Non-Volatile Memory Data Reconstruction Using Multidimensional Parity Arrays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Non-volatile memory systems face challenges in quickly correcting corrupted data, as existing methods like error correction codes (ECCs) are time-consuming and inefficient, especially in cases of clustered errors, and physical organization of data groups in RAID systems can introduce design restrictions that slow down memory systems.
Innovation Solution
A method using parity bits to logically organize data in multidimensional data group arrays, generating parity bit groups through logical operations, and associating them with data group rows for efficient data reconstruction, allowing for faster correction of corrupted data without the need for physical data group separation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If error correction codes (ECCs) are used to detect and reconstruct corrupted data, then data reliability is improved, but the time required for correction increases significantly
Solution Approach 1:
The patent segments data into data groups organized in a multidimensional array structure, allowing selective reconstruction of only corrupted segments rather than processing entire data sets. This segmentation enables the system to apply correction methods more efficiently by focusing computational resources only where needed, thereby reducing overall correction time while maintaining reliability.
Solution Approach 2:
The patent employs a two-tiered correction approach where a first correction method (parity bits) is applied quickly to identify and correct obvious errors, and a second correction method (ECC) is applied only when necessary. This partial application of the more computationally intensive ECC method reduces average correction time while maintaining high reliability for most cases.
2Reliability
If data are physically divided into data groups and stored separately as in RAID, then data reconstruction capability is improved, but design restrictions increase and system speed decreases
Solution Approach 1:
The patent transitions from traditional two-dimensional RAID striping to a multidimensional data group array structure. This additional dimensionality allows for more flexible data organization and reconstruction paths, enabling the system to reconstruct corrupted data through multiple possible routes (different rows, columns, or planes), thereby reducing design restrictions and improving reconstruction capability without sacrificing speed.
Solution Approach 2:
The multidimensional data group array structure serves multiple functions simultaneously: it enables data storage, provides multiple reconstruction paths, supports both parity-bit and ECC correction methods, and allows flexible mapping to physical storage locations. This multi-functionality eliminates the need for separate dedicated reconstruction mechanisms, reducing overall system complexity while maintaining robust reconstruction capability.
3Quantity of substance
If data are read slowly from non-volatile memory, then storage capacity is maintained, but the speed of firmware loading and device startup decreases
Solution Approach 1:
The patent performs preliminary organization of firmware data into the multidimensional data group array structure during the manufacturing or initialization phase. This preliminary organization pre-calculates and stores the logical relationships between data groups, so that during firmware loading, the system can quickly navigate and reconstruct data without performing complex organizational calculations in real-time, thereby accelerating firmware loading speed while maintaining full storage capacity.
Data Source
AI summary
A portion of data stored in a non-volatile memory may be found to be corrupted when it is read. Where parity data is generated from portions of data and the parity data is stored with the portions of data, the corrupted data may be reconstructed from the parity data and uncorrupted portions of data.