Decoupled Real-Time Program Execution State Monitoring
Abstract
The focus of this project is to localize the program execution state from off-processor-chip side-channel sensor streams that are naturally created by the program execution (last level cache -LLC- misses). The side-channel sensor streams evaluated in this project are (1) LLC miss address stream, (2) processor domain power stream, (3) DDR memory domain power stream captured through electromagnetic (EM) emission, and (4) performance monitoring unit (PMU) stream. The localization is performed at a blob level to manage the sampling and computational demands on the power side-channel. An anomaly is detected in the execution when the two consecutive detected paths cannot occur in the golden model of the program. The monitor is evaluated on a Xilinx Zynq Ultrascale XCU 106 board which contains two ARM processors and a sea of FLGA fabric. The monitor uses these trained ML models to classify the sensor stream data into a Blob/Path. The multiple streams' classifications are resolved into a single Blob/Path localization based on confidence values of each stream classification. Individual stream's classification accuracy ranges from 80-90 .
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2021
- Accession Number
- AD1148458
Entities
People
- Akhilesh Tyagi
- Henry Duwe
Organizations
- Iowa State University