Storage Controller Error Prediction for Targeted Read Reclaim

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for managing non-volatile memory devices do not accurately determine the need for read reclaim operations, leading to increased latency and input/output (I/O) load due to inefficient error detection and data relocation strategies.

Innovation Solution

A storage controller with an on-chip valley search (OVS) management module, error predictor, and read reclaim module that uses machine learning to analyze OVS count data and error count data at the read voltage unit level, determining the necessity of read reclaim operations based on distribution type and error prediction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional error detection methods are used for non-volatile memory devices, then the system can detect errors, but the latency and input/output load increase due to inefficient error detection and management

Engineering Contradiction:
Improveerror detection accuracyVSAvoidlatency
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by collecting OVS count data during normal read operations and using this data to predict future error counts before actual errors occur. The error predictor module analyzes historical OVS count data to generate predictions, allowing the system to proactively identify memory regions that are likely to develop errors, thereby avoiding the need for reactive error detection that would increase latency.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The memory system performs self-service through the error predictor module, which autonomously analyzes OVS count data and generates error count predictions without requiring external intervention. The system serves itself by internally monitoring its own health metrics and making intelligent decisions about read reclaim operations, reducing the need for host controller involvement and minimizing I/O load.

Inventive Principle:
Principle #25Self-service

2Reliability

If traditional error detection methods are used for non-volatile memory devices, then the system can detect errors, but the input/output load increases due to inefficient error management

Engineering Contradiction:
Improveerror detection accuracyVSAvoidinput/output load
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system replaces traditional mechanical error detection methods with an intelligent error predictor module that uses machine learning algorithms. Instead of relying on conventional error detection mechanisms that require extensive I/O operations, the error predictor substitutes a computational approach that analyzes OVS count data patterns to predict errors, significantly reducing the I/O load while maintaining or improving detection accuracy.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system changes parameters by transitioning from direct error detection through I/O operations to indirect error prediction through analysis of OVS count data. This parameter change allows the system to infer error likelihood from historical reading patterns without performing additional read operations, thereby reducing I/O load while maintaining reliable error detection capability.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If read reclaim operations are performed frequently to ensure data reliability, then data reliability improves, but latency and I/O load increase

Engineering Contradiction:
Improvedata reliabilityVSAvoidlatency
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system implements feedback through the error predictor module, which continuously monitors OVS count data and uses this feedback to make intelligent decisions about when read reclaim operations are necessary. The feedback mechanism allows the system to adjust read reclaim frequency based on actual error predictions rather than using fixed schedules, ensuring data reliability while minimizing unnecessary operations that would increase latency.

Inventive Principle:
Principle #23Feedback

4Reliability

If read reclaim operations are performed frequently to ensure data reliability, then data reliability improves, but I/O load increases

Engineering Contradiction:
Improvedata reliabilityVSAvoidI/O load
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system replaces mechanical read reclaim operations with an intelligent prediction-based approach. The error predictor module substitutes direct I/O-intensive error detection and correction with a computational analysis of OVS count data, allowing the system to determine when read reclaim is actually needed. This substitution significantly reduces I/O load while maintaining data reliability through targeted rather than frequent operations.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4160603B1Method of operating a storage controller determining error counts
Publication Date: 2026.04.22 SAMSUNG ELECTRONICS CO LTD
  • EP4160603B1 patent drawingFigure 1
  • EP4160603B1 patent drawingFigure 2
  • EP4160603B1 patent drawingFigure 3

AI summary

Disclosed is a method of operating a storage controller which communicates with a non-volatile memory device. The method includes outputting a first command including a request for on-chip valley search (OVS) count data of a memory region of the non-volatile memory device to the non-volatile memory device, wherein the OVS count data include a first count value of a first read voltage and a second count value of a second read voltage, receiving the OVS count data from the non-volatile memory device, determining a first error count value for the first read voltage and a second error count value for the second read voltage, based on the OVS count data, and determining a subsequent operation, based on the first and second error count values.