V. Arvind Rameshwar,V. Lalitha
Abstract:In this paper, we revisit the Recursive Projection-Aggregation (RPA) decoder, of Ye and Abbe (2020), for Reed-Muller (RM) codes. Our main contribution is an explicit upper bound on the probability of incorrect decoding, using the RPA decoder, over a binary symmetric channel (BSC). Importantly, we focus on the events where a single iteration of the RPA decoder, in each recursive call, is sufficient for convergence. Key components of our analysis are explicit estimates of the probability of incorrect decoding of first-order RM codes using a maximum likelihood (ML) decoder, and estimates of the error probabilities during the aggregation phase of the RPA decoder. Our results allow us to show that for RM codes with blocklength $N = 2^m$, the RPA decoder can achieve vanishing error probabilities, in the large blocklength limit, for RM orders that grow roughly logarithmically in $m$.
What problem does this paper attempt to address?
### What problem does this paper attempt to solve?
The main objective of this paper is to provide an explicit upper bound for the error decoding probability of Reed - Muller (RM) codes when using the Recursive Projection Aggregation (RPA) decoder over the Binary Symmetric Channel (BSC). Specifically, the author focuses on the performance analysis in the case where the RPA decoder converges with only one iteration in each recursive call.
#### Main contributions:
1. **Explicit upper bound of error probability**: The paper provides an explicit upper bound of the error decoding probability of the RPA decoder over the BSC.
2. **High - probability correct decoding**: For a sufficiently large block length \( N = 2^m \), the RPA decoder can recover the correct input codeword with an extremely high probability, applicable when the order of RM codes is approximately on the logarithmic level of \( m \).
3. **Theoretical analysis supplement**: This work supplements previous research on RM codes, especially those results indicating that the order of RM codes grows linearly but the rate approaches zero.
4. **Low - complexity decoding algorithm**: It is proved that an explicit, low - complexity decoding algorithm (i.e., the RPA algorithm) can achieve an error probability approaching zero.
#### Specific problem description:
- **Background**: Reed - Muller codes are a class of binary linear codes obtained by evaluating Boolean polynomials on the Boolean hypercube. They are widely used in the fields of information theory and error - correcting coding, especially having capacity - achieving properties on the Binary Erasure Channel (BEC) and more general Binary Memoryless Symmetric (BMS) channels.
- **Existing decoding algorithms**: There are already multiple decoding algorithms, such as the Fast Hadamard Transform (FHT) decoder, recursive decoding algorithms, etc. The recently proposed RPA decoder has shown performance close to the Maximum Likelihood (ML) performance for medium block lengths.
- **Research gap**: Although the RPA decoder performs well in practice, the formal analysis of its theoretical performance is still insufficient.
#### Core problems of the paper:
- **Provide theoretical guarantees**: The paper aims to fill this gap by conducting a strict theoretical analysis of the error probability of the RPA decoder to prove its good performance in the limit of large block lengths.
- **Determine the applicable range**: Study the maximum order of RM codes when the order \( r \) of RM codes grows with the parameter \( m \) while the RPA decoder can still maintain an error probability approaching zero.
#### Results and significance:
- **Upper bound formula**: For any \( 0 < \epsilon < \eta(p) \), when \( r\geq2 \), the upper bound of the error probability of RM codes is:
\[
P_{\text{err}}(\text{RM}(m, r)) \leq 32N^{r + 1}\cdot e^{-2^{-(r + 1)}\cdot N\epsilon^2}
\]
- **Asymptotic results**: For a fixed \( r \), when the block length \( N \) approaches infinity, the error probability approaches zero.
- **Future work**: Further study whether the RPA decoder can achieve a positive rate or even reach the capacity of the BSC on the BSC.
In summary, through strict theoretical analysis, this paper proves that the RPA decoder can achieve a very low error probability under specific conditions, thus providing strong theoretical support for the practical applications of RM codes.