Design of security primitives for trustworthy integrated circuits
Date of Issue2015
School of Electrical and Electronic Engineering
Centre for Integrated Circuits and Systems
The horizontal business model in semiconductor industry has brought economic benefits but relinquished the control that integrated circuit (IC) designer had in chip design and manufacturing, thus making chips expose to greater security threats and vulnerable to various kinds of hardware attacks. As electronic devices becoming increasingly ubiquitous and permeated into our daily lives, critical basic infrastructures and national defense systems, hardware security features that can assure the trustworthiness of an integrated system in a reliable, efficient and inexpensive way are highly desirable. In the dissertation, two embedded hardware security primitives, namely active current sensors and Physical Unclonable Functions (PUFs) are investigated for hardware Trojan (HT) detection, device identification and authentication. Hardware Trojans are the deliberate and malicious alterations to original IC designs that can jeopardize the design integrity, stealing the confidential information or paralyzing the system connected to the subverted chip upon their activation. In this thesis, a transient power supply current sensor to facilitate the screening of an IC for HT infection is proposed. Based on the power gating scheme, it converts the current activity on local power grid into a timing pulse from which the timing and power related side channel signals can be externally monitored by the existing scan test architecture. Its current comparator threshold can be adjusted for calibration against the quiescent current noise floor to reduce the impacts of process variations. Post-layout statistical simulations of process variations are performed on the ISCAS’85 benchmark circuits to demonstrate the effectiveness of the proposed technique for the detection of delay-invariant and rarely switched i HTs. Compared with the detection error rate of a 4-bit counter based HT reported by an existing HT detection method using the path delay fingerprint, the proposed method shows an order of magnitude improvement in the detection accuracy. Physical Unclonable Functions are emerging security primitives that are useful in secure key generation, device authentication, and counterfeit detection and prevention. This thesis presents two main contributions in PUF research community: 1) An ultra-low power and small footprint hybrid RO PUF with very high temperature stability is proposed as an ideal candidate for lightweight applications. The classic ring oscillator (RO) based PUF is resilient to noise impacts, but its response is susceptible to temperature variations. Additional components or complex algorithms are usually employed to address this problem at the expense of large area and power consumption overheads. The proposed PUF exploits the negative temperature coefficient property and the low-power subthreshold operation of current starved inverters to mitigate the variations of differential RO frequencies with temperature. The new architecture uses conspicuously simplified circuitries to generate and compare a large number of pairs of RO frequencies, and facilitate logical reconfigurability to thwart machine learning attacks. The proposed 9-stage hybrid RO PUF was fabricated using 65 nm CMOS technology. Its measured challenge-response pairs (CRPs) possess larger entropy per unit area than the classic RO PUF design. The PUF occupies only 250 m2 of chip area and consumes only 32.3 V per CRP at 1.2 V and 230 MHz. The measured average and worst-case reliabilities of its responses are as high as 99.84% and 97.28%, respectively over a wide range of temperature from −40 ◦C to 120 ◦C. 2). Another new low-cost CMOS image sensor based PUF is also proposed. It targets a variety of security, privacy and trusted protocols that involve image sensor as a trusted entity. The proposed PUF exploits the intrinsic imperfection during the image sensor manufacturing process to generate a rich set of unique and stable digital signatures. The proposed differential readout algorithm stabilizes the response bits extracted from the random fixed pattern noise (FPN) of selected pixel pairs determined by the applied challenge against supply voltage and temperature variations. The threshold of difference can be tightened to winnow out more unstable response bits from the huge challenge-response space offered by modern image sensors to enhance the reliability under harsher operating conditions or it can also be loosened to improve its resiliency against masquerade attacks in routine operating environment. Experimentations conducted on a 64×64 image sensor fabricated in 180 nm 3.3 V CMOS technology demonstrated that robust and reliable challengeresponse pairs can be generated with a uniqueness of 49.37% and a reliability of 99.10% under temperature variations of 15∼115 ◦C and supply voltage variations of 3∼3.6 V.
DRNTU::Engineering::Electrical and electronic engineering::Electronic circuits