CN107017990A - System and method for stable physics unclonable function - Google Patents
System and method for stable physics unclonable function Download PDFInfo
- Publication number
- CN107017990A CN107017990A CN201610892280.0A CN201610892280A CN107017990A CN 107017990 A CN107017990 A CN 107017990A CN 201610892280 A CN201610892280 A CN 201610892280A CN 107017990 A CN107017990 A CN 107017990A
- Authority
- CN
- China
- Prior art keywords
- mismatch
- value
- distribution
- categorized
- puf
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L9/00—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
- H04L9/32—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols including means for verifying the identity or authority of a user of the system or for message authentication, e.g. authorization, entity authentication, data integrity or data verification, non-repudiation, key authentication or verification of credentials
- H04L9/3271—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols including means for verifying the identity or authority of a user of the system or for message authentication, e.g. authorization, entity authentication, data integrity or data verification, non-repudiation, key authentication or verification of credentials using challenge-response
- H04L9/3278—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols including means for verifying the identity or authority of a user of the system or for message authentication, e.g. authorization, entity authentication, data integrity or data verification, non-repudiation, key authentication or verification of credentials using challenge-response using physically unclonable functions [PUF]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L9/00—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
- H04L9/08—Key distribution or management, e.g. generation, sharing or updating, of cryptographic keys or passwords
- H04L9/0861—Generation of secret information including derivation or calculation of cryptographic keys or passwords
- H04L9/0866—Generation of secret information including derivation or calculation of cryptographic keys or passwords involving user or device identifiers, e.g. serial number, physical or biometrical information, DNA, hand-signature or measurable physical characteristics
Landscapes
- Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Storage Device Security (AREA)
- Complex Calculations (AREA)
Abstract
Each embodiment of the present invention allows the natural statistics variations for utilizing the physical attribute in semiconductor devices, so as to create true random, it is repeatable and be difficult to the password position detected.In certain embodiments, this can match mismatch value to realize by the recurrence of PUF elements, to ensure that generated PUF key bits keep insensitive to environmental error in the case where not influenceing the utilization rate of available PUF elements.Pairing process can apply to any given hardware to generate more stable PUF bit sequences, and the PUF bit sequences provide higher margin for error, the quantity or its any combinations of the position of increase assigned error nargin.
Description
The cross reference of related application
This application claims enjoy on October 13rd, 2015 is submitted by Sung Ung Kwak, entitled " Systems
And Methods for Stable Physically Unclonable Functions " U.S. Provisional Application No.62/
240,991 priority, the U.S. Provisional Application is incorporated herein in entirety by reference.
Technical field
The present invention relates to the security system for recognizing and identifying, and it is more particularly related to by means of physics not
System, equipment and the method for the random encryption key generation of function (PUF) can be cloned.
Background technology
The purpose of semiconductor processes is to minimize technique change, technique change be cause the threshold value of transistor, electric capacity,
In physical unit on the wafer of the change of resistance value etc. the reason for produced tolerance limit.Protected by its repeatability used
Hold the advantage of faulty manufacturing technology (including inhomogeneous deposition and etch process), it is impossible to the change between part is completely eliminated
Change.The statistical attribute of these changes is adjoint can not only to obtain the information on part by inspection part layout.In other words, lack
The measurement result being the devil of few component-level, then can not detect or replicate these physical changes.
PUF designs are random to generate using these small but individual manufacture changes in physics semiconductor device
Sequence, unique cryptographic key.In some existing designs, individual is determined based on the mismatch in the polarity of PUF elements
Key bit is to generate individual bit result.It is relatively great amount of in the totality of PUF elements in the representative Gaussian distribution of mismatch
PUF elements are by centered on the midpoint of distribution.In view of other non-fabrication changes (for example, variation, temperature drift, relative
Ageing process, encapsulation stress, noise etc.) influence, frequently with the use of zero-one comparator open following possibility：
Polarity of the key bit based on single PUF elements undesirably changes to one (vice versa) from null value, so as to negatively influence
The repeatability of the code stored.
It is desirable that PUF elements are made up of the circuit block that big mismatch is presented, to make the influence of environmental change in equipment most
Smallization is so as to improve the repeatability and stability of generated key bit.Because the presentation in the totality of PUF elements is relative
The part of small mismatch statistically for more likely undergo the change of symbol, so it is uncomfortable to fall into PUF elements in totality
Share in the purpose of generation random key bit.Therefore, the PUF elements of relatively small mismatch are presented typically from any given
It is excluded in batch to prevent PUF from exporting bit reaction, this will be sensitive for environmental change and produces unstable close
Code key.
Unfortunately, PUF elements are selected by the polarity based on mismatch and the technical tendency of stability are improved in from total
Give up the pith of useful PUF elements in body to realize desired bit error rate (BER).However, excluding the big of PUF elements
Part inherently reduces overall utilization rate.
What is desired is that for the stability in the key bit for not influenceing negatively to be generated by electronics PUF systems and accurately
Ensure that computing system designer maintains the instrument of the high usage of PUF elements in the case of degree.
Brief description of the drawings
Embodiments of the invention will be referred to, the example of the present invention can be shown in the drawings.These accompanying drawings are it is intended that explanation
Property, and it is nonrestrictive.Although being summarized under the background of these embodiments the present invention, but it is to be understood that, this
It is not intended to limit the scope of the present invention to these specific embodiments.
Accompanying drawing (" figure ") 1 shows the drift effect in the normal Gaussian distribution for the PUF elements output for representing mismatch value.
Fig. 2 shows the exemplary Gaussian Profile for using categorized mismatch value according to each embodiment of the present invention
To generate the exemplary pairing process of PUF key bits.
Fig. 3 A and Fig. 3 B show each embodiment according to the present invention by the exemplary survey of the Vgs mismatches in MOS device
Measure and normal Gaussian distribution shifts are distributed into the exemplary bimodal (bi-modal) of the pairing mismatch data using classification.
Fig. 4 shows the expectation allowance (margin) as function error rate of each embodiment according to the present invention.
Fig. 5 shows close by using 128 stable bit PUF of recurrence pairing generation according to each embodiment of the present invention
The illustrative embodiments of the system of key.
Fig. 6 A and Fig. 6 B show that the pairing produced according to each embodiment of the present invention by the conversion of Gaussian Profile is lost
The exemplary recursive pairing of two bi-modal distributions with data.
Fig. 7 A and Fig. 7 B show the effect of the recurrence pairing in Fig. 6.
Fig. 8 shows the exemplary expected error rate of each embodiment PUF arrays when in use according to the present invention.
Fig. 9 is the flow chart for being used to generate the illustrative process of unpaired message of each embodiment according to the present invention.
Embodiment
In the following description, detail is elaborated for illustrative purposes, to provide the understanding of the present invention.So
And, it is obvious to the skilled person that the present invention can be put into practice in the case of without these details.This
Art personnel, which will appreciate that, performs reality of the invention discussed below in a variety of ways and using various means
Apply example.It will also be appreciated by the skilled artisan that other modification, using and embodiment such as other field the present invention
In the range of, the present invention can provide function in the other field.Therefore, embodiments described below exemplifies the present invention
Specific embodiment and be intended to avoid obscuring the invention.
The reference to " one embodiment " or " embodiment " means with reference to the spy described by the embodiment in the description
Determine feature, structure, characteristic or function include at least one embodiment of the present invention in.Each place occurs in the description
Phrase " in one embodiment ", " in embodiment " not necessarily referring to identical embodiment.
In addition, the relation between part in accompanying drawing or between method and step is not limited to the relation directly affected.Phase
Instead, without departing from the teachings of the present invention, it can be changed by adding intermediate member or method and step or with other
Mode changes the relation between part or method and step shown in accompanying drawing.
Herein, term " change " and " mismatch " is used interchangeably." PUF elements " and " PUF equipment " includes physics member
Other elements of part, chemical component and those skilled in the art's accreditation.
Fig. 1 shows the drift effect in the normal Gaussian distribution for the PUF elements output for representing mismatch value.In PUF elements
Mismatch may be caused by many factors, the change of these factors including doping concentration, the change of gate oxide thicknesses and
The change of the geometry tolerance limit produced during MOS device is manufactured due to faulty semiconductor fabrication process.It is present in physics to set
The mismatch information of PUF elements in standby can be obtained in a variety of manners, such as power information, magnetic information or the form of optical information.
In general, based on polarity from the given totality (for example, threshold voltage) of the mismatch value of measured PUF elements
Select PUT bits.In Fig. 1, the PUF elements with small mismatch value are typically given up, to ensure stability and prevent drift
Shifting causes unwanted bit reversal.As used herein, drift can be any change of environmental variance, such as temperature
Drift, noise etc..Therefore, many useful PUF elements in distribution are typically given up, so as to be advantageously implemented desired stabilization
Property.In detail, the upper curve in Fig. 1 is mismatch distribution, but lower curve shows the drift of any given mismatch
Distribution.This information can be obtained from the measurement result of PUF elements.
It is contemplated that the PUF elements with the mismatch at the points of 112 or -1 σ of+1 σ 114 on 102 positions of distribution, then
Mismatch value in the region 120 on the left side of distribution or minus side is considered sufficiently large to ensure position " 0 ".Equally
Ground, the mismatch value in the region 140 on the right side of distribution or positive side is considered sufficiently large to ensure position " 1 ".
However, as shown in curve 104, if the value of the mismatch in the centre portion 130 of distribution 102 is by one or more environment
Variable influence then may greatly change.
Because the edge of curve 104 is relative closer to the center of distribution, which increase the upset of unwanted PUF key bits
Possibility.If for example, any mismatch value outside +/- 1 σ bands 130 is considered as stable, then with positioned at curve 102
Below, the associated all PUF elements of value for falling in +/- 1 σ windows 130 (that is, overall 68% of Gaussian Profile 102) must
It must be rejected.It should be noted that other distributions can not eliminate the problem, because they produce similar suitable result.Favorably
Ground, method and system presented herein is unrelated with the type of the specific statistical distribution of any set of samples of physical unit.
In short, in order to realize desired bit error rate (BER) (such as, it is allowed to the operation in broad temperature range), relatively
Big a part of useful PUF elements must be given up from whole totality 102, potentially drift about and ensure stably to explain
PUF bits.However, the improvement of error rate is reduced to cost with the quantity for sacrificing available PUF elements, the available PUF elements
PUF bits can be produced.Intuitively, more using the PUF elements of the immediate vicinity positioned at distribution, then their mismatch will be smaller
And more not reproducible PUF bits are included within the section, so as to cause bigger error rate.
Therefore, it is desirable to make system and method avoid this compromise proposal between utilization rate and error rate, to cause not
PUF key bits produced by influenceing in the case of the overall utilization rate of available PUF elements keep making environmental error unwise
Sense.
Fig. 2 is shown is used for using classification, undressed PUF component mismatch number according to each embodiment of the present invention
According to exemplary Gaussian Profile produce the exemplary pairing process of PUF key bits.Gaussian Profile 202 includes 256 PUF members
Mismatch value measured by 256 of 16 × 16 bit arrays of part (not shown).The mismatch value of each PUF elements with itself is related
Connection, the mismatch value is exported by PUF element arrays.In embodiment, represent 256 mismatch values of sample measured by 256 according to
It is classified and is combined into to create random value according to the difference to the mismatch value in 250 according to amplitude and polarity.
In detail, the value of the rightmost side in the example in Fig. 2 is first PUF element 210, and the value of the leftmost side is
256 PUF elements 212.According to the distribution, first PUF element 210 and the 256th PUF element 212 are respectively provided with band opposite pole
The relatively large mismatch value of property.By contrast, the 129th PUF element 220 is of approximately zero mismatch value.It is to be understood that
By this sorting technique (assuming that totality of sufficiently large mismatch value), the value obtained it is only about half of will be it is positive, it is and another
Half will be negative.
In embodiment, once (n=256 herein) classifies to the mismatch value of PUF elements from 1 to n, just using following
Mode is matched：PUF elements with maximum positive mismatch (that is, are distributed the first PUF elements at 202 remote right positive side
210) with the PUF elements with the minimal negative mismatch being located at the minus side of the central area of distribution 202 (that is, for constructing
129th PUF element 220 of the first paired value) pairing.Next, by the second maximum positive PUF element (that is, the 2nd PUF elements
240) matched with the second minimal negative PUF elements 230 (that is, the 130th PUF element), so as to obtain second pairing result, etc..
The matching method (herein referred to as non-recursive) is continued until all 256 mismatch values being distributed in 202 all
It is paired.This causes to generate 128 random pair values altogether.Unpaired message is different.By being surveyed for each part
The data measured all are unique, and PUF key bits are made in this way for different piece, wafer, batch or encapsulation etc.
Change is insensitive.
In mathematical form, for n element, pin is performed by the way that i-th of element and (n/2+i) individual element are matched
Pairing to i=1 to n.In embodiment, the pairing of mismatch value includes subtracting mismatch value in couples, to cause relatively large number
Synthesized with less array to creating 128 random and relatively large differences.In embodiment, it is possible to implement selection circuit
The difference of mismatch value between pairing of the (not shown) to select and quantify equipment, such as by the way that numerical value is distributed between each pair
Difference.
In embodiment, for the randomness for also maintaining key bit to generate in addition to maintaining stability, randomly cut
The value that is obtained by subtraction is changed to generate 0 or 1 random value.This can be for example by mutually randomly subtracting two numbers with centering
(that is, by being randomly chosen minuend and subtrahend before subtraction operation is performed) completes.
Assuming that representing 1 to 256 index of the physical layout of the mismatch of physics PUF elements, once being classified, the index will
It is different from original index.In one embodiment, before subtraction, each element is all based on position and is allocated exponential digital, and
If the first exponential digital is more than the second exponential digital, then the first numeral is selected as minuend.If on the contrary, the first index
Numeral is less than the second exponential digital, then the first numeral is selected as subtrahend.In other words, the selection of minuend is also based on such as by referring to
Number digitized representations certain chip mismatch, thus using PUF itself randomness characteristic advantage and prevent unwanted
Random contact scar.It will be appreciated by persons skilled in the art that other mappings and classification schemes can be utilized.
Fig. 3 A and Fig. 3 B show the exemplary measurement and normal Gaussian by mismatch value according to each embodiment of the present invention
Distribution shifts are into exemplary bi-modal distribution.Mismatch is typically by numeral (for example, voltage difference delta VGS) represent.In embodiment
In, the pairing of being recognized or measured and categorized element is by the former Gaussian Profile 302 of Vgs mismatches in MOS device
Histogram changes into the bi-modal distribution 350 including matching mismatch data.Each sample in bi-modal distribution 350 is from distribution
One in 302 assemble to mismatch data point obtain.Thus, for example the generation of 128 PUF keys is needed to 256 mismatch numbers
256 PUF elements that strong point is sampled, because two elements of pairing are combined to generate single PUF bits.
Curve map 300 in Fig. 3 A is the former Gaussian Profile 302 of the mismatch before pairing.Mismatch distribution 302 can be from survey
Amount result (for example, carried out at ambient temperature during wafer sort) obtain.Curve map 350 in Fig. 3 B shows bimodal point
Cloth 350, bi-modal distribution 350 has the allowance of increase, and it is at least 1.5 times of the standard deviation for being distributed 302.In embodiment
In, once completing pairing, two numbers of every centering are just mutually subtracted, to generate relatively large difference.If for example, subtraction is produced
Raw positive number, then produced bit will be allocated the value of " 1 ", but if subtraction produces negative, then the bit will be divided
Value with " 0 ".By subtraction operation, big difference has widened safety allowance, and thereby, it is ensured that result will not be by drift effect
Influence.
The effect for combining the PUF elements of pairing by this way is that the normal distribution 302 of Vgs mismatches is converted into bimodal
Distribution 350, bi-modal distribution 350 does not include Vgs values in the excluded ranges positioned at the immediate vicinity of distribution 350.This satisfies
The standard of the minimum interval of PUF elements pair, it is intended to ensure the PUF bits that generation is stable due to improved expected error rate,
As next by discussion.
Fig. 4 shows the expectation allowance as function error rate of each embodiment according to the present invention.It is desirable that for example
The error rate (that is, the possibility that single position is mispronounced as caused by the change of polarity) produced due to ambient influnence will be zero.
However, empirical data suggests that the crash rate (failure rate) of 128 keys is million/(ppm) 350.Similarly, it is right
In 256 keys, discovery crash rate is 650ppm.Also, in some applications, for the typical failure rate of 128 keys,
1.35 σ intervals cannot also provide enough allowances.
For example, as shown in Figure 4, for 128 keys, from 1ppm to 1,000,000,000/(ppb) 100 ten times of relatively low mistakes
Efficiency 402 needs to be spaced 0.15 standard deviation of increase (herein from 1.65 to 1.8).For 128 keys of identical, from 1ppm
100 times of the relatively low crash rate to 10ppb needs to be spaced 0.25 standard deviation of increase, by that analogy.It should be noted that for this
The purpose of invention, it is not necessary to 100% utilization rate of PUF elements.Advantageously, systematic error does not interfere with result, due to these mistakes
Difference is that whole system is intrinsic and equally put on all mismatches pair.It is compared in the amplitude of the difference with being detected
When, the error as caused by noise is also insignificant.
In embodiment, in order to by the pairing scheme discussed on Fig. 2 and Fig. 3 come in addition further improve stable
Property, the pairing of simulation PUF elements is performed in a recursive manner, is explained as will be referred to Fig. 5, Fig. 5 is shown according to the present invention
Each embodiment be used for the exemplary embodiment party that the system of 128 stable PUF keys is generated by using recurrence pairing
Formula.System 500 includes chip 502,504 and ADC 530.It will be appreciated by persons skilled in the art that system 500 can include
The additional component that data are analyzed, change, amplify, handle and protected, including logical device known in the art and electricity
Source.
Inventor contemplates handles mismatch value by any mathematical operation (for example, by multiplication rather than simple subtraction).
Furthermore it is possible to select and combine any amount of mismatch value to be handled.For example, three mismatch values can be handled to generate
PUF keys.In addition, different algorithms can be used for different physical equipments, to reduce detectability and therefore enhancing peace
Quan Xing.
Chip 502,504 includes two identical but independent 16 × 16PUF arrays 510,520, and it can be used for generation figure
Two single bi-modal distributions shown in 6A and Fig. 6 B.As depicted in the figures, the pairing mismatch in Fig. 6 A and Fig. 6 B
Element at intervals of +/- 1.35 σ, i.e., by obtained from the pairing process that is discussed on Fig. 2 and Fig. 3 with identical in Fig. 3
Value.
Fig. 5 is looked back, each PUF arrays 510,520 are designed to complementary elements 504 as described earlier.Although in Fig. 5
In illustrate only two arrays, but inventor's prediction can use any amount of possible array and any combinations.Implementing
In example, once from one to n, (for example, n=256) classifies to PUF elements 504 for each array, they will be with following
Mode is paired：
By the positive mismatch (for example, number 1 of array 1 510) of maximum for coming from a bi-modal distribution with coming from another
The minimal negative mismatch (for example, number 65 of array 2 520) of bi-modal distribution is matched, to obtain the first recurrence pairing result.
The second largest positive mismatch (for example, number 2 of array 1 510) of the first bi-modal distribution will be come from and come from the second bimodal point
The minimal negative mismatch (for example, number 66 of array 2 520) of cloth is matched, until 128 all paired values are all matched somebody with somebody again
To untill.In other words, for n element 504, by the way that (i+n/2) individual element of i-th of the element and array 2 of array 1 is entered
Row pairing performs pairing to i=1 to n/2, and by by i-th of the element and (i-n/2) individual member of array 2 of array 1
Part is matched performs pairing to i=n/2+1 to n.
This method provides two elements for each bi-modal distribution, i.e., four elements can be by this four element generations 0
With 1 key bit.In embodiment, similar to Fig. 2, as previously, such as, by distributing 0 or 1 value based on index, subtract
Paired value and the random switching of application are to generate the value of 0 or 1.However, compared with Fig. 2 non-recursive method, expecting recurrence
The property relatively large interval of method output, i.e. allowance.
In embodiment, the interval of recursiveness method is that at least twice of non-recursive method is big.For example, such as Fig. 6 A and figure
In 6B as can be seen that and -4 the associated mismatch mismatches associated with+1.35 σ (660) of σ (610) matching generation
5.35 relatively wide total linear spacing allowance, this is 1.35 almost bigger 4 times than specific pairing 630.Even in consideration bi-modal distribution
600th, 650 non-linear attributes and check be located at represent density intermediate point distribution centre mismatch value when, come from
The associated mismatch of bi-modal distribution 600 and -1.46 σ (614) and come from bi-modal distribution 650 and about+1.46
The associated mismatches of σ (662) pairing by produce about 2.93 total linear spacing allowance, this has exceeded can be by non-recursive side
Twice of 2.7 interval that method is obtained.Or even minimum availability interval is also big by two by 1.35 than any given pairing therefore,
Times.
According to an embodiment of the invention recurrence pairing an advantage be this method can iteratively be used in it is any to
To be selectively generating more stable PUF bits or increase the quantity of position on fixed hardware.
Fig. 7 A and Fig. 7 B show the influence of the recurrence pairing in Fig. 6.Fig. 7 A are shown and identical bimodal in Fig. 3 B
Distribution 700, the bi-modal distribution 700 is matched by onrecurrent when applied to Gaussian Profile and produced.By contrast, it is double in Fig. 7 B
Modal distribution 750 is shown by using result obtained from the system of the application recurrence pairing presented in Fig. 5.It can such as see
Go out, the random value in distribution 750 in Fig. 7 B is also scattered in bimodal mode.However, increased spacing distance is advantageously provided
Larger allowance.The empirical data for coming from two kinds of different matching methods is compared, it can be seen that interval is from Fig. 7 A
Onrecurrent pairing 1.35 increase in Fig. 7 B recurrence pairing 2.93.In other words, spacing distance is more than big twice.Cause
This, such as, for 128 keys, can use recurrence matching method to realize 0.4ppb crash rate.
Fig. 8 shows the exemplary expected error rate of each embodiment PUF arrays when in use according to the present invention.Table
Number N in 800 indicates that embodiments in accordance with the present invention are used to perform onrecurrent pairing (that is, N=1) or recurrence pairing (that is, N>1)
Exemplary 16 × 16 array quantity.Although up to 8 arrays are listed in fig. 8, it will be appreciated by persons skilled in the art that
The quantity of possible array and combinations thereof is not limited.(the Δ V in units of σGS) allowance 810 that represents represented based on given
The achievable interval of array 810 of quantity.The standard deviation 820 always drifted about is based on empirical data and the (Δ also in units of σ
VGS) represent.Empirical data of the value of partial failure rate 830 based on exemplary 128 keys.
As shown in figure 8, when recurrence pairing is extended to including multiple PUF arrays, it is double in the quantity of each array 802
When, allowance 810 quadruples incessantly.This result is converted into enhanced PUF stability and the degree of accuracy without more accurate
Measurement result.Further, since the quantity 802 of array is double, the standard deviation 820 always drifted about increases √ 2 factor, and portion
Point crash rate 830 is correspondingly reduced, and further highlights this highly scalable model based on modularization PUF Array Designs
Benefit, modularization PUF Array Designs need minimum or need not add the change of array to improve stability.
In embodiment, some or all of functions in the function of modular system can be implemented with software.This area skill
Art personnel will realize that the degree of accuracy can be got in return with longer key length.For example, can be with constant interval addition more
Array is to increase key bit length (for example, from 128 to 256).Alternatively, for PUF stability and the purpose of the degree of accuracy, it is
Increase spacing distance is to realize bigger allowance, and bit length can keep constant.
It is to be understood that each embodiment of the present invention can apply to any physical characteristic with natural variable, example
Such as threshold voltage, frequency of oscillation, resistance, electric capacity.In one embodiment, the different qualities of composition element pair are to create mathematics
Computing (for example, Vt mismatches and capacitance mismatch).In addition, it will be appreciated by persons skilled in the art that, various memory constructions can
For storage unpaired message.
Fig. 9 is the flow chart of the illustrative process for generating unpaired message according to each embodiment of the present invention." when "
When determining mismatch data for two to multiple PUF elements, the process for generating unpaired message originates at 902.
At step 904, mismatch data is classified by amplitude and polarity, to obtain with certain mathematical distribution
The group of categorized mismatch data.
At step 906, data pair are selected from the group of categorized mismatch data according to some selection mechanisms, to give birth to
Into the bimodal data distribution being made up of paired value., can be from one generated by one or more PUF arrays in embodiment
Or selection pair in multiple bi-modal distributions.In embodiment, can mutually subtract to individual values, between increasing between them
Gauge from.
At step 908, random switching to individual values, and polarity is distributed to paired value to keep randomness.
At step 910, bit sequence is generated by paired value using bi-modal distribution data.
Finally, at step 912, undressed PUF path mismatch data are wiped from memory.
It will be appreciated by persons skilled in the art that, it is without departing from the scope of the invention, less or additional
Step can be collectively incorporated into shown step herein.The arrangement of square frame in flow chart or description herein does not imply that spy
Fixed order.
It will be appreciated by, previous example and embodiment is exemplary and be in order at clear and understandable
Purpose and do not limit the scope of the invention.After specification and studying accompanying drawing is read, to those skilled in the art
Obvious all displacements, enhancing, equivalents, combination and improvement are intended to and are included within the scope of the invention.Cause
This, it is intended to claims are including falling into all such modifications in true spirit and scope of the present invention, displacement and waiting
Effect form.
Claims (20)
1. a kind of device for being used to generate PUF key bits, described device includes：
Sensing element, the sensing element be configured as detect PUF elements between physical characteristic in terms of mismatch and output
The signal associated with the mismatch；
Data converter, the data converter is coupled to the sensing element, and the data converter is by the output signal
It is converted into the mismatch data associated with the first distribution；
Sort module, the sort module is coupled to receive the mismatch data, the selection from the data converter
Module is classified to generate categorized mistake by amplitude and polarity at least some mismatch datas in the mismatch data
With value；
Matching module, the matching module is matched to generate and the second distribution for the mismatch value categorized at least two
Associated paired value；And
Sequencer, the sequencer generates bit sequence using the described second distribution.
2. device according to claim 1, wherein, the matching module is matched to generate to categorized mismatch value
First paired value associated with the second distribution, second paired value associated with the 3rd distribution, and the matching module makes
The 4th distribution is generated with first paired value and second paired value.
3. device according to claim 1, wherein, by by the first categorized mismatch value associated with big positive number
The second categorized mismatch value associated with relatively small negative is combined to generate the paired value.
4. device according to claim 3, in addition to handover module, the handover module are described categorized for switching
The order of at least some categorized mismatch values in mismatch value, the polarity of one represented in zero-sum 1 is distributed to described
Paired value.
5. device according to claim 1, wherein, the difference between two paired values equals or exceeds predetermined value.
6. device according to claim 1, wherein, multiple PUF elements are located on independent array.
7. device according to claim 6, wherein, the described first categorized mismatch value is associated with the first array, and
And the second categorized mismatch value is associated with the second array.
8. device according to claim 1, wherein, the mismatch data storage in the first memory, and works as second
The mismatch data is rewritten during paired data described in memory storage.
9. device according to claim 8, wherein, the first memory includes receiving the deposit of the mismatch data
Device.
10. device according to claim 1, wherein, implemented with software in the sort module and the matching module
At least one module.
11. device according to claim 1, wherein, the sensing element is detected to mismatch in start-up conditions.
12. device according to claim 1, wherein, before classifying to pairing, the data converter will be described
Mismatch data is digitized.
13. device according to claim 10, wherein, handover module includes summation module, and the summation module subtracts base
In the categorized mismatch value switched of the output signal.
14. a kind of method for reliably regenerating the unique bit sequence generated at random, methods described includes：
It is determined that the mismatch data related at least one physical characteristic of multiple PUF elements, the mismatch data and the first distribution
It is associated；
At least some mismatch datas in the mismatch data are classified to generate categorized mistake by amplitude and polarity
With value；
Categorized mismatch value is matched, to generate first paired value associated with the second distribution；And
Bit sequence is generated using the described second distribution.
15. method according to claim 14, wherein, pairing includes categorized by associated with big positive number first
Mismatch value and the associated with relatively small negative second categorized mismatch value are combined to generate paired value.
16. method according to claim 15, wherein, the described first categorized mismatch value is associated with the first array,
And the described second categorized mismatch value is associated with the second array, first array and second array are mutually only
It is vertical.
17. it is at least some through dividing in method according to claim 14, in addition to the switching categorized mismatch value
The order of the mismatch value of class by polarity to distribute to the paired value, and the polarity represents one in zero-sum 1.
18. method according to claim 14, wherein, first distribution is normal distribution and second distribution is
Bi-modal distribution.
19. a kind of method for reliably regenerating the unique bit sequence generated at random, methods described includes：
It is determined that the mismatch data related at least one physical characteristic of multiple PUF elements, the mismatch data and the first distribution
It is associated；
At least some mismatch datas in the mismatch data are classified to generate categorized mistake by amplitude and polarity
With value；
Categorized mismatch value is matched, is distributed with generation first paired value associated with the second distribution and with the 3rd
The second associated paired value；
First paired value and second paired value are matched to generate the 4th distribution；And
Bit sequence is generated using the described 4th distribution.
20. method according to claim 19, wherein, the first interval of the 4th distribution is the of first distribution
At least twice at two intervals is big.
Applications Claiming Priority (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US201562240991P | 2015-10-13 | 2015-10-13 | |
| US62/240,991 | 2015-10-13 | ||
| US15/264,872 US10771246B2 (en) | 2015-10-13 | 2016-09-14 | Systems and methods for stable physically unclonable functions |
| US15/264,872 | 2016-09-14 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107017990A true CN107017990A (en) | 2017-08-04 |
| CN107017990B CN107017990B (en) | 2021-05-04 |
Family
ID=59439082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610892280.0A Active CN107017990B (en) | 2015-10-13 | 2016-10-13 | System and method for stable physically unclonable functions |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107017990B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111865617A (en) * | 2020-08-04 | 2020-10-30 | 上海交通大学 | Method for enhancing system reliability based on physical unclonable function |
| CN114826622A (en) * | 2022-06-27 | 2022-07-29 | 深圳大学 | Optical reconfigurable PUF device based on CMOS image sensor |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101421971A (en) * | 2006-04-11 | 2009-04-29 | 皇家飞利浦电子股份有限公司 | Attack detection with coating puf |
| CN101542496A (en) * | 2007-09-19 | 2009-09-23 | 美国威诚股份有限公司 | Authentication with physical unclonable functions |
| US7702927B2 (en) * | 2004-11-12 | 2010-04-20 | Verayo, Inc. | Securely field configurable device |
| CN102037676A (en) * | 2008-03-20 | 2011-04-27 | 日内瓦大学 | Secure item identification and authentication system and method based on unclonable features |
| CN104521177A (en) * | 2011-12-06 | 2015-04-15 | 本质Id有限责任公司 | Soft decision error correction for memory based puf using a single enrollment |
-
2016
- 2016-10-13 CN CN201610892280.0A patent/CN107017990B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7702927B2 (en) * | 2004-11-12 | 2010-04-20 | Verayo, Inc. | Securely field configurable device |
| CN101421971A (en) * | 2006-04-11 | 2009-04-29 | 皇家飞利浦电子股份有限公司 | Attack detection with coating puf |
| CN101542496A (en) * | 2007-09-19 | 2009-09-23 | 美国威诚股份有限公司 | Authentication with physical unclonable functions |
| CN102037676A (en) * | 2008-03-20 | 2011-04-27 | 日内瓦大学 | Secure item identification and authentication system and method based on unclonable features |
| CN104521177A (en) * | 2011-12-06 | 2015-04-15 | 本质Id有限责任公司 | Soft decision error correction for memory based puf using a single enrollment |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111865617A (en) * | 2020-08-04 | 2020-10-30 | 上海交通大学 | Method for enhancing system reliability based on physical unclonable function |
| CN114826622A (en) * | 2022-06-27 | 2022-07-29 | 深圳大学 | Optical reconfigurable PUF device based on CMOS image sensor |
| CN114826622B (en) * | 2022-06-27 | 2022-09-13 | 深圳大学 | Optical reconfigurable PUF device based on CMOS image sensor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107017990B (en) | 2021-05-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10771246B2 (en) | Systems and methods for stable physically unclonable functions | |
| US10397251B2 (en) | System and method for securing an electronic circuit | |
| KR101377869B1 (en) | Identification of integrated circuits | |
| US9485094B1 (en) | Systems and methods for stable physically unclonable functions | |
| KR101627892B1 (en) | Integrated silicon circuit comprising a physically non-reproducible function, and method and system for testing such a circuit | |
| Baturone et al. | Improved generation of identifiers, secret keys, and random numbers from SRAMs | |
| EP3039547B1 (en) | Systems and methods for analyzing stability using metal resistance variations | |
| WO2013155522A1 (en) | Electronic physical unclonable functions | |
| CN103946854A (en) | Retention based intrinsic fingerprint identification featuring a fuzzy algorithm and a dynamic key | |
| He et al. | Reliable and efficient PUF‐based cryptographic key generator using bit self‐tests | |
| CN107017990A (en) | System and method for stable physics unclonable function | |
| Zalivaka et al. | FPGA implementation of modeling attack resistant arbiter PUF with enhanced reliability | |
| CN114969844A (en) | Detection and mitigation of unstable cells in an array of unclonable cells | |
| KR20200142337A (en) | Security device including physical unclonable function cells and operation method thereof | |
| CN106576040B (en) | Error rate control method for device-specific information | |
| JP2013131869A (en) | Individual information generation device, encryption device, authentication device, and individual information generation method | |
| Gebali et al. | Review of physically unclonable functions (pufs): structures, models, and algorithms | |
| Saqib et al. | ASIC implementation of a hardware‐embedded physical unclonable function | |
| Ahmadi et al. | A machine learning approach to fab-of-origin attestation | |
| KR20220041048A (en) | Systems and methods for using hybrid boolean networks as functions that cannot be physically replicated | |
| CN107646130A (en) | For generate the intrinsic value of electronic circuit method, generate the value electronic circuit and method for being worth as use | |
| Mondal et al. | A new online testing technique for reversible circuits | |
| Lu et al. | Improving fuzzy C‐means clustering algorithm based on a density‐induced distance measure | |
| Sadr et al. | Weighted Hamming distance for PUF performance evaluation | |
| Tariguliyev et al. | Reliability and security of arbiter‐based physical unclonable function circuits |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |