Solid-State Memory Decoding with Multi-Read LLR Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating soft-decision information, such as log-likelihood ratios (LLRs), for solid-state memory is challenging due to varying program/erase cycles, read operations, and operating temperatures, leading to increased data errors and reduced reliability in data storage systems.
Innovation Solution
The system performs multiple read operations on solid-state memory cells to account for inter-cell interference, generating soft-decision information by analyzing multi-read output patterns and using probability calculations to determine LLRs, which are then used by a decoder to enhance data reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple read operations are performed to account for inter-cell interference, then data reliability is improved, but processing time and complexity increase
Solution Approach 1:
The patent performs multiple read operations and generates soft-decision information in advance before decoding. By preparing the soft-decision data (LLRs) beforehand through preliminary reads and probability calculations, the system reduces the time penalty during actual decoding operations, as the complex probability computations are already completed.
Solution Approach 2:
The system dynamically adjusts the number of read operations and soft-decision information generation based on memory conditions. When inter-cell interference is detected or memory degradation occurs, the system increases read operations to improve reliability, while in normal conditions it reduces operations to minimize time loss, creating a dynamic balance between reliability and speed.
2Reliability
If soft-decision information is generated through multiple read operations, then error reduction is improved, but computational complexity increases
Solution Approach 1:
The patent segments the soft-decision information generation process into distinct stages: performing multiple read operations to obtain raw data, calculating probabilities for each read outcome, and then generating LLRs from these probabilities. This segmentation allows each computational task to be optimized independently and processed in parallel where possible, reducing overall computational complexity.
Solution Approach 2:
The patent introduces probability calculations as an intermediary step between raw read operations and final soft-decision LLR generation. Instead of directly computing complex LLRs from multiple reads, the system first computes simpler probability values for each read outcome, then combines these probabilities to generate LLRs. This intermediary approach simplifies the computational path and reduces complexity.
3Measurement precision
If probability calculations are performed for each read operation, then measurement precision is improved, but processing time increases
Solution Approach 1:
The patent maintains continuous probability calculations across all read operations without interruption. Rather than performing discrete batch calculations, the system continuously updates probability values as each read operation completes, keeping the computational pipeline full and efficient. This continuous processing reduces idle time and optimizes the ratio of precision gained to time consumed.
Data Source
AI summary
Embodiments of decoding data stored in solid-state memory arrays are disclosed. In one embodiment, multiple read operations are performed while taking inter-cell interference (ICI) into account. Soft-decision information, such as log-likelihood ratios (LLRs), is determined by using known data and its corresponding multi-read output. Soft-decision information is provided to a detector. Reliability is improved and performance is increased.