Margin Propagation based XOR-SAT Solvers for Decoding of LDPC Codes

Ankita Nandi,Shantanu Chakrabartty,Chetan Singh Thakur

2024-02-07

Abstract:Decoding of Low-Density Parity Check (LDPC) codes can be viewed as a special case of XOR-SAT problems, for which low-computational complexity bit-flipping algorithms have been proposed in the literature. However, a performance gap exists between the bit-flipping LDPC decoding algorithms and the benchmark LDPC decoding algorithms, such as the Sum-Product Algorithm (SPA). In this paper, we propose an XOR-SAT solver using log-sum-exponential functions and demonstrate its advantages for LDPC decoding. This is then approximated using the Margin Propagation formulation to attain a low-complexity LDPC decoder. The proposed algorithm uses soft information to decide the bit-flips that maximize the number of parity check constraints satisfied over an optimization function. The proposed solver can achieve results that are within $0.1$dB of the Sum-Product Algorithm for the same number of code iterations. It is also at least 10x lesser than other Gradient-Descent Bit Flipping decoding algorithms, which are also bit-flipping algorithms based on optimization functions. The approximation using the Margin Propagation formulation does not require any multipliers, resulting in significantly lower computational complexity than other soft-decision Bit-Flipping LDPC decoders.

Information Theory,Signal Processing

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the performance gap issue in low - density parity - check (LDPC) code decoding. Specifically, the existing bit - flipping algorithms have a performance gap compared with the benchmark decoding algorithms (such as the sum - product algorithm SPA) when decoding LDPC codes. The paper proposes an XOR - SAT solver based on margin propagation (MP) for LDPC code decoding, aiming to maximize the number of satisfied parity - check constraints by using soft information to decide bit - flipping, and this solver can achieve performance close to that of the sum - product algorithm with lower computational complexity.

### Main contributions of the paper:
1. **Proposing the MP - XOR - SAT framework**: This framework uses piecewise - linear approximation (margin propagation) to implement the XOR - SAT solver.
2. **Reducing computational complexity**: The MP - XOR - SAT decoder does not require multiplication operations, thus significantly reducing the computational and hardware implementation complexity.
3. **Performance improvement**: When used for LDPC code decoding in the AWGN channel, the MP - XOR - SAT algorithm has a bit - error rate (BER) and frame - error rate (FER) that are only 0.1 dB different from those of the sum - product algorithm (SPA) with the same number of code iterations. At the same time, the number of iterations it requires is 10 times less than that of the state - of - the - art gradient - descent bit - flipping algorithm.

### Solutions:
1. **Objective function design**: First, an objective function \(H_{\text{log}}\) using the log - sum - exponential function is designed, which eliminates the multiplication operations in the objective function.
2. **Normalization factor**: A normalization factor \(\sum_{j = 1}^N r_jd_jq_j\) is introduced to capture the correlation between the decoded bits and the channel information, thereby breaking the symmetry between codewords and making the global maximum correspond to the codeword closest to the channel output.
3. **Gradient - ascent update**: The continuous variable \(q\) is updated using the gradient - ascent method, and the probability scores in the log - sum - exponential function are replaced by margin - propagation (MP) probability scores to simplify the calculation.
4. **Bit - flipping strategy**: According to the gradient - update results, if the value of a certain variable \(q_j\) is below the threshold \(\theta\), then the corresponding binary variable \(d_j\) is flipped.

### Experimental results:
- **Influence of the normalization factor**: The experimental results show that the normalization factor significantly improves the performance of the decoder, especially in a high - noise environment.
- **Multi - bit flipping and state update**: Through the multi - bit flipping and state - update strategies, the convergence speed and decoding performance of the decoder are further improved.

### Conclusion:
The paper proposes an XOR - SAT solver based on margin propagation for LDPC code decoding. This method is not only close to the benchmark algorithm in performance but also has significant advantages in computational complexity and is suitable for efficient decoding in practical applications.