Polar Decoder Column Merging for Low-Latency Parallel Decoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing polar decoders suffer from poor hardware efficiency and high memory bandwidth requirements due to limited parallelism and excessive memory usage, which restricts their ability to fully exploit parallel processing opportunities and increases latency.

Innovation Solution

A novel polar decoder architecture that processes multiple consecutive stages in parallel, decomposing the polar code graph into columns with reduced memory requirements and a simplified mechanism for propagating partial sum bits, allowing flexible adaptation to different block sizes and parameterizations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional polar decoder architecture processes stages sequentially with dedicated hardware for each XOR operation, then decoding accuracy is maintained, but hardware efficiency deteriorates and latency increases

Engineering Contradiction:
Improvedecoding speedVSAvoidhardware structure
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent merges multiple consecutive stages of the polar code graph into a single processing unit. Instead of implementing each stage separately with dedicated hardware, the decoder combines stages and processes them in parallel using a unified set of processing elements, thereby improving hardware efficiency and reducing latency while maintaining decoding accuracy

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent segments the polar code graph into columns rather than processing it stage-by-stage. By organizing the graph structure into vertical columns and processing multiple columns in parallel, the decoder exploits spatial parallelism opportunities that were previously unavailable in sequential stage-based architectures

Inventive Principle:
Principle #1Segmentation

2Productivity

If memory is used to store intermediate results at each stage interface, then data dependencies are satisfied, but memory bandwidth requirements increase and hardware efficiency decreases

Engineering Contradiction:
Improveparallel processing capabilityVSAvoidmemory usage
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent performs preliminary organization of data flow through the column-based structure, arranging computations so that intermediate results are naturally available when needed without requiring extensive memory storage. The column processing order is designed to minimize memory writes and reads by exploiting the inherent data flow patterns in polar decoding

Inventive Principle:
Principle #10Preliminary action

3Adaptability or versatility

If dedicated hardware is allocated for each processing unit in the graph, then processing speed is maintained, but hardware resource utilization deteriorates when block size varies

Engineering Contradiction:
Improveblock size flexibilityVSAvoidprocessing throughput
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent implements universal processing units that can handle multiple stages through time-multiplexing. The same hardware resources are reused across different stages and different block sizes, allowing the decoder to adapt flexibly to varying communication channel requirements while maintaining high processing throughput through efficient resource utilization

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3639374B1Low latency polar coding and decoding by merging of stages of the polar code graph
Publication Date: 2023.08.09 ACCELERCOMM LTD
  • EP3639374B1 patent drawingFigure 1
  • EP3639374B1 patent drawingFigure 2
  • EP3639374B1 patent drawingFigure 3

AI summary

A polar decoder kernal (111) is described. The polar decoder kernal (111) is configured to: receive one or more soft bits from a soft kernal encoded block (113) having a block size of N and output one or more recovered kernal information bits from a recovered kernal information block (114) having a block size of N. The polar decoder kernal (111) comprises a decomposition of a polar code graph (201, 202, 203) into an arbitrary number of columns (1701, 1702) depending on the kernal block size N.