Sliding-Window Polar Decoding With Grouped CRC and Frozen Bits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Polar coding techniques face challenges in efficiently encoding and decoding information bits over noisy channels, particularly in wireless communication applications, where channel polarization and error detection are critical but existing methods struggle with optimal distribution of frozen bits and error-detecting codes.
Innovation Solution
The method involves processing information bits to produce error-detecting codes (EDCs) and generating input vectors for Polar codes using specific generator matrices, such as m-fold Kronecker product matrices, to create codewords that are transmitted or stored, and decoding these codewords using binary decision trees with thresholds to determine probable paths and discard less likely ones, ensuring reliable decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional Polar coding methods are used, then encoding and decoding can be performed, but reliability is insufficient due to suboptimal distribution of frozen bits and error-detecting codes
Solution Approach 1:
The patent segments the information bits into multiple groups and applies different freezing patterns and EDC lengths to each group. Specifically, information bits are divided into first, second, and third groups, with each group having distinct freezing bit positions and EDC configurations. This segmentation allows optimized reliability for different parts of the data stream without requiring complex global optimization.
Solution Approach 2:
The patent implements local quality by assigning different EDC lengths to different information bit groups. The first group uses a first EDC length, the second group uses a second EDC length, and the third group uses a third EDC length. This local differentiation optimizes error detection capability for each segment based on its specific requirements, improving overall reliability without uniform complexity throughout.
2Reliability
If more frozen bits and EDCs are added to improve reliability, then error detection capability increases, but decoding efficiency decreases due to increased processing complexity
Solution Approach 1:
The patent employs dynamic EDC length adjustment based on the group index. Different groups of information bits are assigned different EDC lengths, allowing the system to dynamically adapt the error detection strength according to the specific group's requirements. This dynamic approach prevents uniform over-provisioning of EDC bits, maintaining decoding efficiency while achieving necessary reliability.
Solution Approach 2:
The patent changes the parameter of EDC length across different information bit groups. By varying the EDC length parameter (first EDC length, second EDC length, third EDC length) for different groups, the system optimizes the balance between error detection capability and decoding complexity, avoiding the need for uniformly high EDC lengths that would reduce productivity.
3Reliability
If uniform EDC length is used for all information bits, then implementation is simple, but reliability is suboptimal due to inability to optimize for different data segments
Solution Approach 1:
The patent segments information bits into multiple groups (first, second, and third groups) and assigns different EDC lengths to each segment. This segmentation enables optimized error detection for each group based on its specific characteristics, achieving better overall reliability compared to uniform EDC length, while keeping the configuration manageable through systematic grouping.
Solution Approach 2:
The patent applies partial optimization by using different EDC lengths for different groups rather than uniformly excessive EDC for all bits. This partial action approach achieves sufficient error detection capability for each group without the overhead of excessive EDC bits across the entire data stream, balancing reliability and complexity.
Data Source
AI summary
Methods for encoding and decoding Polar codes are provided, together with apparatuses for performing the methods. An encoding method combines first and second sequences of information bits and CRC bits and a plurality of frozen bits into an input vector. The input vector is multiplied by a generator matrix for a Polar code to produce a concatenated codeword. A decoding method receives such a codeword and produces a decoded vector by generating successive levels of a decision tree. For a first number of levels of the decision tree, paths beyond a first maximum number of most probable paths are discarded. For a second number of levels of the decision tree, paths beyond a second maximum number of most probable paths are discarded. In some cases, the decoding method may have improved performance compared to some decoding methods for non-concatenated codewords.