Embodiments generally relate to countermeasures against leakage power analysis (LPA) attacks. More specifically, embodiments relate to sinking redundant current by a discharging resistor to alter the signature of load power dissipation of a cryptographic circuit.
The power consumption of CMOS-based cryptographic circuits depends strongly on the data being processed by the circuits. A correlation between the power consumed by the circuit and the data being processed by the circuit can be exploited to obtain the stored critical information by a malicious attacker utilizing a side-channel attack (SCAs). Differential power analysis (DPA) attacks are one of the most widely studied SCAs that exploit the switching activities within the cryptographic circuits while the circuits process different input data. Recently, leakage power analysis (LPA) attacks have been disclosed that may be utilized for obtaining the critical information by analyzing the correlation between the input data and leakage power dissipation of the cryptographic circuit. LPA attacks exploit the fact that the leakage current signature of NMOS and PMOS transistors is different than dynamic power, where the amplitude of leakage power is orders of magnitude smaller than the amplitude of dynamic power consumption. To perform a successful LPA attack, the attacker must mitigate the measurement noise that can make the analysis quite difficult due to the small signal-to-noise ratio (SNR) of the monitored leakage power. An effective technique to mitigate the measurement noise is to lower the operating frequency of the cryptographic circuit.
Since the leakage mechanisms in DPA and LPA attacks are quite different, DPA-resistant cryptographic circuits may still be vulnerable against LPA attacks. Therefore, there is a strong need for effective countermeasures against LPA attacks.
A voltage converter with adaptive security features is proposed as a lightweight countermeasure against leakage power analysis (LPA) attacks. When an LPA attack is sensed by the proposed security-adaptive (SA) voltage converter, a discharging resistor starts sinking redundant current to alter the signature of the load power dissipation. The power dissipation induced by the discharging resistor is scrambled by the SA voltage converter to maximize the amount of the inserted noise to the input power profile of the cryptographic against LPA attacks. As compared with a conventional cryptographic circuit that does not house any countermeasure; the lowest measurement-to-disclose value of a cryptographic circuit that employs the proposed voltage converter can be enhanced over 6145 times against LPA attacks.
Countermeasure against leakage power analysis (LPA) attacks. Since this attack is quite new, there is only a few existing countermeasures in the literature. The proposed countermeasures are one the very first countermeasures against LPA attacks. Also, this is an adaptive countermeasure which is quite new in the field.
In some embodiments, a system is provided for security adaptive voltage conversion. The system includes a voltage converter that has an input coupled to a power source and an output coupled to a cryptographic circuit. The voltage converter receives input power from the power source via the input and supplies power to the cryptographic circuit via the output. A redundant discharge load is connectable to the output of the voltage converter by a switch that is operable to connect or disconnect the redundant discharge load to the output of the voltage converter. A clock frequency sensor coupled to the cryptographic circuit that detects a clock frequency of the cryptographic circuit and controls the connection or the disconnection of the redundant discharge load to the output of the voltage converter in response the detected clock frequency of the cryptographic circuit.
In some embodiments, a method is provided for security adaptive voltage conversion. The method includes receiving input power from a power source via an input of a voltage converter and supplying power to a cryptographic circuit via an output of the voltage converter. A clock frequency sensor coupled to the cryptographic circuit detects a clock frequency of the cryptographic circuit and controls a connection or a disconnection of a redundant discharge load to the output of the voltage converter in response the detected clock frequency of the cryptographic circuit. The redundant discharge load is connectable to the output of the voltage converter by a switch that is operable to connect or disconnect the redundant discharge load to the output of the voltage converter.
Other aspects of the invention will become apparent by consideration of the detailed description and accompanying drawings.
Before any embodiments of the invention are explained in detail, it is to be understood that the invention is not limited in its application to the details of construction and the arrangement of components set forth in the following description or illustrated in the following drawings. The invention is capable of other embodiments and of being practiced or of being carried out in various ways.
It should also be noted that a plurality of hardware and software based devices, as well as a plurality of different structural components, may be used to implement various embodiments described herein. In addition, it should be understood that embodiments may include hardware, software, and electronic components or modules that, for purposes of discussion, may be illustrated and described as if the majority of the components were implemented solely in hardware. However, one of ordinary skill in the art, and based on a reading of this detailed description, would recognize that, in at least one embodiment, the electronic based aspects may be implemented in software (e.g., stored on non-transitory computer-readable medium) executable by one or more processors. As such, it should be noted that a plurality of hardware and software based devices, as well as a plurality of different structural components may be utilized to implement various embodiments. Furthermore, and as described in subsequent paragraphs, the specific configurations illustrated in the drawings are intended to exemplify embodiments and that other alternative configurations are possible. For example, “controllers” described in the specification can include standard processing components, such as one or more processors, one or more computer-readable medium modules, one or more input/output interfaces, and various connections (e.g., a system bus) connecting the components. In some instances, the controllers described in the specification may be implemented in one of or a combination of a general processor, an application specific integrated circuit (ASIC), a digital signal processor (DSP), a field programmable gate array (FPGA), combinational logic or state circuitry, or the like.
Converter-reshuffling (CoRe) technique is a countermeasure against DPA attacks with low overhead. CoRe technique utilizes a multi-phase switched capacitor (SC) voltage converter where each phase delivers a portion of the required power to the cryptographic circuit with a different time delay. A pseudo-random number generator (PRNG) is used to scramble the sequence of activation of the converter phases to insert a varying amount of uncertain power noise in each switching period against DPA attacks. However, if the attacker implements a LPA attack on a cryptographic circuit with a CoRe voltage converter, the low leakage power dissipation generated by the cryptographic circuit would only activate a small number of converter phases. The small number of active phases would significantly reduce the entropy of the PRNG in the CoRe voltage converter, making the CoRe technique also vulnerable against LPA attacks.
To increase security against LPA attacks with negligible overhead, a voltage regulator is designed in a security-adaptive fashion. The security-adaptive (SA) voltage converter is designed based on the CoRe voltage converter but modified to sense LPA attacks and insert noise through a discharging resistor only when the device is under an LPA attack. When the SA voltage converter is utilized as the supply voltage of the cryptographic circuit, during the normal and idle modes of operation, no redundant current is being consumed and the SA voltage converter operates conventionally as the CoRe voltage converter. The SA voltage converter is triggered to provide redundant current when the operating clock frequency fc is within a certain range. The activity of the discharging resistor is then reshuffled by the PRNG to scramble the inserted noise profile. Since the proposed SA converter operates conventionally and is only triggered to sink redundant current when the device is under an LPA attack, the power overhead of this countermeasure is negligible.
The SC converter 120 is operable to function as a CoRe voltage converter and may be referred to as a CoRe voltage converter. The SC converter 120 includes an N-bit pseudorandom number generator (PRNG) 122 and N switched capacitor stages 124. Each of the N switched capacitor stages 124 corresponds to 1 of N phases of the SC converter 120 and may be referred to as a switched capacitor phase. Therefore, the SC converter 120 may be referred to as an N stage SC converter, an N phase SC converter, or a multi-phase SC converter. N is the total number of phases and the total number of stages, and N is even. The SC converter 120 is coupled to the power source 110 to receive power from the power source 110. The SC converter 120 is coupled to the cryptographic circuit to deliver power to the cryptographic circuit 112.
The SC converter 124 is connected to and receives power from the power supply 110 (Vin), and outputs power to the cryptographic circuit 112 (Vout). The output power is delivered via a random number of the N phases of the SC converter 120 that are selected by the N-bit PRNG 122. An amount of power drawn from the power supply 110 depends on a demand for power consumption by the cryptographic circuit 112 and by the discharging resistor 140 (RC) when the discharging resistor is coupled to the output of the SC converter 120. The discharging resistor 140 can introduce a second (or redundant) load relative to the load of the cryptographic circuit 112. The discharging resistor 140 may be referred to as a redundant discharging load.
The N-bit PRNG 120 generates a random pattern of output signals for gating the SC converter stages 124 such that an appropriate number of stages 124 are charged and discharged for meeting a demand for power by the cryptographic circuit 112 and by the discharging resistor 140 (RC) when the discharging resistor 140 is coupled to the output of the SC converter 120. Each signal in the random pattern of outputs from the N-bit PRNG 122 is sent to a respective SC converter stage 124. In some embodiments, the N-bit PRNG 122 includes a controller or may be coupled to a controller. The controller may include an electronic processor and a memory that stores instructions, such that when the instructions are executed by the electronic processor, the instructions cause the controller to perform the functions described with respect to the SC converter 120. The controller includes suitable logic, circuitry, interfaces, and/or or code that are operable to determine how many stages of the SC converter 124 need to be activated (charged and/or discharged) to meet the demand for power by the cryptographic circuit 112 and by the discharging resistor 140 (RC) when the discharging resistor 140 is coupled to the SC converter 120.
The cryptographic circuit 112 receives a clock input that determines an operational frequency fc of the cryptographic circuit 112. The cryptographic circuit 112 may utilize a secret key (also referred to as a correct key) to encrypt or decrypt data, such as a password or other data. An attacker may observe power input to the SC converter 124 (Vin) to determine data (such as the secret key) and/or operations performed by the cryptographic circuit 112.
The first clock frequency sensor 130 and the second clock frequency sensor 134 also receive the clock input at the operational frequency fc that is provided to the cryptographic circuit 112. The first clock frequency sensor 130 includes a frequency comparator 132 that compares the operational frequency fc to a first frequency threshold Fc,1 and controls a switch M1 in response to the results of the comparison. The second clock frequency sensor 130 includes a second frequency comparator 134 that compares the operational frequency fc to a second frequency threshold Fc and controls a switch M2 in response to the results of the comparison. In some embodiments the switches M1 and M2 are be transistors. When the switches M1 and M2 are both closed/on (M1=1, M2=1) the discharging resistor 140 (RC) is coupled to the output (Vout) of the SC converter 120. When the discharging resistor 140 (RC) is coupled to the output (Vout) of the SC converter 120 the demand for power from the SC converter 120 is increased above the demand for power from the cryptographic circuit 112, such that power input to the SC converter 120 (Vin) from the power supply 110 also increases. In some embodiments, the first clock frequency sensor 130 and/or the second clock frequency sensor 134 includes a controller that controls the opening and closing of the switches M1 and M2 respectively.
The SC converter 120 is configured to implement a converter reshuffling (CoRe) technique where each of the N phases 124 delivers a portion of the power delivered to the cryptographic circuit 112 with a different time delay. The PRNG 122 functions to scramble the sequence of activation of the SC converter phases 124 to insert a varying amount of uncertain power noise during each switching period of the SC converter 120 and serve as a countermeasure against DPA attacks.
When the cryptographic circuit 112 is in a normal working mode, the cryptographic circuit 112 exhibits a high dynamic power consumption (i.e., the clock frequency fC is high), M1 transistor is in the off-state to let the SA voltage converter system 100 operate similar to the CoRe voltage converter. Under an LPA attack, however, the attacker would lower the clock frequency fC to mitigate measurement noise. If the clock frequency fC is lower than the critical frequency threshold FC,1 and higher than the critical frequency threshold FC, both of the M1 transistor and the M2 transistor would be in on-state, letting some amount of redundant current flow through the discharging resistor 140 (RC). The redundant power dissipation induced by RC is then reshuffled by the N-phase CoRe converter to scramble the inserted power noise.
When the clock frequency fC of the cryptographic circuit 112 is lower than the critical frequency threshold FC, the M2 transistor would be turned-off, deactivating the discharging resistor 140 (RC) as shown in
Referring to
In step 215, the cryptographic circuit 112 is in becomes active and operates within a normal operating frequency range (Fc<fc<FC,1). The discharge resistor 140 is decoupled from the output of the SC converter 120 and the SA voltage converter system 100 functions as a converter reshuffling system while providing power to the cryptographic circuit 112. For example, the first clock frequency sensor 130 detects that the operating frequency is above the high frequency threshold (fc≥FC,1), and sets the transistor switch M1 in an off-state (M1=0). The second clock frequency sensor 132 detects that the operating frequency is above the low frequency threshold and sets the second switch M2 to an on-state (M2=1).
In step 220, the cryptographic circuit 112 becomes under attack and the operating frequency is slowed down. The switches M1 and M2 are both set to an on-state to couple the discharging resistor 140 to the output of the SC converter 120. With the discharge resistor 140 engaged, noise is inserted to the SA converter system 100 input at Vin as the discharging resistor 140 (RC) sinks current from the SC voltage converter 120 output Vout. The power dissipation induced by the discharging resistor 140 is scrambled by the SC voltage converter 120 to increase the amount of the noise inserted into the input power profile of the SA voltage converter system 100 while the system 100 provides power to the cryptographic circuit 112. For example, the first clock frequency sensor 130 detects that the operating frequency is below the high frequency threshold and sets the first switch M2 to an on-state (M2=1). The second clock frequency sensor 132 detects that the operating frequency is above the low frequency threshold and sets the second switch M2 to an on-state (M2=1).
When the SC converter 120 is utilized as the supply voltage for the cryptographic circuit 112 during the normal and idle modes of operation, no redundant current is being consumed by the discharging resistor 140 (RC) and the SC converter 120 operates conventionally as a CoRe voltage converter. The SA voltage converter system 100 is triggered to provide redundant current to the discharging resistor 140 (RC) when the operating clock frequency fC is within a range above an idle frequency threshold Fc and below a normal operation frequency threshold FC,1. The activity of the discharging resistor 140 (RC) is then reshuffled by the PRNG 122 to scramble the inserted noise profile at the input to the SC converter 120 (Vin). Since the SA converter system 100 operates conventionally and is only triggered to sink redundant current by the discharging resistor 140 (RC) when the cryptographic device 112 is under an LPA attack, the power overhead of this countermeasure is negligible. Therefore, the system 100 is designed in a security-adaptive fashion.
Parameter Design
To maximize the entropy of the N-bit PRNG that resides within the SA voltage converter, the number of active phases of an SA voltage converter in each switching period should be around N/2 (the entropy of the N-bit PRNG reaches the maximum value
Assuming the mean value of leakage power dissipation of the cryptographic circuit within a switching period under LPA attacks is μc and the output voltage of an N-phase CoRe converter within the SA voltage converter is Vout. When the cryptographic circuit employs an SA voltage converter, if the discharging resistor RC is activated, the power dissipation PC consumed by the discharging resistor RC can be denoted as Pc=Vout2/RC. The mean value μt of the total load power dissipation of the SA voltage converter within a switching period can be approximated as
The output current Iout delivered by a single SC converter phase is
Iout=2Cf(Vin−2Vout)kfs, (2)
where Cf is the flying capacitance within each phase, Vin is the input voltage from the power source, fs is the switching frequency of the SC converter, and k is the fs and Cf dependent parameter.
Since around half of the total phases should be active in each switching period to maximize the entropy of the N-bit PRNG, the following approximated equation should be satisfied
where R′c is the optimized resistance value of the discharging resistor RC that maximizes the security of the cryptographic circuit. R′C therefore, can be determined as
Security Evaluation Against LPA Attacks
To quantify the security of a cryptographic circuit that employs the proposed SA voltage converter against LPA attacks, the correlation coefficient between the input and load power profiles of the SA voltage converter needs to be modeled. The correlation coefficient γ of a voltage converter is
where n is the total number of the input or load power data samples, Pl,i (Pin,i) is the ith, (i=1, 2, . . . , n) load (input) power of the voltage converter, and
A. Sampling a Single Clock Period as One Sample of Input Power Data
In LPA attacks, in order to filter the measurement noise, the clock frequency fc of the cryptographic circuit needs to be sufficiently reduced
where F0 is an integer that can reasonably filter out the measurement noise). However, when a cryptographic circuit implemented with a CoRe or an SA voltage converter is under LPA attacks, in addition to filtering the measurement noise, the reshuffling noise induced by PRNG can also be filtered if the clock frequency fc is further reduced. For example, the clock frequency fc can be further reduced to
(F is an integer and F>F0) to also filter the reshuffling noise.
If the attacker selects a single clock period (F number of switching periods) as one sample of the input power data as shown in
where nc is the power efficiency of the N-phase CoRe converter in the SA voltage converter, P0 is the power consumed by a single active phase in the SA voltage converter, and Pi is the leakage power dissipation of the cryptographic circuit induced by the ith input data. HY
As compared to a conventional cryptographic circuit (i.e., without any countermeasure), the MTD enhancement ratio R(FTs) of a cryptographic circuit that employs a voltage converter is
is the average correlation coefficient between the input and output power profiles of the voltage converter.
As compared to an LPA attack on a conventional cryptographic circuit with clock frequency
the MTD value would be enhanced by F/F0 times if the attacker implements an LPA attack on a cryptographic circuit which employs a voltage converter with a slower clock frequency
As a result, the MTD enhancement ratio R1(FTs) of a cryptographic circuit that employs a voltage converter with a variable clock frequency can be written as
Advanced encryption standard (AES) utilizes multiple S-Boxes to perform non-linear mathematical transformations to mask the relationship between the ciphertext and the secret key. To validate the mathematical analysis, a 130 nm CMOS Sbox is used as the cryptographic circuit that is powered, respectively, by a CoRe voltage converter and by an SA voltage converter. Both circuits are simulated in Cadence. {F0=10}3 and N=32. The average correlation coefficient of the SA voltage converter is quite lower than the average correlation coefficient of the CoRe voltage converter when the attacker selects a fast clock frequency to perform the LPA attack, as shown in
B. Sampling Multiple Clock Periods as One Sample of Input Power Data
The technique of sampling multiple clock/switching periods as one sample of input power data is quite efficient for filtering the power noise generated from reshuffling-based voltage converters in DPA attacks. When an attacker implements an LPA attack on a cryptographic circuit that houses a CoRe voltage converter or an SA voltage converter, the attacker can also filter the reshuffling noise by sampling K, (K≥2) number of clock periods as one sample of input power data instead of lowering the clock frequency
further, as shown in
where P(i-1)K+j, (j=1,2, . . . ) is the leakage power dissipation of the cryptographic circuit induced by the ((i−1)K+j)th input data. WX
As compared to sampling a single clock period as one sample of input power data, sampling K number of clock periods as one sample of input power data would enhance the MTD value to K times. Therefore, the MTD enhancement ratio R2(KF0Ts) of a cryptographic circuit that employs a voltage converter is
when utilizing K number of clock periods as one sample of input power data.
When the attacker increases the sampling time period to KF0Ts, the average correlation coefficient of the SA voltage converter has a marginal enhancement, as shown in
D(Pl,i(θ,FTs))=D(FPi)=D(KF0Pi)=K2F02σs2, (13)
where σs2 is the variance of the leakage power dissipation of the cryptographic circuit. However, the variance of load power of a voltage converter while sampling K number of clock periods as one sample of input power data D (Pl,i(θ; KF0Ts)) is (F0>1)
As compared to sampling K number of clock periods as one sample of input power data, further lowering clock frequency fc can therefore enhance the variance of the load power of the voltage converter over K times. A larger variance of the load power enhances the SNR of the voltage converter and decreases the lowest MTD enhancement ratio.
Lowering clock frequency fc further is more efficient than sampling multiple clock periods as one sample of input power data to enhance the power of LPA attacks. When the attacker further lowers clock frequency fc, as shown in
Circuit Level Verification
To validate the proposed countermeasure with circuit level simulations, a 130 nm CMOS S-box is used as the load to simulate the correlations between the input and load power profiles of different voltage converters. A 32-phase 2:1 CoRe voltage converter and a 32-phase 2:1 SA voltage converter are used in the simulations. The input voltage Vin and output voltage Vout of the voltage converters used in the simulations are, respectively, 2.4 V and 1.2 V. Additionally, the clock frequency fc of the S-box to perform an LPA attack is reduced to 2 MHz and the variation range of the switching frequency fs of the voltage converter is fs∈[19 MHz, 21 MHz].
The load current of the SA voltage converter is significantly higher than the CoRe voltage converter when the S-box is under LPA attacks, as shown in
generating a large amount of uncertain power noise in input power profile against LPA attacks.
LPA Attacks Simulation
When LPA attacks are implemented (simulated) on an Sbox that does not house any Countermeasure, the correct key (which is (66)10 in this example) is leaked to the attacker after analyzing 500 leakage power traces, as shown in
In conclusion, a security-adaptive (SA) voltage converter is utilized as a lightweight countermeasure against LPA attacks. The discharging resistor in the SA voltage converter can significantly increase the amount of noise insertion in the input power profile when LPA attacks are sensed by the proposed technique. Through scrambling the redundant load power dissipation in the input power profile, the MTD value of a cryptographic circuit that employs the SA voltage converter is enhanced over 6145 times as compared to the MTD value of a conventional cryptographic circuit that has no countermeasure.
Guidelines on the Selection of a Suitable Critical Frequency FC,1 to Maximize Security
Two different kinds of noise may impact the MTD enhancement ratio of a cryptographic circuit that employs a CoRe voltage converter: i) measurement power noise from devices that are used to perform the measurement and ii) reshuffling power noise from the CoRe voltage converter.
When a cryptographic circuit is in a normal working mode (i.e., clock frequency fc≈F1fs and F1 is an integer), the measured input power PMIP,i of the CoRe voltage converter induced by the ith input data is
PMIP,i=P*in,i(θ,1/(F1fs))+PM,i, (15)
where P*in,i (θ, 1/(F1fs)) is the actual input power of the CoRe voltage converter induced by the ith input data and PM,i is the corresponding measurement power noise. When the variance of P*in,i (θ, 1/(F1fs)) is σ12(θ, 1/(F1fs), the average variance σ12(θ, 1/(F1fs) of P*in,i (θ, 1/(F1fs)) becomes
Accordingly, the signal-to-noise ratio (SNR) of the input power profile SNRM(1/(F1fs) can be written as
where σM2 is the variance of the measurement power noise.
However, when the attacker lowers the clock frequency from F1fs to fc (i.e., F1fs/fc is an integer, the attacker can measure F1fs/fc number of leakage power data), the total measured input power PTMIP,i of the CoRe voltage converter induced by the ith input data is
where PM,i,j1 is the corresponding measurement power noise related with the j1th measurement under the ith input data. Therefore, the SNR of the input power profile SNRM(1/fc) can be written as
The correlation coefficient γM(1/fc) between the actual input power and measured input power of the CoRe voltage converter with measurement power noise when the clock frequency is fc can be written as
When the clock frequency is fc and the average correlation coefficient between the actual input power and load power of the CoRe voltage converter is
γt(1/fc)=γM(1/fc)×
The total MTD enhancement ratio MTDt(1/fc) induced by the measurement power noise and reshuffling power noise from the CoRe voltage converter is
As compared to a cryptographic circuit with the clock frequency of (1/f0) fs, the MTD value of a cryptographic circuit with the clock frequency of fc would be enhanced fs/(fc F0) times. MTDt(1/ff) therefore becomes
As shown in
Various features and advantages of the invention are set forth in the following claims.
This application claims the benefit of priority to U.S. Provisional Patent Application No. 62/491,347, which was filed on Apr. 28, 2017, the entire contents of which is hereby incorporated by reference in its entirety.
This invention was made with government support CCF1350451 awarded by the National Science Foundation. The Government has certain rights to the invention.
Number | Name | Date | Kind |
---|---|---|---|
20180004944 | Nagata | Jan 2018 | A1 |
Entry |
---|
Ahmad et al., “Low-power compact composite field AES S-Box/Inv S-Box design in 65 nm CMOS using novel XOR gate,” Integr. VLSI J., 2013, 46(4): 333-344. |
Alioto et al., “Effectiveness of leakage power analysis attacks on DPA-resistant logic styles under process variations”, IEEE Trans. Circuits Syst. I Reg. Papers, 2014, 61(2): 429-442. |
Alioto et al., “Leakage power analysis attacks: A novel class of attacks to nanometer cryptographic circuits,” IEEE Transactions on Circuits and Systems I: Regular Papers, 2010, 57(2): 355-367. |
Andersen et al., “A 4.6 W/mm 2 power density 86% efficiency on-chip switched capacitor DC-DC converter in 32 nm SOI CMOS,” Proc. Appl. Power Electron. Conf. Expo. (APEC), 2013, pp. 692-699. |
Avirneni et al., “Countering power analysis attacks using reliable and aggressive designs,” IEEE Transactions on Computers, 2014, 63(6): 1408-1420. |
Saddam et al., “Evaluation of dynamic voltage and frequency scaling as a differential power analysis countermeasure,” Proc. 20th Int. Conj VLSI Des., 2007, pp. 854-862. |
Jeon, “Fully integrated on-chip switched capacitor DC-DC converters for battery-powered mixed-signal SoCs,” Ph.D. dissertation, Dept. Electr. Comput. Eng., Northeastern Univ., Boston, MA, USA, Oct. 2012. |
Kar et al., “Exploiting fully integrated inductive voltage regulators to improve side channel resistance of encryption engines,” Proc. ISLPED, 2016, pp. 130-135. |
Kopf et al., “An information-theoretic model for adaptive side-channel attacks,” Proc. CCS, 2007, pp. 286-296. |
Kopf et al., “Vulnerability bounds and leakage resilience of blinded cryptography under timing attacks,” Proc. IEEE CSF, 2010, pp. 44-56. |
Maghrebi et al., “Entropy-based power attack,” Proc. IEEE Int. Symp. HOST, pp. 1-6. |
Moradi, “Side-channel leakage through static power-should we care about in practice?,” Springer Cryptographic Hardware and Embedded Systems, 2014, pp. 562-579. |
Pozo et al., “Side-channel attacks from static power: When should we care?,” Proc. Design, AutoIllation and Test in Europe (DATE), 2015, pp. 145-150. |
Ramadass et al., “A fully-integrated switched-capacitor step-down dc-dc converter with digital capacitance modulation in 45 nm CMOS,” IEEE J. Solid-State Circuits, 2010, 45(12): 2557-2565. |
Seeman, “A design methodology for switched-capacitor DC-DC converters,” Ph.D. dissertation, Electr. Eng. Comput. Sci., Univ. California Berkeley, Berkeley, CA, USA, May 2009. |
Standaert et al., “An overview of power analysis attacks against field programmable gate arrays,” Proc. IEEE, 2006, 94(2): 383-394. |
Tokunaga et al., “Securing encryption systems with a switched capacitor current equalizer,” IEEE J. Solid-State Circuits, 2010, 45(1): 23-31. |
Uzun et al., “Converter-gating: A power efficient and secure on-chip power delivery system,” IEEE J. Emerging Sel. Topics Circuits Syst., 2014, 4(2): 169-179. |
Yu et al., “A voltage regulator-assisted lightweight AES implementation against DPA attacks,” IEEE Transactions on circuits and Systems I: Regular Papers, 2016, 63(8): 1152-1163. |
Yu et al., “Charge-Withheld Converter-Reshuffling: A Countermeasure Against Power Analysis Attacks,” in IEEE Transactions on Circuits and Systems II: Express Briefs, 2016, 63(5): 438-442. |
Yu et al., “Exploiting voltage regulators to enhance various power attack countermeasures,” IEEE Transactions on Emerging Topics in Computing, 2016, 6(2): 244-257. |
Yu et al., “False Key-Controlled Aggressive Voltage Scaling: A Countermeasure Against LPA Attacks,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2017, 36(12): 2149-2153. |
Yu et al., “Leveraging on-chip voltage regulators as a countermeasure against side-channel attacks,” Proc. IEEE DAC, 2015, pp. 1-6. |
Yu et al., “Security-Adaptive Voltage Conversion as a Lightweight Countermeasure Against LPA Attacks,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2017, 25(7): 2183-2187. |
Yu et al., “Time-delayed converter-reshuffling: An efficient and secure power delivery architecture,” IEEE Embedded Syst. Lett., 2015, 7(3): 73-76. |
Zhu et al., “Employing symmetric dual-rail logic to thwart LPA Attack,” IEEE Embedded Systems Letters, 2013, 5(4): 61-64. |
Number | Date | Country | |
---|---|---|---|
20180316489 A1 | Nov 2018 | US |
Number | Date | Country | |
---|---|---|---|
62491347 | Apr 2017 | US |