Parallel Convolutional Encoding with Block Partitioning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional convolutional encoders face performance bottlenecks in high-speed applications due to linear encoding complexity, which becomes impractical for long packets as the logic complexity and critical path increase significantly with look-ahead techniques.

Innovation Solution

Partitioning input information bits into smaller blocks for parallel convolutional encoding, where each block is encoded independently with different initial states, and selecting the appropriate output based on the final encode state of the previous block, allowing for increased throughput by encoding in parallel.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If look-ahead techniques are applied to increase encoding throughput, then encoding speed is improved, but logic complexity and critical path increase significantly

Engineering Contradiction:
Improveencoding throughputVSAvoidlogic complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The input bit sequence is divided into multiple sub-blocks that can be processed in parallel. Each sub-block is encoded independently using separate encoding paths, allowing the encoder to process multiple segments simultaneously rather than sequentially, thus increasing throughput without requiring a single complex look-ahead unit

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from temporal parallelism (look-ahead processing within a single encoding path) to spatial parallelism (multiple independent encoding paths processing different sub-blocks simultaneously). This dimensional shift allows throughput improvement while keeping individual path complexity manageable

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Device complexity

If conventional serial encoding is used, then logic complexity is kept low, but encoding throughput becomes a performance bottleneck for high-speed applications

Engineering Contradiction:
Improvelogic complexityVSAvoidencoding throughput
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The input bit sequence is divided into multiple sub-blocks that can be processed in parallel. Each sub-block is encoded independently using separate encoding paths, allowing the encoder to process multiple segments simultaneously rather than sequentially, thus increasing throughput without requiring a single complex look-ahead unit

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Multiple simple encoding paths are merged into a single parallel processing architecture. Each path uses straightforward encoding logic similar to conventional serial encoders, but they operate simultaneously on different sub-blocks, combining their outputs to achieve high throughput while maintaining low individual path complexity

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP2215723B1Convolutional encoding with partitioned parallel encoding operations
Publication Date: 2019.04.24 QUALCOMM INC
  • EP2215723B1 patent drawingFigure 1~2
  • EP2215723B1 patent drawingFigure 3
  • EP2215723B1 patent drawingFigure 4

AI summary

Convolutional encoding throughput is increased by partitioning input information bits into a plurality of blocks that are convolutionally encoded in parallel. A plurality of convolutional encoding operations which have respective initial encode states that are mutually different from one another are applied in parallel to one of the blocks to produce a respectively corresponding plurality of convolutional encoding results. One of the convolutional encoding results is selected based on a convolutional encoding operation applied to another of the blocks.