Circuit, method and product for synthesizing physically unclonable functions of connection
By designing the connected synthetic physically uncloned function (CSPUF), using the statistical properties of sensor and actuator signals and external data sources to generate digital signatures, the existing PUFs are solved instability and aging management problems in complex environments, and realizes the implementation of digital signatures. More efficient computer security and networking solutions.
Patent Information
- Application Number
- CN201980046401.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2019-06-27
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2039-06-27
AI Technical Summary
The existing physically uncloned function (PUF) is unstable in complex environments, difficult to efficiently manage aging, and has strong noise dependence, which limits its application in computer security.
A connected synthetic physically non-cloneable function (CSPUF) is designed to determine the statistical properties of the noise distribution by receiving sensors and actuators of computer equipment, and generate digital signatures based on data from external data sources to achieve the response. Stabilize and manage. The CSPUF can be calibrated statically or dynamically, disable circuit options, supports self-cycling, iterative and blockchain arrangements, enhancing its unclonability and controllability.
By reducing dependence on noisy environments, the response is more stable, the non-cloneability and management efficiency of PUF is improved, new networking solutions and devices are supported, and computer security is enhanced.
Smart Images

Figure CN112805767B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates generally to physical functions, such as Physical Unclonable Functions (designated hereinafter by the acronym PUF), and in particular to concatenated synthetic PUFs. Background Art
[0002] Computer security has become a major challenge over the past few decades. Attacks have become increasingly sophisticated and efficient. At the same time, computer science applications are expanding, for example using the Internet of Things. In various technological fields, it is known to use physical features to ensure computer security.
[0003] The existing literature on PUFs is emerging and has studied both the underlying PUF physics and PUF properties (eg, aging, reliability).Several PUF-based implementations have been provided.
[0004] For example, patent document US8694778B2 describes a method of using a PUF device that receives a communication including a first challenge value, a second challenge value, and a remote message authenticity value. The method includes generating additional challenge-response pairs in a secure manner. The additional challenge-response pairs are securely communicated between the PUF device and an authentication server or other device subsequently used in authentication. However, this method has limitations.
[0005] Therefore, there is a need for advanced methods and systems for building and managing improved physically unclonable functions. Summary of the invention
[0006] A connected synthetic physically unclonable function (acronym CSPUF) is disclosed, the CSPUF consisting of a circuit configured to: receive signals from one or more sensors and / or actuators in a computer device / one or more sensors and / or actuators of a computer device; determine one or more statistical properties of the noise distribution of said selected one or more of said sensors and / or actuators; receive data IN from one or more external data sources; determine one or more digital signatures (responses) based on said statistical properties and said selected external data. In one embodiment, along with a response R when challenged by a challenge C, the circuit is configured to receive data IN from one or more external data sources and / or transmit data OUT. Developments describe the uses and advantages of the data IN and data OUT channels, such as static or dynamic calibration, options to disable the circuit. Other embodiments contemplate variations of the interconnection of two CSPUF circuits, thereby providing "self-looping", "iterative", "cascading" and other "blockchain" arrangements.
[0007] Thus, a "synthetic PUF" (acronym SPUF) according to the present disclosure can appropriately utilize existing or otherwise available sensors and / or actuators in consumer electronic devices. A connected SPUF (CSPUF) adds connectivity to the SPUF, allowing for several advantages applicable to a variety of use cases.
[0008] The response of a CSPUF can be stabilized, for example, by reducing dependence on noisy environments. The unclonability of the PUF can be enhanced. Aging of the PUF can be managed more efficiently. Connected SPUFs can enable new networking schemes, devices, or networks (e.g., series and / or parallel arrangements, distributed networks, self-looping CSPUFs, cascaded networks, etc.). The connectivity of CSPUFs may mean greater controllability of PUF devices. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate various embodiments of the invention and, together with the general description of the disclosure given above and the detailed description of the embodiments given below, serve to explain embodiments of the invention but in no case limit the scope thereof.
[0010] Figure 1 A connected synthetic physically unclonable (CSPUF) device according to the present invention is shown.
[0011] Figure 2 The effect of noise is shown.
[0012] Figure 3 A representation of a CSPUF with a bidirectional communication channel is shown.
[0013] Figure 4 Some embodiments of the invention are shown in which data IN are received.
[0014] Figure 5 Some embodiments of the invention are shown where data is received and further modifies the SPUF itself.
[0015] Figure 6 Some embodiments of the invention are shown in which data is sent out.
[0016] Figure 7 , Figure 8 and Fig. 9 Various embodiments of the invention are shown according to different retroactions.
[0017] Figures 10 to 14 Various embodiments of the invention according to different reactions are shown.
[0018] Figures 15 to 18An exemplary combination of two CSPUFs is shown. DETAILED DESCRIPTION
[0019] As used herein, "computer device" designates a consumer electronic apparatus (e.g., a smart phone, television, microwave, refrigerator, robot, drone, connected bicycle, car, airplane, etc.) having communication resources that enable access to computing resources.
[0020] As used herein, a "physically unclonable function" (or "PUF") designates a physical function. A PUF is inherently tangible. A PUF is a hardware function that provides a response when a challenge is applied to the hardware in question. While the associated properties of a PUF (e.g., "reliability" or "unclonability" properties) are subtle and still under debate in the art, the term PUF appears to be a naming convention. In this disclosure, the method steps described do characterize a PUF. Embodiments of the present invention enable the creation of a "physically unclonable function," which is otherwise known as a synthetic PUF: the generally accepted properties of a PUF match the properties that emerge once the method steps are performed.
[0021] A challenge-response pair (acronym CRP) specifies a pair (a single pair). Given the hardware under consideration, a PUF corresponds to all possible CRPs. Knowing all the pairs may allow perfect simulation of a (unique by construction) PUF, or more appropriately knowing all the pairs is sufficient to handle replay attacks.
[0022] Therefore, hardware is essential for PUF, because hardware is the source of entropy and, therefore, the source of unclonability. In other words, PUF cannot be "pure" software. Software can help exploit the underlying hardware entropy in a reliable way. Such software will be referred to as F or F_HW in the following.
[0023] As used herein, a "synthetic physically unclonable function" (acronym "SPUF") specifies a synthetic or composite assembly of sensors and / or actuators (hereinafter referred to as "sensors" for simplicity) selected for the statistical properties of the distribution of noise of the sensors and / or actuators. Such an embodiment can usually be constructed in a given physical device, but distributed embodiments are still possible. Patent document EP3340214 filed by the applicant describes an example of SPUF.
[0024] Other examples of SPUFs include CMOS SPUFs.
[0025] A first example of a CMOS SPUF can be obtained by performing the following method and / or its variations. This method comprises the following steps:
[0026] - define a subset of sensor components of a sensor;
[0027] - Interrogate a subset of sensor components under consistent conditions;
[0028] - receiving output signal values from a subset of sensor components and for each component in the subset of sensor components, determining an i-th order statistical moment of a temporal distribution of the output signal values of each sensor component;
[0029] - determining one or more outlier sensor components, the outlier sensor component being a component whose i-th order statistical moment differs from the mean of the spatial distribution of moments selected in a subset whose absolute value is better than a threshold, estimating the i-th order statistical moment of a sensor component over the time distribution associated with the sensor component.
[0030] In some embodiments, the sensor may be an imaging sensor and the sensor components may be active pixels, whereby the imaging sensor includes a plurality of active pixels.
[0031] In one embodiment, the step of interrogating a subset of active pixels under uniform conditions comprises the step of acquiring a plurality (N) of images.
[0032] The sensor may be a combination of independent sensors.
[0033] In some embodiments, the method may further include the step of generating an identifier based on the determined stray sensor component.
[0034] In one embodiment, the method may further comprise the step of comparing the generated identifier and / or its hash value with other identifiers to authenticate the imaging sensor. In development, a threshold value is received or predefined.
[0035] The threshold may be a function of one or more desired properties of the generated identifiers, including bit length and / or bit errors.
[0036] The number of images N and the threshold may be set so that the same outlier pixels are repeatedly determined for a given statistical order i.
[0037] In one embodiment, information about the outlier sensor assembly may be stored in a helper data file. The coincidence condition may be a spatially coincidence condition obtained by closing a shutter associated with the imaging sensor and / or by grounding a photodiode associated with a pixel in a subset of pixels and / or by adjusting a light source and / or by shaping a light beam. The subset of pixels may be determined based on structural information about the shared electronics of the imaging sensor.
[0038] A first example of a CMOS SPUF may be obtained by a system configured to process a sensor, the system comprising:
[0039] - a processor configured to define or receive information about a subset of sensor components of the sensor, the subset of sensor components being interrogable under consistent conditions;
[0040] - a receiver configured to receive output signal values from a subset of the sensor components.
[0041] For each component in the subset of sensor components, the processor may be configured to determine an i-th order statistical moment of the temporal distribution of the output signal value of each sensor component. The processor may also be configured to determine one or more outlier sensor components, the outlier sensor component being a component whose i-th order statistical moment has a difference from the mean of the spatial distribution of the selected moments in the subset whose absolute value is better than a threshold, estimating an i-th order statistical moment of a sensor component over the temporal distribution associated with the sensor component.
[0042] In one embodiment, the processor may be further configured to generate an identifier according to the determined outlier pixels.
[0043] A second example of a CMOS SPUF can be obtained by performing another method comprising the following steps:
[0044] - define a subset of sensor components of a sensor;
[0045] - Interrogate a subset of sensor components under consistent conditions;
[0046] - receiving output signal values from a subset of the sensor components;
[0047] - determining, for each component in the subset of sensor components, an i-th order statistical moment of the temporal distribution of the output signal values of each sensor component;
[0048] - determining one or more pathological sensor components, the pathological component being a component for which the sum of distance values to all other components in the subset is greater than a threshold, the distance between two sensor components being determined by the difference of i-th statistical moment values of two time distributions associated with the components obtained when the subset of sensor components is interrogated under consistent conditions.
[0049] In one embodiment, the sensor may be an imaging sensor, and the sensor assembly may be an active pixel, the imaging sensor thus comprising a plurality of active pixels. In one embodiment, the step of interrogating a subset of active pixels under a consistent condition may include the step of acquiring a plurality (N) of images. The sensor may be a combination of independent sensors. The method may further include the step of generating an identifier according to the determined pathological sensor assembly. The method may further include the step of comparing the generated identifier and / or its hash value with at least another identifier to authenticate the sensor. In one embodiment, a threshold may be received or predefined. The threshold may be a function of one or more desired attributes of the generated identifier, the attribute including bit length and / or bit error. The number and / or threshold of output signal values received from a subset of sensor assemblies may be set so as to repeatedly determine the same pathological sensor assembly for a given statistical order i. In one embodiment, at least some partial information about the coordinates of the pathological sensor assembly may be stored in a helper data file. The sensor may be an imaging sensor, and the consistent condition may be a spatially consistent condition obtained by closing a shutter associated with the imaging sensor. The sensor may be an imaging sensor, and the consistent condition is a spatially consistent condition obtained by grounding a photodiode associated with a pixel in a subset of pixels. In one embodiment, the sensor may be an imaging sensor and the uniform condition is a spatially uniform condition obtained over a subset of pixels by adjusting the light source and / or by shaping the light beam. In one embodiment, the image may be in a lossy compressed format. In an embodiment, the image may be a video frame.
[0050] A second example of a CMOS SPUF may be obtained by a system comprising:
[0051] - a processor configured to define or receive information about a subset of sensor components of the sensor, the subset of sensor components being interrogable under consistent conditions;
[0052] - a receiver configured to receive, for each component in the subset of sensor components, an output signal value from the subset of sensor components.
[0053] Such a processor may be configured to determine an i-th order statistical moment of the temporal distribution of the output signal values of each sensor component, and the processor is further configured to determine one or more pathological sensor components, the pathological component being a component whose sum of distance values to all other components in the subset is greater than a threshold. The distance between two sensor components may be determined by the difference of the i-th statistical moment values of the two temporal distributions associated with the components obtained when the subset of sensor components is interrogated under consistent conditions. In an embodiment, the processor is further configured to generate an identifier or key based on the determined pathological pixels.
[0054] As used herein, the term "connected SPUF" (hereinafter designated by the acronym CSPUF) is a SPUF provided with communication channels (IN and / or OUT) where data is interposed at different physical and / or logical levels within the SPUF. A non-communicating or silent CSPUF is a SPUF.
[0055] As used herein, the term "Distributed Connected Synthetic Physical Unclonable Function" (hereinafter designated by the acronym DCSPUF) is a CSPUF that utilizes parts of different physical devices (which are located in different locations, for example, distributed in an apartment or servers in the cloud). The DCSPUF requires communication. It can be considered that C is implicit in the DCSPUF, so it is also called DSPUF.
[0056] DCSPUF is a type of CSPUF (it is a specific CSPUF). CSPUF is a type of SPUF, which in turn is a type of PUF.
[0057] According to an embodiment of the present invention, there is provided a circuit, which is configured as follows:
[0058] - receiving signals from one or more sensors and / or actuators of a computer device;
[0059] - determining one or more statistical properties of a noise distribution of a selected one or more of the sensors and / or actuators;
[0060] - receiving data IN from one or more external data sources;
[0061] -Determining one or more digital signatures based on statistical properties and selected external data.
[0062] Such a circuit may be called a "connected synthetic physically unclonable function," or CSPUF.
[0063] In one embodiment, the circuit may be configured to receive data IN and / or transmit data OUT from one or more external data sources.
[0064] In one embodiment, selection may be performed between one or more sensors and / or selection may be made between external data.
[0065] In one embodiment, the received data IN can statically calibrate one or more sensors. The data IN can provide additional information about the statistical moments of the sensors, and this information can be used for calibration purposes. For example, for GNSS positioning, the gravity factor for a given altitude can be taken into account. For example, this data can be provided via an API.
[0066] In another embodiment, the received data IN can dynamically calibrate one or more sensors. In addition to a single calibration purpose, external data IN can be provided over time and thus can continuously modify the internal workings of the signature generation of the SPUF (which can then remain stable). For example, the external data can be related to the external environment in which the SPUF device is placed and in which the PUF response is generated. Information about the environment can enable or allow or at least attempt to offset, neutralize or otherwise eliminate environmental fluctuations. Advantageously, environmental instabilities can be measured and further neutralized to stabilize the PUF response.
[0067] In one embodiment, the received data IN may include information that directly or indirectly results in disabling the circuit. By receiving one or more signals or values (e.g., hashes, passphrases, strings, etc.), various results may be implied: a sub-portion (if not all of the portion) of the circuit may be modified (e.g., disabled, deactivated, attenuated, overruled, destroyed, ignored, etc.). The modification may be logical and / or physical. For example, a "kill switch" command may be sent to disable a targeted sub-portion of the circuit (if not the entire SPUF arrangement). Since CSPUFs may be chained or otherwise arranged, addressability of a particular SPUF or portion thereof may be advantageous.
[0068] The modifications may be of various degrees or forms. In an intermediate form, an adaptive and / or reactive networked CSPUF may be formed. For example, a veto mode (e.g., backup mode) may be activated in lieu of a kill switch command. Such an embodiment may be advantageous in an IoT environment where the boundaries of devices and systems may vary. For example, a magnetic sensor may monitor a magnetic field; if a predefined threshold (e.g., 4 Tesla) is exceeded, a circuit may be disabled or an alarm may be sounded.
[0069] In one embodiment, the received data IN or parts thereof may be used as or contribute to a challenge to a circuit which provides a response in response to the challenge. The external data may "enrich" or "complement" or "complete" the challenge. The combinatorial space of all possible challenges is too large for brute force exploration: the external data IN may include a secret or key or token to allow a response to be generated.
[0070] In one embodiment, the received data IN or a portion thereof may sign a challenge to a circuit in response to which the circuit provides a response. A subset of the challenge may be known or predefined (eg, a succession of challenges pseudo-randomly generated by a LFSR).
[0071] In one embodiment, the circuit may also be configured to transmit data OUT, which is related to the internal information of the circuit. For example, the data OUT may be used to provide internal information to a third party (human and / or machine), such as status data or aging information. For example, if the sensor and / or actuator drifts too fast, the data sent out may directly or indirectly lead to ignoring the response generated by the circuit 100 (which may be interrupted, degraded or otherwise malfunction). For example, in a CMOS SPUF, some areas of pixels may not be addressable or out of order. Then, the response to the challenge may be ignored. Advantageously, many downstream strategies become possible: trust thresholds, ignoring responses or ranges of responses, sending a cut-off switch command in response to a combination of excessive aging of sensors and / or actuators, another option to recombining sensors and / or actuators, receiving data with weighting or estimation, etc.
[0072] In one embodiment, the circuit may also be configured to transmit data OUT which is used as context or environmental information for the circuit. The environment sensed by the sensors making up the SPUF may be measured in terms of several of its physical properties. For example, measurements may be performed for temperature, ambient electromagnetic fields, etc. Given these measurements, an analysis of the distribution of values may be performed (e.g., i-th moment, e.g., kurtosis), and sweat spots may be identified: noise may be exploited.
[0073] In one embodiment, the circuit may also be configured to transmit data OUT to one or more third parties, which are computers or PUFs or SPUFs. As used herein, a "third party" may specify: a) a non-PUF entity, such as a data server in the cloud or a local machine that monitors the aging of a SPUF (e.g., in a smartphone for authentication during a payment process); or b) a PUF (a classic one); or c) an SPUF.
[0074] This latter linking opens up many opportunities (IoT devices, collections of servers in the cloud). In such an embodiment, it is possible to envision building a constantly changing aggregation of SPUFs interconnected via the cloud (e.g., the statistical property Alpha of the noise of sensor A in smartphone number 1 can be combined with the statistical property Beta of sensor F in smartphone 2).
[0075] In one embodiment, the submitted challenge C may modify or control or otherwise affect the transmitted data OUT. For example, the data OUT may be the result of a test performed on C.
[0076] In one embodiment, the transmitted data OUT may modify the submitted challenge C. The data OUT may modify C (complement, filter, add, etc.) before, during, or after generating R. For example, in the CMOS example, all active pixels may be set to white (via data IN) as a challenge.
[0077] In one embodiment, data IN can modify response R while or after it is generated. For example, in a CMOS CSPUF, a third party or operator can validate or invalidate the response R being generated depending on their needs. Regardless of what OUT is, key R can be manipulated (e.g., split, supplemented, etc.).
[0078] In one embodiment, the response R may react on the data IN. This embodiment implements an "iterative" CSPUF ("auto-chaining").
[0079] In some embodiments, the response R may modify the challenge C. This embodiment implements a "self-looping" SPUF along an independent communication channel IN-OUT. The response R1 to the challenge C1 may be used as a challenge C2, and so on. In one embodiment, the challenge C may be modified on the fly once the generation of the response R begins. Depending on the embodiment under consideration, part(s) or the entirety of the response may be used to modify the challenge (as an on-the-fly modification of the incoming challenge).
[0080] There is also provided a system comprising two or more of the previously described circuits, wherein the second circuit takes the transmitted data OUT of the first circuit as input data IN. Such an embodiment can be used to propagate or diffuse a specific value.
[0081] In one embodiment, such a system may include two or more of the previously described circuits, the second circuit taking the response R of the first circuit as input data IN. The link may be used as a blockchain.
[0082] In one embodiment, such a system may include two or more of the previously described circuits, the second circuit taking the transmitted data OUT of the first circuit as a challenge C. This arrangement may be called "cascading" and may be used to manage fallback, for example, as aging increases or reliability decreases.
[0083] In one embodiment, such a system may comprise two or more of the previously described circuits, the second circuit taking the response R of the first circuit as challenge C. This embodiment corresponds to a serial arrangement (or signature chain).
[0084] There is also provided a (computer-implemented) method comprising the steps of:
[0085] - receiving in the circuit a signal from one or more sensors and / or actuators of the computer device;
[0086] - determining one or more statistical properties of a noise distribution of a selected one or more of the sensors and / or actuators;
[0087] - receiving data IN from one or more external data sources;
[0088] -Determining one or more digital signatures based on statistical properties and selected external data.
[0089] A computer program product is provided comprising instructions for implementing the steps of the described method when the computer program is executed on a computer device.
[0090] Figure 1 A connected synthetic physically unclonable (CSPUF) device according to some embodiments of the present invention is shown.
[0091] In some embodiments, the SPUF 100 specifies a material combination of a selected sensor (a sensor and / or actuator represented by 1 or 2, etc.), and an operation or operable logical operation on a signal of the selected sensor (sensor and / or actuator), more precisely, an operation or operable logical operation on processing a property of a noise distribution of the sensor (sensor and / or actuator).
[0092] The connected SPUF (or CSPUF 100) adds a communication channel 110, which is either unidirectional or bidirectional depending on the embodiment.
[0093] For a given CSPUF, a variety of reactions can be described (ie, involving C, R, F, Data IN, and Data OUT). The coupling of multiple CSPUFs in turn enables a number of different advantageous embodiments (below).
[0094] Although the challenge C, response R and the physical unclonable function of the CSPUF 100 may remain the same (CRP pair or simply CRP), it is advantageous to provide the data IN and / or OUT (110) independently of the CRP.
[0095] Figure 2 The effect of noise is shown.
[0096] In practice, in industrial reality, a PUF or SPUF or CSPUF according to the present disclosure is subject to noise 200. Hardware (e.g., sensors, actuators, challenges and other parts of electronics) and / or software (e.g., signals, errors, calculations, etc.) are actually subject to noise 200 (perturbations, interferences, defects, limitations, up to quantum noise).
[0097] The term "noise" covers environmental fluctuations (e.g., the car with the SPUF embedded in it has stopped; the electromagnetic field is above a certain threshold, etc.).
[0098] In the SPUF framework (composed of sensors), for a given stable, controlled challenge C, the response R may depend on noise or environmental conditions, thereby distorting the challenge (introducing some variability) and thus reducing the reliability of the PUF implementation. For example, noise 200 may affect the challenge C and / or the (S)PUF (via its effect on the sensors that make up the composite PUF).
[0099] Thus, the inventors have analyzed (210) the distribution of noise 200 at the level of the sensor (particularly at high statistical moments 220). This analysis can be performed locally and / or remotely (in the cloud). For local embodiments, the calculations can be performed by dedicated hardware circuits, for example, and / or general purpose circuits can be used (e.g., the CPU or GPU of a smartphone requesting its sensors), the latter being direct ("pure hardware") or indirect (e.g., a software program requesting underlying computing resources). The understanding of the analysis can be used downstream to adjust or otherwise modify 230 the data IN and / or the data OUT.
[0100] Figure 3 A representation of a CSPUF with a bidirectional communication channel is shown.
[0101] Challenge C 101 is provided as input to a SPUF / CSPUF implementing CRP 100, which in turn provides an output or response R 102. Data IN 111 may be received. Data OUT 112 may be sent out. The relationship between C 101, R 102, IN 111 and OUT 112 is complex and will be described below.
[0102] Advantageously, processing of DATA IN and / or DATA OUT may improve processing of the SPUF (eg, better R stability, handling of aging, etc.).
[0103] Figure 3 An embodiment of a CSPUF with two communication directions (IN and OUT) is shown. Some CSPUFs may be provided with this dual capability, while others may not (some may be provided with IN, others only with OUT).
[0104] Questions
[0105] C represents a challenge (for a SPUF, the format may be an "image"; in other examples, it may be a value, a set of values, a signal, a shock, etc.).
[0106] In the described framework, C is given (by a human and / or machine third party, client, end user, etc.).
[0107] C is independent of the sensor(s) of the SPUF / CSPUF 100.
[0108] There is usually no way to prevent C from challenging the CSPUF 100 .
[0109] response
[0110] R is constructed by constructing a unique response of the SPUF's HW to a given challenge C. Regarding the format, in the case of SPUF, R is a key or ID or signature or bit string. The response R can be classified or machine-learned. Machine learning can actually help maximize the entropy extracted from a given HW device.
[0111] Once the hardware has been challenged, there is usually no way to prevent the response R from being generated.
[0112] Challenge-Response Pair, CRP
[0113] F specifies a CRP (challenge-response pair). F corresponds to a physical unclonable function. Each hardware (HW) has its own F HWi . The function cannot be cloned. It cannot be predicted, thus preventing attacks (by replay). Attacks on a PUF (or SPUF or CSPUF) can include exploring the possible space (scanning) and replay attacks.
[0114] The F in SPUF may be a software embodiment (reprogrammable) so as to manipulate the statistical properties (statistical moments) of the distribution. Alternatively or additionally, F may be embodied in hardware, for example as a circuit, which is also sometimes reprogrammable.
[0115] The software (SW) of the SPUF sits on top of the CSPUF HW, which analyses the distribution of the statistical properties of the noise coming from the sensor. This SW may be "part" of the CSPUF, or may involve additional resources.
[0116] Function F in SW specifies which part of the HW sensor is involved (eg which part of the CMOS, which combination of sensors), and which noise statistical moment / property is to be utilized or monitored. A- Consider a CSPUF for the CSPUF itself (no upstream reaction)
[0117] When the CSPUF itself is considered alone, ie, the CSPUF is not in a chained configuration with other devices, the relationship between C, R, data IN and data OUT is described in the first frame.
[0118] Data IN
[0119] The form and / or substance of data IN may vary. Data IN may be a parameter or value without intrinsic meaning to the SPUF. Data IN may be "external" or "absolute", originating from the outside (for whatever reason). Data IN may be random. Data IN may be determined to be equal to 1 by an operator. Data IN may be a function of another value. Data IN may be secret. Data IN may be static (blind correction and constant correction). Data IN may be dynamic (parameterized).
[0120] The data IN itself cannot "modify" C, because C is "given" by construction.
[0121] The data IN can statically and / or dynamically influence or correct C (add or append to C) to make it meaningful under different environmental conditions (offset, distortion-free). In fact, the correction depends on the environmental conditions and makes the resulting challenge truly independent of the environmental conditions.
[0122] Data IN can modify function F in the SPUF: Data IN can affect the internals, thereby determining the composite CRP challenge-response pair.
[0123] Data IN can modify the selection or limits of the sensors that make up the SPUF. Data IN can also modify the statistical analysis being performed, so Data IN can "switch" the model implemented in the SPUF (e.g., the F of the first CMOS SPUF example can be changed from 1 Switch to the second CMOS example F 2 ).
[0124] Data IN cannot lead to a multiplicative response. Instead, data IN can improve the stability (aging) of R by manipulating appropriate properties (in the noise distribution).
[0125] Data OUT
[0126] Data OUT cannot "modify" C (a priori, unless there is a reactionary loop in which data OUT affects a given way of C).
[0127] Data OUT cannot modify F of PUF or SPUF. In other words, data OUT from downstream will not affect the internals of CSPUF.
[0128] The data OUT may "leak" or provide information about the SPUF and / or its context. The data OUT may be "contextual" or "inside the SPUF", or may be "related to the SPUF", or "concerned with", i.e., related to the environment of the SPUF. For example, electromagnetic conditions or ISO values may be provided by the data OUT. The data OUT may include metadata: i.e., data about the data (providing contextual information, e.g., about environmental conditions).
[0129] The uses of data OUT are numerous and advantageous. Data OUT enables downstream regulation (trust of results to verify, authorize, mitigate, confirm, attenuate, mitigate, attenuate, weight and modulate responses). For example, data OUT can inform ISO whether the parameter is CMOS SPUF or electromagnetic ambient conditions. Such data can be used for further control (e.g., if the measured electromagnetic field is better than 4 Tesla, it can be inferred that the CSPUF is unreliable or should be recalibrated). Downstream uses of data OUT can be local (e.g., in a smartphone) and / or remote (e.g., decision logic in the cloud).
[0130] Data IN and data OUT are independent: there is no direct or indirect link between data IN and data OUT.
[0131] There is no a priori link (physically and / or logically, spatially and / or temporally) between R and the data OUT.
[0132] B- Consider chaining a CSPUF or using other counteractions to regulate in other ways
[0133] In the second framework, the CSPUF is part of the system and can be part of the regulation including feedback, feedforward and other mechanisms.
[0134] In fact, in some instances, Data IN, Data OUT, C, and R (and downstream feedback) can be tied together.
[0135] Data IN and data OUT may be made dependent or related to each other in some cases, such as where data IN indirectly modifies OUT or where data OUT is re-injected as part of data IN. In some cases, data IN may be equal to data OUT.
[0136] If there is a reaction loop (not shown) in which data OUT affects the way a given C is, then data OUT may "modify" C. For example, data OUT indicating "ISO=100" may result in a change to challenge C.
[0137] If Data OUT is re-injected into CSPUF, Data OUT from downstream can affect the interior of F or CSPUF.
[0138] Figure 4 Some embodiments of the invention are shown in which data IN are received.
[0139] The figure shows a first type of CSPUF (a “controllable” CSPUF) which features only a data entry (data IN 111 ).
[0140] exist Figure 4 In the example shown in , the data IN can modify the function F. For example, in a CMOS SPUF, the number of images acquired can be changed, or the area of pixels used to determine the signature can be changed, thereby affecting the quality.
[0141] In the example, C is independent of IN (parallel configuration).
[0142] In one embodiment, data IN can be a cut-off switch signal. In one embodiment, data IN can suppress F, and thus suppress R.
[0143] In one embodiment, the data IN may correspond to or include a secret channel.
[0144] In one embodiment, the data IN may be used for calibration purposes (static or dynamic correction), see below.
[0145] Figure 5 Some embodiments of the invention are shown where data is received and further modifies the SPUF itself.
[0146] Should Figure 5 An important feature of a CSPUF according to the present invention is shown: in some embodiments, the CSPUF may be programmable. It may be programmed once, a limited number of times, or at will (reprogrammable).
[0147] In one embodiment, the HW 510 and / or SW 520 in the CSPUF may be modified (in whole or in part). In other words, the physical unclonable function or function F or CRP may be modified by modifying the HW and / or SW.
[0148] The CSPUF is made of hardware 510 (HW), but the analysis performed on the data is specified in software 520 (SW). The SW includes executable instructions that, when executed on a computer, perform logical operations (e.g., statistical analysis). The SW utilizes or uses the underlying HW (and possibly other circuits, not shown) of the CSPUF. The SW and HW are dual in some respects: the SW can be encoded as the HW, and the HW can encode the SW. There are multiple ways to change the HW and / or SW (only the HW, only the SW, or both the HW and SW).
[0149] In some embodiments, data IN may cause the HW circuitry of the CSPFF to "wrap" or "burn," or otherwise have a tangible effect thereon.
[0150] In one embodiment, for any parameter as data IN (eg, "the weather is good" or data IN is equal to 1), it may result in a change F.
[0151] In some embodiments, data OUT may also return SW rules that affect the CSPLF. For example, if the CSPUF is too old, too degraded, or dysfunctional, the internal HW may be changed (or the way the available HW is used may be changed, i.e., the SW may be changed). For example, in a CMOS CSPUF, if there are too many or too many pathological pixels, the CSPUF may be manipulated differently.
[0152] In other words, the F or CRP or the implemented PUF method itself can be changed: for example, if and when predefined criteria are met (e.g., aging, environmental conditions, user selection, etc.), the method according to the invention can include the following steps: switching the selection (bounds) of the sensor and / or the way of analyzing or using the distribution of the sensor noise. For example, the F of the first CMOS SPUF example can be changed to 1 "Switch" to the second CMOS example F 2 In other words, given the same underlying hardware, praised for its entropy and unclonability flaws, parts of the logic that exploit that hardware can be modified or altered.
[0153] Figure 6 Some embodiments of the invention are shown in which data is sent out.
[0154] The figure shows another type of CSPUF (a "chat CSPUF").
[0155] In such an embodiment, the CSPUF may send out (or generate or circumvent or create or define or determine) data OUT.
[0156] The data OUT may be directed to various parties (e.g. including third parties that may be or include a (standard) PUF, an SPUF according to an embodiment of the invention, and a connected SPUF or CSPUF). The data OUT may also be sent to a server in the cloud for further data processing, e.g. a CSPUF implemented in a car may send data to a smartphone.
[0157] Figure 7 , Figure 8 and Fig. 9 Various embodiments of the invention according to different reactions are shown.
[0158] Depending on the use case, a variety of counteractions may be justified and advantageous. Some examples are described below.
[0159] Figure 7 An embodiment of the invention is shown in which a challenge C challenges (a) a CSPU which in turn generates (b) a response R. Before, during or after generating the response R, data OUT or a portion of data OUT is sent (c) to data IN and re-injected (d) into the CSPUF.
[0160] Such an embodiment may be advantageous. For example, in a CMOS CSPUF, the data IN may specify a subset of pixels. If a drift in the response R is detected (e.g., dead pixels or a large drift in the average value, etc.), the reaction loop may be reactive or adaptive. For example, more or different information may be injected into the data IN in order to modify F to handle future incoming challenges. In the CMOS CSPUF example, different subsets of CMOS pixels may be used. In this case, R is a function of F that is challenged by C depending on IN.
[0161] In one embodiment, the data OUT may be a counter of the number of responses R that have been generated. If too many responses have been generated, the device is likely being "scanned" or attacked. The counter may be implemented in HW and / or SW (SW observes the HW responses to the challenge).
[0162] Figure 8 An embodiment is shown in which the communication channel is affected but the CSPUF is not. For example, the data IN may originate from a third party (e.g., a server, another PUF, an SPUF, another CSPUF, etc.). In one embodiment, the data IN may be equal to the data OUT (message transmission through a chain or sequence of connected CSPUFs). In this case, the CSPUF device may carry or support data transmission in addition to its corresponding physically unclonable function (which may be used for counteraction at some point). In one embodiment, the data IN may modify the data OUT. "Modify" may mean one or more of the following actions: replace, suppress, add, change, substitute, alter, transform, vary, alter, compress, minimize, convert, append, offset, mitigate, inhibit, activate, deactivate, encode, encrypt, cipher, decipher, obfuscate, enhance, decode, transcode, adjust, update, restrict, compensate, etc. In particular, a feed-forward mechanism may be enabled.
[0163] Fig. 9An embodiment is shown in which the data OUT acts against the SW portion of the CSPUF. Such an embodiment may be advantageous, for example, if the CSPUF shows signs of aging; in this case, the parameter model used in the CSPUF may be changed to switch to a more robust or more reliable CRP pair.
[0164] Figures 10 to 15 Various embodiments of the invention according to different reactions are shown.
[0165] Fig.10 An embodiment is shown in which a challenge C is submitted that modifies or controls or affects the data OUT. For example, the data OUT may be the result of a test performed on C. If the test is successful (given predefined criteria), then R may be marked as acceptable. If the test is not confirmed (e.g., if C belongs to a domain or interval that is outside the nominal range or acceptable domain or interval for the challenge), then the response may be rejected or ignored. For example, if the challenge is unacceptable, this may mean that the CSPUF is attacked or otherwise explored / scanned, and in this case, a forged (or random) response may be provided.
[0166] Fig.11 An embodiment is shown in which data OUT may or may not affect R via an effect on C. C generates R via F. Data OUT may modify C (complement, filter, add, etc.) before, during, or after generating (b) R. For example, in a CMOS CSPUF, all pixels may be set to "white" via data IN.
[0167] Fig.12 Embodiments are shown in which data IN is applied to R (at the time of generating R, once R is generated, during, simultaneously with, or after generating R). For example, in a CMOS CSPUF, a third party or operator can validate or invalidate the response R being generated depending on their needs. The key R can be manipulated (e.g., split, supplemented, etc.) regardless of what OUT is.
[0168] Fig.13 Embodiments are shown in which the response R reacts to the data IN (at the time of generating R, upon generating R, during, simultaneously with, or after generating R). Such an embodiment has the advantage of enabling an "iterative" CSPUF ("auto-chaining"). In such an embodiment, R reacts to the data IN which modifies F, which in turn generates R when challenged with C. The sequence of responses R can be analyzed, modified, and re-injected in F. In the CMOS CSPUF example, such an embodiment allows convergence towards a subset of pixels (e.g., from the entire matrix to a stable subset), etc.
[0169] Fig.14An embodiment is shown in which the CPUF can be "self-looping" with independent channels IN and OUT. In this case, the independent channels IN and OUT constitute a data communication channel (e.g., wires, etc.). In this specification, the data communication channel will be designated by the IN and OUT signals that can be associated with the CSPUF.
[0170] Figures 16 to 18 An example of a combination of two CSPUFs is shown.
[0171] Such a "pair" of two CSPUFs can constitute a basic brick ("endpoint") for building a larger network. As can be seen, a series sequence or chain with a higher number of CSPUFs can be inferred from the proposed basic bricks. No parallel arrangement is shown. No further arrangements combining series and / or parallel arrangements are shown. However, diagrams and network topologies corresponding to different configurations can be determined. PUFs are valuable today due to their simplicity, low cost and low energy consumption. In further embodiments of the present invention, even though larger networks can lead to higher complexity, thereby increasing the presence of attacks, millions of CPUFs can be coupled or interconnected, providing a resilient and intelligent network.
[0172] With respect to the coupling of two CSPUFs, the communication may be local and / or remote (from short to long distances). For example, the first CSPUF may be in Paris, while the second CSPUF may be in New York. The two CSPUFs may be embedded in the same smartphone or car. The interaction may occur quickly, or may be performed at different time intervals over time.
[0173] Fig.15A possible coupling of two CSPUFs according to one embodiment of the present invention is shown. In such an embodiment, data can be sent in a network of linked CSPUFs. Data IN and data OUT of the corresponding devices are connected, for example, enabling a "master" or "edge" CSPUF 1 to control other CSPUFs. In fact, such an arrangement can be used to propagate or diffuse specific values (for example, data IN equals "4 Tesla") or command or indicate other instructions. In this case, each CSPUF can perform its physical unclonable function independently of each other, but there can still be a communication channel. In a specific sub-case, data IN can be equal to data OUT (for example, the expression "ISO=100" can be propagated in the network). Non-linear effects may also occur because data IN can modify F in the CSPUF. For example, if data IN is equal to "4 Tesla" in the first CSPUF1, data OUT of the first CSPUF1 can set data IN in the second CSPUF to be equal to "5 Tesla", or set data IN in the third CSPUF to be equal to "6 Tesla", and so on. In some embodiments, data OUT of CSPUF1 can be completely different from data IN 2 of CDSPUF2. In some embodiments, the data OUT of CSPUFn may be a linear function of CSPUF1.
[0174] Fig.16 One embodiment is shown in a "blockchain" configuration. In this arrangement, Ri (the response of the i-th device) is used to modify Fj via data INj (the data IN of the next device). This chain does in fact function as a blockchain: the first device outputs a first signature, which is used as a challenge C for the second device, which outputs a second signature, which in turn is used to determine a third signature, and so on. For independent devices A, B, and C, the chain ABC will be different from the ordered chain CAB. This feature may be advantageous if composition is important. Attacking one device may prove far from sufficient; as with Bitcoin, a hash among multiple hashes must be attacked.
[0175] In such an embodiment, Ri can be used, for example, to control the downstream chain. In some embodiments, [Ri ... Rj] can be XORed or manipulated (split, permuted, etc.). Use cases may be advantageous: for example, if the response or signature Ri is intercepted, demonstrated or proven to be wrong, the data INj can react by changing the S-PUF function (as previously described, the F of the first CMOS SPUF example can be obtained from the SPUF function). 1 "Switch" to the second CMOS example F 2 ).
[0176] Such embodiments combine the characteristics and properties of physically unclonable functions (physical functions) with the characteristics and properties of blockchains (trustless computation, distributed consensus, etc.).
[0177] Fig.17 An embodiment is shown in which the data OUT of the first CSPUF 1 is taken as a challenge C for CSPUF 2. In other words, the first CSPUF or node keeps its operation unchanged, where C1 is a given. The response is then "cascaded" through the neighbors of the first CSPUF. Such an embodiment can be advantageous. For example, in the CMOS CSPUF example, an interrupt or out-of-order pixel as data OUT in the i-th iteration can cause the Cj of the next node to change or refocus. If aging or reliability decreases, other CR pairs can be changed via C (but CSPUFj is independent of i). Such an embodiment can be used to enable backup circuits. Ri and Rj are independent and can be combined or considered in other ways.
[0178] Fig.18 An embodiment is shown in which the response R of the first CSPUF1 is taken as the challenge C for the next CSPUF2. Fig.14 A self-looping CSPUF is shown, but the embodiment represents a serial combination of CSPUFs. Such an embodiment depicts a chain of signatures, thereby implying uniqueness of multiple linked signatures. If there is a weak (or damaged or defective or aged or otherwise changed) node, the entire chain is compromised. There cannot be any substitution on a chain because each chain is, by definition, unclonable.
[0179] In some embodiments, data OUT of CSPUF1 may optionally be connected to data IN of CSPUF2. The communication channel is optional. In some embodiments, data INi and data OUTj may have their own rules or behaviors.
[0180] In one embodiment, the sensors may be selected to maximize reliability and / or entropy.
[0181] Depending on the embodiment, the sensor can be selected from the group consisting of: pressure sensor, ultrasonic sensor, humidity sensor, gas sensor, motion sensor, acceleration sensor or accelerometer, displacement sensor, force measurement sensor, gyroscope sensor or gyroscope, temperature sensor, image sensor, video sensor, UV sensor, magnetic sensor, CMOS image sensor, silicon microphone, inertial measurement unit (IMU), pressure sensor, micromirror, radio frequency sensor, magnetic field sensor, digital compass, oscillator, photometer or light sensor, proximity sensor, GNSS (e.g., GPS), barometer sensor, Wi-Fi sensor, Bluetooth sensor, NFC sensor, pedometer, pulse oximeter sensor, heart rate sensor and fingerprint sensor.
[0182] Some embodiments of the present invention may be entirely hardware embodiments. In other words, CSPUF components according to the present invention may be entirely hardware (e.g., embedded sensors and circuits). Some embodiments may contain both hardware elements and software elements. In other words, the CSPUF may combine hardware and software. The present invention may also take the form of a computer program product accessible from a computer-usable or computer-readable medium that provides program code for use by or in conjunction with a computer or any instruction execution system. Computer-usable or computer-readable may be any device that can contain, store, communicate, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or apparatus. The medium may be an electronic, magnetic, optical, electromagnetic, semiconductor system (or device or apparatus), or a propagation medium.
[0183] In space, some steps may be performed locally and / or remotely (e.g., in the smartphone itself, where some steps may be performed over a network or in the cloud). Local execution may involve (specific / dedicated and / or general) hardware blocks and / or software. The corresponding hardware circuits may be distributed within one or more computer devices, or aggregated in a particular circuit. In time, some steps may be performed in the past (e.g., previous steps, "offline", etc.) and / or at the current time (e.g., "online"). The hardware may be intangible (i.e., does not evolve over time (at least not intentionally)) or may be reprogrammed.
Claims
1. A PUF circuit comprising one or more sensors, the PUF circuit comprising a processor configured to receive information about a subset of sensor components of the sensor, the PUF circuit being configured to apply one or more challenges to the subset of sensor components under consistent conditions, the PUF circuit comprising a receiver configured to receive output signal values from the subset of sensor components in response to the challenges applied to the subset of sensor components, in, The PUF circuit (100) comprises a data communication channel, and the PUF circuit is configured to: - determining one or more statistical properties of a noise distribution of said output signal values; - receiving (131) input data IN from one or more external data sources using said data communication channel; - determining a response to said challenge based on said statistical properties and selected input data IN received from said one or more external data sources, said response being a digital signature, The input data IN is used to process the PUF circuit.
2. The PUF circuit according to claim 1, wherein: The PUF circuit is configured to transmit data OUT (132) to one or more external entities using the data communication channel.
3. The PUF circuit according to claim 1, wherein: The received data IN calibrates the one or more sensors.
4. A PUF circuit according to any one of the preceding claims, wherein: The received data IN includes information that directly or indirectly results in disabling the PUF circuit.
5. The PUF circuit according to any one of claims 1 to 3, wherein: The received data IN or a portion thereof is used as or contributes to a challenge to the PUF circuit (100), and the PUF circuit (100) provides a response in response to the one or more challenges.
6. The circuit according to any one of claims 1 to 3, wherein: The received data IN or a part thereof signs a challenge of the PUF circuit (100).
7. The PUF circuit according to claim 2, wherein: The at least one external entity comprises at least one computer or another PUF circuit.
8. The PUF circuit according to any one of claims 1 to 3, wherein: The applied challenge C modifies and / or controls and / or influences the transmitted data OUT.
9. The PUF circuit according to any one of claims 1 to 3, wherein: The transmitted data OUT modifies the applied challenge C.
10. The PUF circuit according to any one of claims 1 to 3, wherein: The response R is modified by the data IN either simultaneously with or after it is generated.
11. The PUF circuit according to any one of claims 1 to 3, wherein: The response R reacts to the data IN.
12. The PUF circuit according to any one of claims 1 to 3, wherein: The response R modifies the challenge C.
13. A system comprising two or more PUF circuits (100) according to any one of the preceding claims, the PUF circuits comprising a first PUF circuit and a second PUF circuit, wherein: The second circuit takes the data OUT transmitted by the first circuit as input data IN.
14. A system comprising two or more PUF circuits (100) according to any one of the preceding claims 1-12, the PUF circuits comprising a first PUF circuit and a second PUF circuit, wherein: The second circuit takes the response R of the first circuit as input data IN.
15. A system comprising two or more PUF circuits (100) according to any one of the preceding claims 1-12, the PUF circuits comprising a first PUF circuit and a second PUF circuit, wherein: The second circuit takes the data OUT transmitted by the first circuit as a challenge C.
16. A system comprising two or more PUF circuits (100) according to any one of the preceding claims 1-12, the PUF circuits comprising a first PUF circuit and a second PUF circuit, wherein: The second circuit takes the response R of the first circuit as a challenge C.
17. A method for generating a digital signature, the method being implemented by a PUF circuit including one or more sensors, the method comprising: receiving information about a subset of sensor components of the sensor; applying one or more challenges to a subset of the sensor components under consistent conditions; and receiving output signal values from the subset of sensor assemblies in response to a challenge applied to the subset of sensor assemblies, Wherein, the method comprises: - determining one or more statistical properties of a noise distribution of said output signal values; - receiving (131) input data IN from one or more external data sources using a data communication channel; - determining a response to said challenge based on said statistical properties and selected input data IN received from said one or more external data sources, said response being a digital signature, The input data IN is used to process the PUF circuit.
18. A computer program product comprising instructions for implementing the steps of the method according to claim 17 when said computer program is executed on a computer device.
Citation Information
Patent Citations
Synthetic physically unclonable function
EP3340214A1
Enrollment of physically unclonable functions
US8694778B2
Proofs of vicinity using cpufs
WO2007063475A2
Identification of devices using physically unclonable functions
WO2009024913A2
