Solid State Storage Device Using Dynamic State Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Solid state storage devices face read failure issues due to shifting threshold voltage distribution curves in non-volatile memory cells after repeated programming and erasing, leading to decreased read speed and accuracy, as existing state prediction methods are not always accurate, causing unnecessary read retry processes.
Innovation Solution
A solid state storage device employing a state prediction method that uses machine learning algorithms to obtain failure mode prediction functions and update prediction results dynamically, based on collected state parameters, to accurately predict and manage different failure modes, thereby reducing read failure occurrences and enhancing read speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a conventional state prediction method is used to predict future states of non-volatile memory, then the read retry process can be triggered in advance, but the prediction accuracy is insufficient leading to unnecessary read retry processes that decrease read speed
Solution Approach 1:
The patent applies dynamics by transitioning from a static prediction model to a dynamic one that adapts to changing failure modes. The system dynamically identifies failure modes based on real-time state parameters and selects corresponding prediction functions, allowing the prediction mechanism to evolve with memory degradation patterns and improve accuracy without triggering unnecessary read retries
Solution Approach 2:
The patent changes parameters by introducing failure mode identification as an additional dimension to state prediction. Instead of using a single prediction function for all states, the system adjusts the prediction approach based on identified failure modes (cycling, temperature, read disturb, program disturb, data retention), optimizing prediction accuracy for each specific degradation scenario while minimizing false positives that would slow down read operations
2Measurement precision
If the threshold voltage distribution curves shift after repeated programming and erasing, then the read accuracy decreases, but the conventional read retry process increases the time required for data retrieval
Solution Approach 1:
The patent applies preliminary action by performing state prediction and failure mode identification before actual read operations. The system proactively identifies potential read failures based on current state parameters and predicted future states, allowing it to prepare appropriate read voltage adjustments in advance. This prevents read accuracy degradation without triggering unnecessary read retry sequences that would increase read time
Solution Approach 2:
The patent implements feedback by continuously monitoring state parameters and using prediction results to adjust read operations. The system feeds back prediction information about future memory states to optimize current read voltage settings, ensuring high read accuracy while avoiding the time penalty of conventional read retry processes through intelligent, prediction-based voltage adjustment
Data Source
AI summary
A solid state storage device includes a non-volatile memory and a control circuit. The non-volatile memory includes a specified region. The control circuit is connected with the non-volatile memory, and includes a function storage circuit. A state prediction function for a first failure mode and a state prediction function for a second failure mode are stored in the function storage circuit. If the control circuit confirms that the specified region is changed from the first failure mode to the second failure mode, the control circuit predicts the specified region according to current state parameters of the specified region and the state prediction function for the second failure mode.