Skip to main content
Video s3
    Details
    Presenter(s)
    Kushal Headshot
    Display Name
    Kushal
    Affiliation
    Affiliation
    North Dakota State University
    Country
    Author(s)
    Display Name
    Kushal
    Affiliation
    Affiliation
    North Dakota State University
    Affiliation
    Affiliation
    North Dakota State University
    Display Name
    Nimish Mathure
    Affiliation
    Affiliation
    North Dakota State University
    Abstract

    Always-On Denial of Service (DoS) Trojans with power drain payload can be disastrous in systems where on-chip power resources are limited. These Trojans are designed so that they have no impact on system behavior and hence, harder to detect. A formal verification method is presented to detect sequential always-on DoS Trojans in pipelined circuits and pipelined microprocessors. Since the method is proof-based, it provides 100% accurate classification of sequential Trojan components. Another benefit of the approach is that it does not require a reference model, which is one of the requirements of many Trojan detection techniques (often a bottleneck to practical application). The efficiency and scalability of the proposed method have been evaluated on 36 benchmark circuits. The most complex of these benchmarks has as many as 135,898 gates. Detection times have been found to be very efficient and rate-of-detection was found to be 100\\%, i.e., all Trojan sequential elements were detected and all non-trojan sequential elements were classified as such.

    Slides
    • Formal Verification Approach to Detect Always-on Denial of Service Trojans in Pipelined Circuits (application/pdf)