On-Chip Intelligent Sensor Network for Real-Time Anomaly Detection
Abstract
With the advent of the Internet of Things (IoT) and cyber physical systems (CPS), malware, ransomware, distributed denial-of-service (DDoS), etc. have become significant concerns. Meanwhile, the integration of intellectual property (IP) from third party vendors and outsourcing of chip fabrication in advanced technology nodes have made it difficult to protect critical systems from hardware (HW) sabotage and hardware Trojans (HTs). These threats can change a system#s intended functionality, lower its reliability, or leak sensitive information. Conventional software (SW)-based approaches for real-time malware detection consume significant memory and processing resources while side-channel analysis requires equipment so large and expensive it is unsuitable outside of the laboratory and in real-time settings. HT detection methods such as reverse engineering, pre-silicon verification, and post-manufacturing tests are expensive, unscalable, or ineffective. Side-channel analysis, while more promising, is typically performed off-chip where the signal-to-noise ratio (SNR) is low and HTs can hide within process variations.The proposed project seeks to build upon the success of existing side-channel anomaly detection approaches while overcoming their limitations. Specifically, we propose measuring power and EMside-channels on the chip itself and performing anomaly detection at run-time. By combining the complementary data from traditionalEM, EM backscattering, and power modalities, anomaly detection rates can be improved while also resisting EM- and power-specific countermeasures. Run-time anomaly detection complements pre-silicon and post-manufacturing approaches by finding those malwares and HTs that evaded them. At run time, once malware and HTs are triggered, their impacts on side-channels increase making them easier to detect. Bringing these capabilities onto the chip shall improve time and spatial resolution, allowing one to not only detect but alsolocalize malware and HTs. Localization at run time creates new opportunities to bypass threats rather than discarding the entire infected system. Using open-source IPs and SoC designs as our experimental platforms along with sample malware and HTs, we will aim to1) Investigate the anomaly detection and localization capabilities of these modalities measured through an on-chip programmable sensor array (PSA) and an low dropout regulator (LDO)-based power distribution network; 2) Quantify the information relevance and redundancy of each modality, and combine all three modalities to achieve superior detection rates and mean time to detection; 3) Design, optimize, and implement the on-chip sensor network within an SoC; and 4) Demonstrate the effectiveness and efficiency of the proposed real-time sensor framework by fabricating and testing a 65nm chip as proof-of-concept.This project will support the Navy#s AppliedCyber Resiliency program for advancing fundamental technology to ruggedize naval data and SW systems to achieve mission objectives in the face of adversarial cyber interference. It will do so by facilitating a better understanding of the true threat posed by malware and HTs and, consequently, the inadequacy of current detection methods. Further, the proposed on-chip sensors shall support the deployment of lifelong HW and SW protection schemes in a broad range of systems. Practical methods will be generated whose utility can extend to system designers, IP vendors, and end-users in the Navy to fulfil and ensure mission dependability. Finally, considering that almost all Naval systems rely on integrated computing systems, the research outcomes of this project will have ultimate importance in ensuring the trustworthiness of the fundamental hardware platforms within these systems.This abstract is approved for public release
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 15, 2023
- Source ID
- N000142412048
Entities
People
- Domenic Forte
Organizations
- Office of Naval Research
- United States Navy
- University of Florida