LDPC Convolutional Packet Rearrangement for Burst Erasure Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current erasure correction methods, such as Reed-Solomon codes, are inadequate for handling a large number of packet erasures in applications like moving image streaming, especially when erasures occur due to fading in radio communication paths, and there is a need for improved erasure correction capabilities in packet communication systems.
Innovation Solution
The implementation of an erasure correction coding apparatus and method using a low-density parity-check convolutional code (LDPC-CC) that arranges information data according to a constraint length and coding rate, generating parity packets to enhance erasure correction capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Reed-Solomon code is used for erasure correction, then correction performance can be improved by increasing block length, but calculation amount and circuit scale increase
Solution Approach 1:
The patent changes the fundamental parameters of the error correction code from Reed-Solomon block code to LDPC convolutional code, adopting a different coding structure with sparse parity-check matrices that enables effective erasure correction without requiring large block lengths, thus avoiding the increase in circuit scale while maintaining correction performance
Solution Approach 2:
The patent substitutes the traditional Reed-Solomon code mechanism with an LDPC-CC mechanism that uses belief propagation decoding and sum-product algorithms, replacing the mathematical structure and decoding approach to achieve better erasure correction with reduced computational complexity and smaller circuit implementation
2Reliability
If Reed-Solomon code block length is increased to improve correction performance, then more packets can be corrected, but calculation amount in encoding and decoding processing increases
Solution Approach 1:
The patent changes the code structure from block code to convolutional code with memory, using a sparse parity-check matrix H that relates current and past information packets to parity packets through a structured relationship, enabling efficient encoding and decoding with fixed computational complexity regardless of packet sequence length
Solution Approach 2:
The patent introduces dynamic elements through the convolutional structure where the encoder maintains state information (memory) and the decoder uses iterative belief propagation that dynamically updates probability estimates, allowing the system to adapt to varying erasure patterns without increasing overall computational burden
3Reliability
If packets are sequentially erased over a relatively long period due to fading, then burst erasure is caused, but Reed-Solomon code cannot effectively correct such erasures
Solution Approach 1:
The patent employs a convolutional code structure with memory that dynamically adapts to sequential erasures by maintaining state information across time, allowing the decoder to leverage temporal correlations and propagate belief information through the erasure burst to recover lost packets that would be independent in block code
Solution Approach 2:
The patent ensures continuous useful action through the convolutional encoding process where each output packet depends on current and previous input packets, creating a continuous chain of dependencies that allows incremental decoding and recovery even when packets are sequentially erased over extended periods
Data Source
AI summary
A loss correction encoding device having an improved capability of loss correction using LDPC-CC includes a rearranging unit that rearranges information data contained in n information packets according to the constraint length Kmax and the encoding rate (q−1)/q of a check polynomial of the loss correction code used in a loss correction encoding unit. Specifically, the rearranging unit rearranges the information data in such a way that continuous Kmax×(q−1) pieces of information data after rearrangement are contained in different information packets. The rearranging unit distributes the information data to information blocks from n information packets, where n satisfies the formula Kmax×(q−1)≤n.