Novel Feature Selection for Non-destructive Detection of Hardware Trojans Using Hyperspectral Scanning

Jorgensen, Erik J.; Kacmarcik, Andrew; Prvulovic, Milos; Zajić, Alenka

doi:10.1007/s41635-022-00127-7

Cited by 1 publication

(12 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Prior work has shown relatively smooth variation of the backscattered EM side-channel signal spatially across the face of an IC [16], but that is not necessarily the case across the frequency dimension as shown in Figure 3. For this reason, we break the reconstruction problem up into H separate 2D image reconstruction problems y h ≈ A h x h where H is the number of harmonics that are measured.…”

Section: A Hyperspectral Image Recovery With Reliability-weighted Sam...mentioning

confidence: 92%

“…Here we sample with replacement because it can be useful to have multiple samples of the same point when those points have high noise or large variance between ICs. Prior work has shown that the backscattering EM side channel may vary significantly between copies of the same IC [16], [37]. While some regions in the hyperspectral space may look similar across all ICs, the regions that look different can hold information that may reveal whether the IC is infected with an HT or simply varies due to a natural imperfections of the manufacturing process.…”

Section: A Hyperspectral Image Recovery With Reliability-weighted Sam...mentioning

confidence: 99%

“…The hyperspectral locations that reliably show statistically significant differences between an uninfected and infected IC may be very sparse for small Trojans. This can be as few as 0.2% of locations for small, dormant Trojans as prior work has shown [16]. If we simply capture samples uniformly at random as is the standard approach to compressed sensing problems of this type, these 369 locations are likely to go undiscovered.…”

Section: A Hyperspectral Image Recovery With Reliability-weighted Sam...mentioning

confidence: 99%

“…While these methods have been shown to be successful for relatively large or active HTs which draw significant electrical current, limited progress has been made for detecting smaller, dormant HTs with non-destructive methods. The authors of [37] and [16] develop a framework for detecting dormant HTs through backscattering EM emanations, but those methods either require manual signal probing at individual IC locations or time-intensive hyperspectral measurements of the entire IC.…”

mentioning

confidence: 99%

“…Hyperspectral imaging has received significant attention 202 as an application of compressed sensing due to the time 203 required to acquire potentially hundreds of 2D images across a range of frequencies. As noted in [16], the use of a technique called point-scanning to capture hyperspectral images can exacerbate the issue of image capture time to the point where measurements can take over four hours for a single IC. This technique entails raster-scanning a measuring device spatially in two dimensions while measuring the response at specific frequencies, one at a time, and can potentially take hours per scan depending on the resolution and dimensions of the desired scan.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Hyperspectral Image Recovery via Reliability-Weighted Compressed Sensing for Hardware Trojan Detection

et al. 2022

Self Cite

View full text Add to dashboard Cite

Hyperspectral imaging is a popular style of image acquisition which captures a multitude of two-dimensional images at different frequencies to yield more information from a scene. For imaging in certain frequency ranges, point-scanning images pixel-by-pixel is the only option for imaging and consequently can require prohibitively long measurement times. This paper introduces a non-uniform compressed sensing strategy for recovering whole hyperspectral images with much fewer samples than normally required. Our methods use the known variability of measurements across training subjects to weight random sampling toward unreliable hyperspectral points. This strategy is used to heavily reduce remote hyperspectral image acquisition time while maintaining or improving recovery performance. To test our methods, we focus on the problem of recovering hyperspectral images for non-destructively detecting malicious dormant hardware Trojans hidden in integrated circuits. These non-destructive detection techniques require remote hyperspectral measurements of the backscattering electromagnetic side channel; which can only be performed by point-scanning with existing technologies. We detect covert hardware Trojan circuit modifications with state-of-the-art performance while requiring up to ten times fewer measurements than prior methods. We compare performance of a uniform-random and our weighted-random compressed sensing strategy for the 2-dimensional discrete cosine transform bases as well as with learned dictionary bases. INDEX TERMSCompressed sensing, non-uniform sampling, hyperspectral imaging, hardware Trojan, hardware security, backscattering EM side channel. I. INTRODUCTION 19 In the past two decades, compressed sensing (CS) has evolved 20 from a novel theoretical outline to become a popular frame-21 work to recover high-dimensional images from relatively few 22 samples in a variety of problem domains. Under the assump-23 tion that the data of interest can be represented sparsely in 24 some basis, CS enables the accurate recovery of signals even 25 when sampling far less than Shannon-Nyquist theory would 26 imaging in a variety of fields such as archaeology [11], 49 geoscience [12], biomedicine [13], [14], and hardware secu-50 rity [15], [16]. HSI involves acquiring potentially hundreds 51 of two-dimensional images across a range of frequencies 52 and can require long capture times or complicated hardware. 53 Since the development of the compressed sensing frame-54 work, hyperspectral imaging (HSI) has received significant 55 attention as a target for CS due to time required to scan 56 the high-dimensional feature space in which a hyperspectral 57 image exists. Early works used coded apertures to capture 58 transformed images through random pixel masks paired with 59 assumptions of spatial smoothness to reconstruct hyperspec-60 tral images [17], [18]. These physically project incoming 61 light from a scene to a compressed space directly, which is 62 then reconstructed algorithmically. Instead of hyperspectral 63 imaging wit...

show abstract

Section: A Hyperspectral Image Recovery With Reliability-weighted Sam...mentioning

confidence: 92%

Section: A Hyperspectral Image Recovery With Reliability-weighted Sam...mentioning

confidence: 99%

Section: A Hyperspectral Image Recovery With Reliability-weighted Sam...mentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Hyperspectral Image Recovery via Reliability-Weighted Compressed Sensing for Hardware Trojan Detection

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

Novel Feature Selection for Non-destructive Detection of Hardware Trojans Using Hyperspectral Scanning

Cited by 1 publication

References 47 publications

Hyperspectral Image Recovery via Reliability-Weighted Compressed Sensing for Hardware Trojan Detection

Hyperspectral Image Recovery via Reliability-Weighted Compressed Sensing for Hardware Trojan Detection

Contact Info

Product

Resources

About