [8], that is, it is proved that when the rate configuration of PAC codes follows the polarization cut - off rate constraint, the computational complexity of its sequential decoding can be limited by the Pareto distribution. Specifically:
1. **Theoretical Analysis**: Through detailed mathematical derivations, it is proved that when the bias term and rate \(R^-\) of the first half - code block are less than the cut - off rate \(R_0^-\) of the weak polarization channel, the computational distribution required for decoding the first data bit is upper - bounded by the Pareto distribution. Similarly, a similar proof is carried out for the second half - code block.
2. **Recursive Polarization Transformation**: Through the analysis of recursive polarization transformation, sufficient conditions for ensuring bounded decoding complexity are established.
3. **Numerical Analysis**: Based on the proposed rate configuration criteria, a high - rate PAC (1024, 899) code is constructed. At a frame error rate of \(10^{-5}\), compared with the state - of - the - art 5G polar codes and LDPC codes, an encoding gain of more than 0.75 dB is obtained while maintaining a low decoding complexity.
### Conclusion
The research results of this paper provide a theoretical basis for designing PAC codes with low decoding complexity and show their potential in practical applications. By following the polarization cut - off rate constraint, efficient sequential decoding can be achieved at high signal - to - noise ratios, thus providing a new coding scheme for future wireless communication systems.
### Related Formulas
1. **Bias Term and Branch Metric**:
\[
\gamma^-(s_i)\triangleq\phi^-(s_i; y_i, y_{N/2 + i})-b^-
\]
\[
\phi^-(s_i; y_i, y_{N/2 + i})\triangleq\log_2\frac{P(y_i, y_{N/2 + i}|s_i)}{P(y_i, y_{N/2 + i})}
\]
2. **Pareto Distribution Upper Bound**:
\[
P(C_n^-\geq L)\leq\left(\frac{4}{L(1 - 2^{-\epsilon/\beta})^2}\right)^\beta
\]
3. **Expected Value and Moment - Generating Function**:
\[
h(r)\triangleq\log_2E[2^{rX}]
\]
These formulas help explain how PAC codes achieve bounded decoding complexity under specific conditions.