Hassan Noghrei

Hassan Noghrei, Mohammad-Reza Sadeghi, Wai Ho Mow

The quality and quantity of data used for training greatly influence the performance and effectiveness of deep learning models. In the context of error correction, it is essential to generate high-quality samples that are neither excessively noisy nor entirely correct but close to the decoding region's decision boundary. To accomplish this objective, this paper utilizes a restricted version of a recent result on Importance Sampling (IS) distribution for fast performance evaluation of linear codes. The IS distribution is used over the segmented observation space and integrated with active learning. This combination allows for the iterative generation of samples from the shells whose acquisition functions, defined as the error probabilities conditioned on each shell, fall within a specific range. By intelligently sampling based on the proposed IS distribution, significant improvements are demonstrated in the performance of BCH(63,36) and BCH(63,45) codes with cycle-reduced parity-check matrices. The proposed IS-based-active Weight Belief Propagation (WBP) decoder shows improvements of up to 0.4dB in the waterfall region and up to 1.9dB in the error-floor region of the BER curve, over the conventional WBP. This approach can be easily adapted to generate efficient samples to train any other deep learning-based decoder.

Hassan Noghrei, Murad Abdullah

This paper presents a refined analysis of the block error rate (BLER) of polar codes over symmetric binary-input discrete memoryless channels under successive cancellation (SC) and successive cancellation list (SCL) decoding. A novel expression for the BLER under SC decoding is derived directly in terms of the decoder's LLRs. Building on this formulation, we propose a polar code construction algorithm optimized for SC decoding and evaluate its performance under SC and dynamic SC flip (DSCF) decoding against established SC-optimized constructions, including Gaussian approximation (GA)-based and Tal-Vardy polar codes. Furthermore, by decomposing the BLER into path loss and path selection components, we derive a novel LLR-based expression for the path loss probability, which enables an SCL-optimized polar code construction method. The proposed constructions are evaluated under SCL decoding with list sizes 2, 4, and 8, and are compared with 5G standard polar codes, GA-based designs, and Reed-Muller polar codes. Simulation results show that the proposed SC-optimized polar codes achieve up to a 0.2 dB performance gain under DSCF decoding over the AWGN channel compared to benchmark constructions, and exhibit superior performance over binary symmetric channels. For SCL-optimized polar codes, the proposed method achieves comparable or improved performance across all considered list sizes, with gains of up to 0.4 dB relative to benchmark designs.