Sliding-Window Polar Decoding With Asymmetric CRC Bit Placement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Polar coding techniques face challenges in optimizing the distribution of error-detecting code (EDC) bits and frozen bits within the input vector to enhance decoding performance, particularly in asymmetric concatenated Polar coding scenarios, where channel reliability varies and optimal bit position selection is complex.
Innovation Solution
The method involves producing an input vector with distinct sequences of information bits and EDC bits, where the first sequence precedes the second, and both are encoded with frozen bits, using a Polar code generator matrix to create a codeword that is transmitted over a physical channel, allowing for improved decoding through asymmetric concatenated Polar coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If EDC bits and frozen bits are randomly distributed in the input vector, then the encoding process is simple, but decoding reliability is poor
Solution Approach 1:
The patent applies local quality by assigning different types of bits (information bits, EDC bits, frozen bits) to specific positions in the input vector based on channel reliability characteristics. The encoder strategically places EDC bits in positions corresponding to less reliable synthetic channels and frozen bits in positions corresponding to more reliable synthetic channels, optimizing decoding performance for each bit type according to its specific function and reliability requirements.
Solution Approach 2:
The patent changes the parameter of bit position assignment from random to systematic based on channel reliability. By computing the reliability of each synthetic channel and dynamically assigning bits according to these reliability parameters, the system transforms the static random assignment into a dynamic parameter-driven assignment that adapts to channel conditions.
2Reliability
If conventional Polar coding is used without asymmetric concatenation, then the coding structure is simple, but decoding performance deteriorates in varying channel conditions
Solution Approach 1:
The patent implements asymmetric concatenated Polar coding by treating different segments of the input vector differently. The first sequence uses a different coding structure (with EDC bits) compared to the second sequence, creating an asymmetric coding scheme that adapts to varying channel reliability across different bit positions, thereby improving overall decoding performance in non-uniform channel conditions.
Solution Approach 2:
The patent divides the input vector into distinct sequences (first sequence with K1 information bits and EDC bits, second sequence with K2 information bits and EDC bits) and applies different encoding strategies to each segment. This segmentation allows the system to optimize each segment independently according to its specific reliability requirements and channel conditions.
3Reliability
If optimal bit position selection is performed for each bit type, then decoding reliability improves, but computational complexity increases
Solution Approach 1:
The patent performs preliminary computation of synthetic channel reliability before the actual encoding process. By pre-calculating and storing the reliability characteristics of each synthetic channel position, the system avoids repeated complex computations during encoding, reducing real-time computational complexity while maintaining optimal bit position selection for maximum decoding reliability.
Data Source
Figure 1
Figure 2
Figure 3
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.