PAC Code SSCL Decoding With Special Node Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The application of special node decoding to polarization-adjusted convolutional (PAC) codes is not possible due to the convolutional code encoding, as it results in non-linear codes and non-zero outputs at special nodes, which complicates decoding processes.
Innovation Solution
A simplified successive cancellation list (SSCL) decoder is developed for PAC codes, processing special nodes at the node level by considering the impact of convolutional code encoding, with predefined processing for rate-0, repetition, rate-1, and single parity check nodes, and using least likelihood ratio scrambling for SPC nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If special node decoding is applied to PAC codes, then decoding latency is reduced, but the decoding complexity increases due to non-linear codes and non-zero outputs
Solution Approach 1:
The decoding process is segmented into two distinct parts: special node processing and general node processing. Special nodes (rate-0, repetition, rate-1, and SPC nodes) are identified and processed using simplified predefined operations, while other nodes undergo full traversal. This segmentation allows the system to reduce latency for specific node types without unnecessarily complicating the overall decoding process.
Solution Approach 2:
The patent changes the processing parameters for special nodes by using predefined output values based on node type rather than performing full tree traversal. For example, rate-0 nodes always output zero, repetition nodes output based on a single information bit, rate-1 nodes pass through channel values directly, and SPC nodes use parity check equations. This parameter change simplifies the decoding operation for special nodes while maintaining accuracy.
2Reliability
If full tree traversal is performed for all nodes, then decoding accuracy is maintained, but decoding latency increases
Solution Approach 1:
The patent extracts special nodes from the general decoding process and applies dedicated simplified processing rules to them. By identifying nodes with known characteristics (rate-0, repetition, rate-1, SPC) and removing them from the full traversal requirement, the system maintains decoding accuracy for these nodes through predefined operations while significantly reducing the time spent on their processing.
Solution Approach 2:
For special nodes, the patent applies partial action by performing only the necessary minimal processing required to achieve correct decoding results. Instead of full tree traversal for all nodes, the system performs exactly what is needed for each special node type (e.g., simple parity check for SPC nodes, direct output for rate-1 nodes), avoiding unnecessary computations while maintaining accuracy.
Data Source
AI summary
Methods and devices are provided in which a channel encoded with a polarization-adjusted convolutional (PAC) code is received. A decoded codeword is generated based at least in part on simplified successive cancellation list (SSCL) decoding performed on the channel via a decoding tree. The decoding tree includes a node that generates candidate codeword output based on predefined processing using convolutional code (CC) state input and channel vector input. A sub-tree of the node remains unprocessed.