DRAM Training Procedure Change Detection for Attack Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Memory devices, particularly DRAM components, are vulnerable to attacks where unauthorized users modify or remove them to access secure information, and existing systems lack effective methods to detect such attacks in real-time, compromising data security.
Innovation Solution
A system that performs training procedures on DRAM components to establish reference values for channel characteristics, allowing for the detection of changes that indicate an attack by comparing current training values to stored references, enabling corrective actions like disabling features to protect secure information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If training procedures are performed to detect attacks by comparing channel characteristics, then security detection capability is improved, but device complexity increases due to additional training and comparison operations
Solution Approach 1:
The system performs training procedures during manufacturing to establish reference values for channel characteristics before the device is deployed. These reference values are stored and later used for comparison during operation to detect attacks, eliminating the need for complex real-time training and reducing operational complexity
Solution Approach 2:
The system creates copies of channel characteristic data during training procedures and stores them as reference values. These copied reference values are then used for comparison against current channel characteristics to detect attacks, simplifying the detection process while maintaining high reliability
2Reliability
If real-time monitoring of channel characteristics is implemented to detect attacks, then security is improved, but processing time increases due to continuous training and comparison operations
Solution Approach 1:
Training procedures are executed during manufacturing to pre-establish reference values for channel characteristics. This preliminary action moves the computationally intensive training process away from real-time operation, enabling fast comparison-based attack detection during device usage without real-time processing delays
Solution Approach 2:
The system performs comparison operations at specific intervals or triggered events rather than continuously monitoring channel characteristics. This periodic approach reduces processing time while maintaining effective real-time attack detection capability
3Measurement precision
If reference values are stored for comparison to detect attacks, then measurement precision is improved, but information storage requirements increase
Solution Approach 1:
The system extracts only the essential channel characteristic parameters needed for attack detection and stores them as reference values. By taking out only the critical measurement data rather than storing complete channel state information, the system achieves high measurement precision while minimizing storage requirements
Data Source
AI summary
Methods, systems, and devices for training procedure change determination to detect an attack are described. A host device may perform one or more training procedures to train aspects of a memory device (e.g., a dynamic random-access memory (DRAM) component). A training procedure may depend on a current (e.g., present, within a threshold duration) metric associated with the memory device, such as a current channel metric for a channel between the memory device and the host device. The host device, memory device, or another device, may store a set of reference values associated with a training procedure and may compare a result of a training procedure to a reference value of the set to determine whether the training procedure has changed. If the training procedure or a related value has changed, the memory device may disable one or more features of the memory device to protect against a potential attack.