Soft Reed-Solomon Decoder for Low-Power NVM Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Reed-Solomon (RS) decoders for non-volatile memory systems consume excessive power and area when correcting errors exceed the Hamming distance bound, limiting their efficiency in error correction.
Innovation Solution
A soft-decision RS decoder is implemented, which computes syndrome polynomials, finds error locator polynomial candidates using affine spaces and Gaussian elimination, and corrects errors by selecting the optimal candidate, thereby reducing computational complexity and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current RS decoders are used to correct errors exceeding the Hamming distance bound, then error correction capability is improved, but power consumption and area increase significantly
Solution Approach 1:
The decoder is divided into multiple functional modules including syndrome computation unit, affine space generator, basis finder, and ELP candidate generator. Each module handles a specific aspect of the decoding process, allowing for optimized resource utilization and reduced overall power consumption while maintaining enhanced error correction capability.
Solution Approach 2:
The patent employs soft-decision decoding which changes the parameter representation from hard decisions (0/1) to soft values representing confidence levels. This parameter change enables the decoder to handle errors beyond the Hamming distance bound more efficiently with reduced power consumption by utilizing reliability information from the channel.
2Reliability
If current RS decoders are used to correct errors exceeding the Hamming distance bound, then error correction capability is improved, but decoder area increases significantly
Solution Approach 1:
The decoder architecture is segmented into specialized units (syndrome computation, affine space generation, basis finding, ELP candidate generation) that can be implemented using shared computational resources. This modular segmentation reduces the overall decoder area while maintaining the capability to correct errors beyond the Hamming distance bound.
Solution Approach 2:
The decoder implements universal computational blocks that can perform multiple functions. For example, the same computational infrastructure is used for syndrome computation, affine space generation, and ELP candidate evaluation, reducing the total area required compared to having separate dedicated hardware for each function.
3Use of energy by moving object
If soft-decision decoding is implemented to reduce power consumption, then energy efficiency is improved, but computational complexity increases
Solution Approach 1:
The decoder performs preliminary computations to generate the affine space and find the basis before the actual decoding operation. These pre-computed structures are then reused during decoding, reducing the real-time computational complexity while maintaining the power-efficient soft-decision approach.
Solution Approach 2:
The patent uses computation sharing where intermediate results and computational structures are copied and reused across different stages of the decoding process. This copying approach reduces redundant computations and lowers overall computational complexity while maintaining energy efficiency.
Data Source
AI summary
A soft-decision decoding computes a first syndrome polynomial in accordance with a received word, computes a second syndrome polynomial by multiplying the first syndrome polynomial by a locator polynomial based on locations of erasures within the received word, finds a basis and private solution to an affine space of polynomials that solve key equations based on the second syndrome polynomial, determines a weak set of a locations of symbols in the received word with confidence below a certain confidence level, computes a matrix from the basis, the private solution and the weak set, determines sub-matrices in the matrix whose rank is equal to a rank of the matrix, determines error locator polynomial (ELP) candidates from the sub-matrices, the basis, and the private solution, and corrects the received word using a selected one of the ELP candidates.