PowerTrip: Characterizing Ransomware Through Software-Accessible Power Side Channels

Crypto-ransomware remains a major cybersecurity threat because it encrypts victim data and pressures organizations to pay for recovery. The existing defenses largely rely on software-visible indicators such as file system behavior, system calls, or process activity. These signals are useful, but they can be weakened by using different encryption algorithms or modes of operation or low-rate encryption strategies that reduce observable behavioral bursts. This paper investigates whether software-accessible power telemetry can provide an additional signal for detecting ransomware-like cryptographic activity. We monitor CPU energy through Intel’s Running Average Power Limit (RAPL) interface and GPU board power through nvidia-smi. Rather than executing live ransomware, we use controlled OpenSSL-based encryption workloads that reproduce the dominant computational phase of crypto-ransomware: sustained bulk encryption over diverse file types. This design enables safe, reproducible, and policy-compliant evaluation while isolating the power signature of repeated cryptographic computation. Across controlled, exploratory, and streaming experiments, the results show that encryption-heavy workloads produce distinguishable statistical and temporal power signatures even under concurrent benign activity. File type, algorithm-level, and mixed-activity analyses show that power traces encode workload structure and an MLP-based detector can separate ransomware-like encryption from benign tasks with useful accuracy. The reported streaming replay used the exploratory model’s 25-sample observation window (approximately 0.25 s of acquisition at 100 Hz); no optimal observation window length was established. We implemented a power telemetry detector and evaluated temporal alert policies; downstream behavioral confirmation and containment remain proposed components. An extended alert policy analysis reveals substantial false positive exposure at some settings and reduced sensitivity to low-duty-cycle encryption, motivating corroboration by independent host behavior. Operational deployment should therefore pair power-based alerts with orthogonal behavioral confirmation. Overall, software-accessible power telemetry offers a complementary endpoint signal without requiring external power probes, although runtime overhead and end-to-end containment effectiveness remain to be evaluated.

Authors

Institutions

Publication Details

Journal
Computers
Published
2026-11-16
DOI
https://doi.org/10.3390/computers15110838
Primary Topic
Advanced Malware Detection Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

PowerTrip: Characterizing Ransomware Through Software-Accessible Power Side Channels

Tarun Aditya Kusupati, Sayed Erfan Arefin
Computers
Advanced Malware Detection Techniques
article

PowerTrip: Characterizing Ransomware Through Software-Accessible Power Side Channels

Tarun Aditya Kusupati, Sayed Erfan Arefin
article en

Abstract

Crypto-ransomware remains a major cybersecurity threat because it encrypts victim data and pressures organizations to pay for recovery. The existing defenses largely rely on software-visible indicators such as file system behavior, system calls, or process activity. These signals are useful, but they can be weakened by using different encryption algorithms or modes of operation or low-rate encryption strategies that reduce observable behavioral bursts. This paper investigates whether software-accessible power telemetry can provide an additional signal for detecting ransomware-like cryptographic activity. We monitor CPU energy through Intel’s Running Average Power Limit (RAPL) interface and GPU board power through nvidia-smi. Rather than executing live ransomware, we use controlled OpenSSL-based encryption workloads that reproduce the dominant computational phase of crypto-ransomware: sustained bulk encryption over diverse file types. This design enables safe, reproducible, and policy-compliant evaluation while isolating the power signature of repeated cryptographic computation. Across controlled, exploratory, and streaming experiments, the results show that encryption-heavy workloads produce distinguishable statistical and temporal power signatures even under concurrent benign activity. File type, algorithm-level, and mixed-activity analyses show that power traces encode workload structure and an MLP-based detector can separate ransomware-like encryption from benign tasks with useful accuracy. The reported streaming replay used the exploratory model’s 25-sample observation window (approximately 0.25 s of acquisition at 100 Hz); no optimal observation window length was established. We implemented a power telemetry detector and evaluated temporal alert policies; downstream behavioral confirmation and containment remain proposed components. An extended alert policy analysis reveals substantial false positive exposure at some settings and reduced sensitivity to low-duty-cycle encryption, motivating corroboration by independent host behavior. Operational deployment should therefore pair power-based alerts with orthogonal behavioral confirmation. Overall, software-accessible power telemetry offers a complementary endpoint signal without requiring external power probes, although runtime overhead and end-to-end containment effectiveness remain to be evaluated.

ComputersVol. 15(11)
University of Dayton (US)
Advanced Malware Detection Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon