LDPC Targeted Symbol Flipping With Precomputed LEH Address Lookup

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current targeted symbol flipping methods in LDPC codes suffer from long latency and complex mode switching due to the need for memory elements to request symbol addresses before each trial, resulting in inefficient hardware with a complicated state machine.

Innovation Solution

A method and apparatus utilizing a combination of LE queue and hard decision memory (LEH) to store all symbol addresses before run mode, allowing for the identification of suspicious checks and providing additional decoding trials, thereby simplifying targeted symbol flipping.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If memory elements request symbol addresses before each trial in current targeted symbol flipping, then decoding can proceed, but latency increases and hardware complexity increases

Engineering Contradiction:
Improvedecoding correctnessVSAvoidlatency
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent pre-calculates and stores all possible symbol addresses in the LEH memory structure before decoding begins. This preliminary action eliminates the need for memory elements to request addresses during each trial, thereby reducing latency while maintaining decoding correctness.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The LEH (Look-Up Table for Hard decisions) acts as an intermediary structure that stores pre-computed symbol addresses. Instead of direct requests between memory elements and address generation logic, the LEH table mediates by providing ready-made addresses, simplifying the hardware and reducing access time.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If memory elements request symbol addresses before each trial in current targeted symbol flipping, then decoding can proceed, but device complexity increases

Engineering Contradiction:
Improvedecoding correctnessVSAvoidhardware complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the address generation function from the runtime decoding process and separates it into a static pre-computed table (LEH). This extraction removes the complexity of dynamic address generation logic from the critical decoding path, simplifying the hardware architecture.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

By pre-computing and storing all symbol addresses in the LEH table before decoding, the patent eliminates the need for complex runtime address generation logic. This preliminary computation simplifies the hardware by replacing dynamic generation with static lookup.

Inventive Principle:
Principle #10Preliminary action

3Adaptability or versatility

If targeted symbol flipping uses complicated mode switching, then decoding can adapt to different scenarios, but latency increases

Engineering Contradiction:
Improvedecoding adaptabilityVSAvoidlatency
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The LEH table serves multiple functions: it stores symbol addresses for all possible scenarios, provides lookup for different trial conditions, and supports various decoding modes without requiring separate address generation logic. This multi-functionality maintains adaptability while eliminating mode switching overhead.

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

Data Source

PatentUS8977926B2Modified targeted symbol flipping for non-binary LDPC codes
Publication Date: 2015.03.10 AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LTD
  • US8977926B2 patent drawing
  • US8977926B2 patent drawing
  • US8977926B2 patent drawing

AI summary

A LDPC decoder includes a processor for targeted symbol flipping of suspicious bits in a LDPC codeword with unsatisfied checks. All combinations of check indices and variable indices are compiled and correlated into a pool of targeted symbol flipping candidates and returned along with symbol indices to a process that uses such symbol indices to identify symbols to flip in order to break a trapping set.