On-Die Power Supply Droop Detector for Security Threat Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hackers can attack processors or system-on-chip (SoC) by probing the power supply rail, measuring the supply droop profile, and extracting cryptographic keys or security keys, posing a threat to secure data and system integrity.
Innovation Solution
Implementing a hardware-based unsupervised machine-learning approach that uses an on-die power supply droop detector as a feature extractor, in combination with a deep neural network (DNN) for feature classification, to identify and mitigate security threats by classifying normal operation, aged device behavior, and security threats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hackers probe the power supply rail to measure supply droop profile, then they can extract cryptographic keys or security keys, but this compromises system security and data protection
Solution Approach 1:
The patent converts the harmful probing action into a beneficial security feature by detecting the very act of probing. The power supply droop detector monitors for abnormal droop profiles that indicate probing attempts, and the machine learning classifier distinguishes between normal variations and attack patterns. This transforms the hacker's measurement capability into a detection mechanism that protects cryptographic keys and security data.
2Ease of operation
If traditional power supply monitoring is used, then basic voltage levels are tracked, but it cannot distinguish between normal operation, aged device behavior, and security threats
Solution Approach 1:
The patent introduces machine learning logic as an intermediary between the power supply droop detector and the security system. This intermediary layer processes the raw droop measurements and applies trained classification models to distinguish between normal operation, aged device behavior, and security threats. The machine learning component acts as a mediator that interprets complex power supply patterns and translates them into actionable security decisions.
3Reliability
If complex machine learning models are deployed for threat detection, then classification accuracy improves, but hardware complexity and power consumption increase
Solution Approach 1:
The patent segments the threat detection system into distinct functional modules: a power supply droop detector for feature extraction, machine learning logic for classification, and a response mechanism for mitigation. This segmentation allows each component to be optimized independently, enabling the deployment of sophisticated machine learning models while managing hardware complexity through modular architecture.
Data Source
AI summary
Hardware based unsupervised based machine-learning (ML) approach to identify a security threat to the processor (e.g., caused by probing of a power supply rail). An apparatus is provided which includes an on-die power supply droop detector as a feature extractor. The droop detector detects a droop in the power supply caused by a probe physically coupling to the power supply rail. The droop detector in combination with machine-learning logic detects change in power supply rail impedance profile due to a probe coupled to the power supply rail. A deep-neural network (DNN) is provided for feature classification that classifies a security threat from normal operation and from operations caused by aging of devices in the processor. The DNN is trained in a training phase or production phase of the processor. An aging sensor is used to distinguish classification of aged data vs. normal data and data from security attack.