Pre-configured Error Pattern Ordered Statistics Decoding for CRC-Polar Codes

Xuanyu Li,Kai Niu,Yuxin Han,Jincheng Dai,Zhiyuan Tan,Zhiheng Guo

2023-09-23

Abstract:In this paper, we propose a pre-configured error pattern ordered statistics decoding (PEPOSD) algorithm and discuss its application to short cyclic redundancy check (CRC)-polar codes. Unlike the traditional OSD that changes the most reliable independent symbols, we regard the decoding process as testing the error patterns, like guessing random additive noise decoding (GRAND). Also, the pre-configurator referred from ordered reliability bits (ORB) GRAND can better control the range and testing order of EPs. Offline-online structure can accelerate the decoding process. Additionally, we also introduce two orders to optimize the search order for testing EPs. Compared with CRC-aided OSD and list decoding, PEPOSD can achieve a better trade-off between accuracy and complexity.

Information Theory

What problem does this paper attempt to address?

The paper attempts to address the problem of how to improve the decoding performance of short block codes in Ultra-Reliable Low-Latency Communication (URLLC), especially when using Cyclic Redundancy Check (CRC) aided Polar Codes. Existing decoding methods such as CRC-aided Successive Cancellation List (CA-SCL) decoding, although effective, still have room for improvement in terms of complexity and performance. To this end, the authors propose a new decoding scheme—Pre-configured Error Pattern Ordered Statistics Decoding (PEPOSD), which aims to optimize the decoding process by pre-generating a large number of error patterns (EPs), thereby reducing decoding complexity while ensuring decoding accuracy.

Specifically, the main innovations of PEPOSD include:

1. **Decoding Process**: Unlike traditional Ordered Statistics Decoding (OSD) methods, PEPOSD does not focus on querying the most reliable independent symbols but instead uses a large number of pre-configured error patterns to test the transformed information bits. These error patterns can be continuously read and tested on hard decision bits to determine if they can correct errors in the information bits. After filtering out valid candidate codewords through CRC checks, the most likely result is selected through Euclidean distance competition. Due to the characteristics of CRC-Polar codes, this method can terminate decoding early by setting a maximum number of valid codewords, thus achieving lower complexity.

2. **Error Pattern Pre-configuration Process**: Error patterns can be pre-configured once for all types of codes (different lengths or rates) to achieve higher decoding speed, or they can be dynamically generated before decoding to save hardware resources. To further optimize the testing order using soft information, the paper introduces two sorting methods: Index Weight (IW) & Hamming Weight (HW) sorting and Priority Weight (PW) sorting. IW reflects the error probability of a specific error pattern, while PW combines IW and HW, designing an efficient way to utilize possible error patterns.

Through the above methods, PEPOSD not only improves the decoding performance of short CRC-Polar codes but also significantly reduces decoding complexity, demonstrating potential in practical applications.