Classifications

• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M1/00—Analogue/digital conversion; Digital/analogue conversion
  • H03M1/001—Analogue/digital/analogue conversion
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
  • G06F7/58—Random or pseudo-random number generators
  • G06F7/588—Random number generators, i.e. based on natural stochastic processes
• • G—PHYSICS
  • G07—CHECKING-DEVICES
  • G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
  • G07C15/00—Generating random numbers; Lottery apparatus
• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M1/00—Analogue/digital conversion; Digital/analogue conversion
  • H03M1/12—Analogue/digital converters
  • H03M1/48—Servo-type converters
• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M1/00—Analogue/digital conversion; Digital/analogue conversion
• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M1/00—Analogue/digital conversion; Digital/analogue conversion
  • H03M1/12—Analogue/digital converters

Definitions

• This relates to a computation system including a digital entropy signal generation apparatus and method.
• entropy is randomness that may be generated by hardware or software and is typically output in a sequence of random numbers or symbols.
• Such entropy has various uses, such as in statistical sampling, computer simulation, and securing data from interception or theft, including by cryptography, by ways of example.
• Cryptography typically includes one or more keys that must be randomly generated, where the keys are used to encrypt data when transmitted and decrypt the encrypted data when received, and the greater the randomness associated with the key(s) the more difficult unauthorized third party decryption ⁇ i.e., without the authorized key) becomes.
• randomness may use used for other inputs, such as for generating digital signatures or challenges in authentication protocols.
• NIST National Institute of Standards and Technology
• processors such as a microprocessor, digital signal processor (DSP), microcontroller unit (MCU) or other comparable integrated circuit device.
• DSP digital signal processor
• MCU microcontroller unit
• typical considerations include cost in terms of area and power consumed on or by the device, where such considerations may impede or prohibit the use of an integrated random number generator.
• processor clock speed may be incompatible with the clock speeds necessary to generate random numbers according to conventional approaches.
• the system comprises an input for receiving an analog signal comprising a frequency and noise, that noise including input referred noise, and the noise fluctuates in a range.
• the system also comprises a signal path.
• the signal path comprises: (a) an analog to digital converter for providing a digital output value in response to a clock period; (b) a feedback node; and (c) circuitry for limiting a signal swing at the feedback node, during a period of the clock period, to be no greater than an RMS value of the noise.
• the analog to digital converter is further for providing the digital output value in response to the analog signal and the signal swing at the feedback node.
• FIG. 1 illustrates an electrical block diagram of a network system 10, according to a preferred embodiment.
• FIG. 2 details two network elements NE A and NE B that may be representative of any of the network elements of FIG. 1.
• FIG. 3 illustrates a functional block diagram of random number generator RNGA.
• FIG. 4 illustrates a functional block diagram of random number generator RNGA, with additional schematic representations now shown in certain blocks.
• FIG. 5 illustrates a functional block diagram of random number generator RNGA, with additional schematic representations shown for a DAC circuit 26' as an alternative to DAC circuit 26 shown in FIG. 4.
• FIG. 6 again illustrates a functional block diagram of an alternative random number generator RNGA I relative to that of FIG. 3, with the alternative presenting a reversal of the order of devices in the feedback path.
• FIG. 7 again illustrates a functional block diagram of an alternative random number generator RNGA 2 relative to that of FIG. 3, with the alternative presenting a different clocking arrangement and a flip flop 26 ⁇ .
• FIG. 1 illustrates an electrical block diagram of a network system 10, according to one example of a preferred embodiment.
• System 10 includes a local area network (LAN) shown generally above a horizontal dashed line in the figure, and the LAN includes a number of conventional items.
• the LAN also is connected via an Internet Service Provider (ISP) to the Internet, which itself represents and/or is connected to a network of additional networks, including the elements connected to those networks.
• ISP Internet Service Provider
• the LAN has a physical interface 12, such as a multi-conductor cable, to which are connected a number N of network elements shown as NEi . i, NEi .2 , . . ., NE IJV .
• Each of these (and other) network elements referred to herein is a logical entity that communicates onto a telecommunication or computing system network; such elements, therefore, may be computers, tablets, personal devices, and modules or the like that are anticipated to be prolific in the development of the Internet-of-Things (IoT). Such devices may be constructed of myriad forms of hardware and software (including firmware and the like), with at least the minimal functionality to communicate to another device in the LAN. Further, the network elements may include various other computational or feature functionality. Also connected to physical interface 12 is a router 14, which is further coupled to (or includes) a wireless adapter 16. Wireless adapter 16 is operable to bi-directionally communicate wirelessly with a number M of network elements shown as NE 2.
• Router 14 controls the forwarding of data packets among the network elements in the LAN and between the LAN and the ISP. In all events, any of the network elements of FIG. 1 are operable to communicate with one another, or with what is generally shown as OTHER NETWORK(s)) that have compatible network elements, where such other networks are remotely located but likewise may communicate via the Internet.
• FIG. 2 further details two network elements NEA and NE B , which may be representative of any of the network elements of FIG. 1 (or of elements on the OTHER NETWORK(s).
• Network element NEA includes a processor ⁇ e.g., microcontroller) MCUA, which includes a random number generator RNGA.
• random number generator RNGA is operable to provide an entropy based random number, thereby providing, or from which may be determined, an encryption key KEYA. Both KEYA and DATA are connected to an encryption process also performed as part of processor MCUA. The encryption process may be achieved by any of various conventional techniques, and the result is encrypted data DATA ENC .
• This encrypted data DATA EN C is communicated from network element EA to network element E B , via any of the various communication manners described hereinabove.
• the encrypted data DATA EN C is thus received by network element E B and processed by its respective processor MCU B , which includes a decryption process that receives the encrypted data DATA E NC and a data decryption key KEY B .
• Data decryption key KEY B may be symmetric or asymmetric with respect to the decryption key KEY A used by network element E A , with the appropriate protocol or the like establishing the choice so that by design the data is secure when communicated between the network elements, so as to deter or eliminate the possibility of third party unauthorized detection of the original data DATA processed by network element NEA.
• FIG. 2 illustrates encryption/decryption in a path from network element NEA to network element NE B
• the reverse path i.e., encryption/decryption in a path from network element NE B to network element NE A
• the reverse path is likewise possible.
• FIG. 3 illustrates a functional block diagram of a preferred embodiment random number generator RNGA, with additional schematic representations as to certain aspects provided later.
• Generator RNGA includes a physical entropy source 20 that provides an analog noise signal N PE s, with a randomly fluctuating magnitude.
• Source 20 may be one of various different forms.
• source 20 may be thermal noise, which is the electronic noise generated by charge carriers through a conductor; hence, as described hereinbelow, source 20 may be a terminal of a resistor, whereby electronic noise is generated at that terminal by current passing through the resistor (commonly referred to as shot noise).
• the voltage fluctuation generated by entropy source 20 will be in order of few microvolts and thereby representing a relatively low input signal swing, such as compared to a typical input signal of an analog-to-digital conversion system.
• Random number generator RNGA also includes a comparator 22, as may be implemented by various forms so as to compare two different signals and provide a digital output of high or low based on the relative amplitude of the compared signals, thereby accomplishing in part an analog-to- digital conversion based on the comparison inputs.
• comparator 22 has two inputs shown as IN 1 and IN 2 , where input IN 1 is connected to receive the analog noise signal N PE S from source 20, and input IN 2 is connected to receive a feedback analog signal S FB from a limited signal swing node 24, as described hereinbelow.
• a sample clock CLK S is also connected to comparator 22, and in response to a same edge (e.g., rising edge) of each cycle of sample clock CLKs, comparator 22 compares N PE S at input INi to S FB at input IN 2 and, as shown in the Figure by the use of (+) and (-) conventions, comparator 22 outputs a binary digital signal (e.g., high) if N PE S > S FB or comparator 22 outputs a complementary binary digital signal (e.g., low) if N PE s ⁇ S FB - For successive cycles of the sample clock CLKs, therefore, comparator 22 outputs successive binary values, which in succession thereby in total provide the digital random numeric sequence DRNS, that is, the output of random number generator RNGA.
• comparator 22 is shown as comparing a positive and negative input, but in an alternative embodiment signals N PE s and S FB may be input to a differential circuit with the difference output to a single input comparator, where that single input comparator then compares the difference to a reference, again outputting either one of two digital states based on the result of the comparison of the difference to the reference.
• the digital random numeric sequence DRNS is both an output of generator RNGA and is also fed back into a feedback path that eventually provides the feedback analog signal S FB at limited signal swing node 24.
• the digital random numeric sequence DRNS is connected in the feedback path as an input to a digital-to-analog (DAC) circuit 26.
• DAC circuit 26 is constructed to convert its digital input to an analog output signal A DA c- Such conversion, and structures to achieve it, are known in the DAC (and ADC) art.
• the particular devices chosen to implemented DAC circuit 26 may be relatively simple, given that the precision of the conversion is less critical in that the ultimate output sequence DRNS of RNGA is a random sequence. Accordingly, in various other circuit devices, often the precision and corresponding resolution of a DAC is chosen to reduce potential errors in the conversion of data, such as quantization errors; in the preferred embodiment, and by contrast, the introduction of such noise is favorable, because it further randomizes the output signal, which is in part driven by the randomness of analog noise signal N PE S- [0018]
• the analog output signal A DA c from DAC circuit 26 is input to an analog signal integrator 28.
• quantization noise reduction techniques for DAC resolution or frequency shifting typically associated with modulators teach away from what is implemented in the preferred embodiment which, as described in various examples hereinbelow, permits the general function of digital-to-analog and integrative functions to be controlled and timed in a matter that increases, rather than reduces, the presence (or addition) of noise in the signal path.
• integrator 28 its output is connected to a limited signal swing node 24 described hereinabove.
• Node 24 is referred to as having a "limited signal swing” for reasons now further defined.
• various aspects of a circuit including components and even parameters of conductors, each contribute noise to the overall circuit or system. Under circuit analysis, the collective of such noise can be determined or approximated and referred back to the circuit input, and is usually referred to as "input referred noise,” that is, the total measure of noise as would affect an input signal.
• step size of the device established to be considerably larger than the input referred noise, where step size is the amount of signal swing that can occur in the feedback path signal during one sample period of the modulator; this design criteria allows the signal swing in a clock cycle to dominate over any noise, so as to allow the modulator to properly track or approximate the modulator forward path input signal.
• DAC 26 and integrator 28 are established so as to limit the signal swing at node 24 to be equal to or less than the RMS value of the input referred noise of generator RNGA, where that noise includes any noise effects from devices 20, 22, 26, 28, and the interconnects between them.
• node 24 is intentionally designed to swing, during any period of CLKs, a limited amount so that the noise is actually able to dominate the signal at node 24, rather than just the opposite as would be the case in a conventional approach.
• the range (e.g., between the points of three standard deviations from the center of a Gaussian distribution) of input referred noise is generally known or ascertainable, where for sake of an example, consistent with a schematic provided later, assume that such a range is between -100 ⁇ and +100 ⁇ .
• the total signal swing within this input signal range is 200 ⁇ .
• the value of NPES is 'x(O)' and the value of SFB is 'y(O)' .
• Comparator 22 produces an output based on x(0) and y(0). Let us assume x(0) >y(0) and so comparator 22 produces an output of 'high' . When this 'high' is integrated in integrator 28 it raises the voltage on SFB at node 24) to a new value 'y(l)' . Now when comparator 22 triggers the next time in response to clock CLKs, the input random noise value changes to x(l) and comparator 22 makes a new decision based on x(l) and y(l). As the system is in negative feedback the integrator 28 output SFB always attempts to reach the input signal N PE s- Comparator 22 digitizes the residual error signal (x(t)-y(t)). We know x(t) is a true random signal. So as long as x(t) is comparable to y(t), the difference will be random and digital equivalent of it (DRNS) will be also be random.
• DRNS digital equivalent of it
• FIG. 4 again illustrates a functional block diagram of random number generator RNGA, with additional schematic representations now shown in certain blocks.
• physical entropy source 20 is formed by a resistor divider that includes a resistor Rl in series with a resistor R2, with the series resistance connected between a fixed voltage source VDD and ground.
• the resistance of resistors Rl and R2 are equal, but alternative resistances may be used. In any event, given the potential across the resistor pair, a current flows, and a node 20 N between the resistors provides an output signal.
• the voltage at node 20 N could be expected as fixed at a DC level, fixed DC voltage, but (in a more detailed analysis) current flow through a conductor gives rise to analog noise signal in addition to the DC level that occurs due to the voltage division as between the resistors.
• VDD and the resistances of resistors Rl and R2 are preferably selected so that the analog fluctuation or noise swings between approximately -100 ⁇ and +100 ⁇ .
• such an input value would usually be far below the resolution of a conventional delta modulator, so as not to be sufficient to cause a transition in such a device.
• this value may be possible to quantize.
• Comparator 22 is not shown in FIG. 4, because it may be readily constructed according to conventional principles, particularly given the context of the teachings of this description.
• each of inputs INi and IN 2 may be connected to a respective transistor gate in a differential pair of connected transistors, whereby one or the other of the transistors is enabled at a time based on the relative potentials at those inputs and the result is gated by the frequency of sample clock CLKs.
• the frequency of sample clock CLKs is less than the frequency bandwidth of the input signal, as again the preferred embodiment endeavors to further randomize the input rather than to model it with a digital value that tracks it.
• the bandwidth of N PE s might be 1 MHz, while sample clock CLK is 1 kHz, such that one or more orders of magnitude exist between the two (e.g., three orders of magnitude as between 1 MHz and 1 kHz, whereas two orders of magnitude may suffice in some preferred embodiments).
• amplifier circuitry may be included in comparator 22, so as to reduce input offset. Additional circuitry may be included so as to ensure that, in response to the differential input, the output is either fully high or fully low, so as to provide a digital output signal (i.e., logic high or logic low). Further, the digital signal is driven to amplitudes for other circuitry using the sequence, such as use in connection with cryptography as described hereinabove.
• an example logic high voltage may be 1.0 volt. Therefore, as described hereinabove, the digital output is therefore a single binary value at a time, which over successive values provides digital random numeric sequence DRNS. Accordingly, for a relatively small input signal with an order of magnitude of 100 ⁇ , a considerable dynamic range is realized in the output of 1 volt. For example, this may be contrasted to costly and complex conventional modulators where the output signal amplitude has a high correlation to the amplitude of the input signal.
• FIG. 4 further illustrates schematic details for one preferred embodiment of DAC circuit 26.
• an analog buffer 26 B that receives the digital output from comparator 22 and, in response, outputs an analog voltage signal.
• it is assumed as a rail signal of either a high rail analog signal or a low rail analog signal, based on a respective digital input of 1 or 0.
• the output of buffer 26 B is connected through a resistor R3, which from the analog voltage of buffer 26 B causes a current I D AC to flow. This current is provided to integrator 24.
• FIG. 4 further illustrates schematic details for one preferred embodiment of analog signal integrator 28.
• integrator 28 may elegantly and without complexity include a single capacitor CI, thereby minimizing design considerations such as area and power consumed by the device.
• current I D AC is connected to limited signal swing node 24, and capacitor CI is connected between node 24 and ground.
• capacitor CI and resistor R3 of DAC circuit 26
• the respective capacitance and resistance values of these devices is selected, in combination with the frequency of sample clock CLKs, so as to limit the step size at node 24 described hereinabove, which is the amount of voltage swing that can occur at node 24 during any single period of sample clock CLK S .
• the step size at node 24 N is 150 ⁇ .
• This step size may be considered an effective resolution of the combination of DAC 26 and integrator 28, that is, the amount of voltage change that may be represented in a single cycle of sample clock CLKs.
• this combination does have a very high resolution as would be expected in a conventional delta modulator that endeavors to accurately quantize considerably larger input analog signals while effectively reducing (or shifting to a different frequency) quantization noise.
• the preferred embodiment is not required to accurately quantize a large (or any) input, but instead is dithering the DAC 26 voltage with the already-random input signal (e.g., thermal noise) from source 20, so as to produce further randomization into the eventual RNG output signal, DRNS.
• the already-random input signal e.g., thermal noise
• the charge stored at node 24 (by capacitor CI) will be in some sense related to previous output signals as fed back, but the step size at that node is less than potential swings in NPES; thus, for a given cycle of sample clock CLK S , the amount of signal swing input to input IN 1 of comparator 22 can exceed the limited step size at input IN 2 (i.e., from node 24), whereby an accurate analog-to-digital conversion is not achieved, but instead the already-randomized input is further randomized by a feedback signal that is influenced by, but not equal to, a previous output and fed back signal.
• FIG. 5 again illustrates a functional block diagram of random number generator RNGA, with additional schematic representations shown for a DAC circuit 26' as an alternative to DAC circuit 26 shown in FIG. 3.
• DAC 26' again receives the binary output signal from comparator 22, and that signal provides a control signal CTRLi to a switch SWi, and it also is connected through an inverter 30 IN which, therefore, outputs provide a control signal CTRL 2 to a switch SW 2 .
• the switch is closed when its respective control signal is high, and the switch is open when its respective control signal is low.
• a first current source ISi is connected from VDD to the pole of switch Si, and the throw of switch Si is connected to node 24.
• the pole of switch S 2 is connected to node 24, while a second current source IS 2 is connected from the throw of switch S 2 to ground. Therefore, given the above-described intended control, when the output of comparator 22 is a binary high value, switch SWi closes and switch SW 2 opens, so that current source ISi charges node 24 N . In opposite fashion, when the output of comparator 22 is a binary low value, switch SWi opens and switch SW 2 closes, so that current source IS 2 sinks charge from node 24 N .
• the capacitance of capacitor CI, connected to node 24, may be selected again so as to limit the step size, per clock period of CLKs, of voltage swing at node 24 N , consistent with the teachings of this description for further randomizing the signal path of generator RNGA.
• FIG. 6 again illustrates a functional block diagram of an alternative random number generator RNGA I relative to that of FIG. 3, with the alternative presenting a reversal of the order of devices in the feedback path.
• random number generator RNGA I the output of comparator 22 is connected as an input to a digital signal integrator 30.
• the output of digital signal integrator 30 is preferably a multibit digital signal that is connected as an input to a DAC 32, which provides an output to node 24 and thereby again provides the limited signal swing that is connected to the negative input of comparator 22.
• digital integrator 30 operates as a counter that increments when the output of comparator 22 is "high” and decrements when the output of comparator 22 is “low.”
• the number of bits in integrator 30 is matched with the resolution of DAC 32.
• DAC 32 is a multibit DAC that converts the output of digital integrator 30 into an analog signal.
• the least significant bit (LSB) size of DAC 32 is again chosen so as to limit the voltage swing at node 24 for a given cycle of clock CLKS, that is, to be equal to or less than the RMS value of the input referred noise at the positive input to comparator 22.
• the maximum and minimum values of DAC are designed to include several (e.g., three) standard deviations of the RMS input referred noise input to comparator 22.
• FIG. 7 illustrates a functional block diagram of an alternative random number generator RNGA 2 relative to that of FIG. 3, with the alternative presenting a different clocking arrangement and a flip flop 26.
• the clock signal to comparator 22 is shown as a first sample clock CLK S i, while a second sample clock CLK S2 , operating at a frequency differing and preferably slower than, CLKsi, clocks a flip flop 26EF that has its data input connected to the output of comparator 22, while the output of flip flop 26EF provides the digital random numeric sequence DRNS.
• comparators are usually operable to perform a comparison at a speed that is faster than is specified, or alternatively for additional design operational assurance a comparator is usually clocked by a system clock at a frequency slower than the actual comparative speed of the device.
• a comparator such as comparator 22 may include internal signaling to indicate when each comparison it achieves is complete, while the overall system is synchronized to a system clock.
• random number generator RNGA 2 therefore, the internal signaling of comparator 22 is used to trigger first sample clock CLKsi, that is, the output of comparator 22 is sampled as soon as the comparison function of comparator 22 is complete.
• flip flop 26FF is clocked by sample clock CLKS 2 , which itself can be implemented as the system clock. Given the relative speeds of sample clocks CLKSi and CLKS 2 , and further in view of the lack of synchronization between the two, additional randomness is therefore imparted as between the feedback path and ultimate output value for generator RNG 42 , as compared to preferred embodiments described hereinabove.
• the preferred embodiments provide a computation system for creating a digital entropy signal.
• the preferred embodiments provide numerous benefits.
• the preferred embodiments include a robust random signal available to preferred embodiment or other circuits, systems, and processes, such as for securing data through encryption or the like.
• the preferred embodiments are relatively straightforward to implement and accomplish a desirable random signal with relative low cost and low power consumption. These results are achievable despite the otherwise difficulty in the art of achieving metastability in faster device nodes.
• a preferred embodiment may include the teachings herein with additional functionality integrated into a whole, such as a system level processor, including a microprocessor, DSP, or MCU.
• the preferred embodiment has been shown with a signal path that includes a feed forward path and a feedback path, with analog-to-digital (e.g., comparator) and digital-to-analog conversion in those paths, along with a controlled node so as introduce additional randomness into the conversion of the data in the signal path. While particular examples have been shown of location of these aspects along the path, various aspects may be reordered or located in different locations of the total signal path thereby producing still other preferred embodiments.
• analog-to-digital e.g., comparator

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Computational Mathematics (AREA)
• Mathematical Analysis (AREA)
• Mathematical Optimization (AREA)
• Pure & Applied Mathematics (AREA)
• General Engineering & Computer Science (AREA)
• Analogue/Digital Conversion (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)

Priority Applications (5)

Applications Claiming Priority (2)

Publications (1)

Family

ID=59929295

Family Applications (1)

Country Status (6)

Families Citing this family (4)

Citations (3)

Family Cites Families (20)

• 2016
  • 2016-11-01 US US15/339,931 patent/US9780798B1/en active Active
• 2017
  • 2017-08-31 US US15/691,827 patent/US10491228B2/en active Active
  • 2017-11-01 KR KR1020197012219A patent/KR102585834B1/ko active Active
  • 2017-11-01 JP JP2019523718A patent/JP6940737B2/ja active Active
  • 2017-11-01 WO PCT/US2017/059527 patent/WO2018085396A1/en not_active Ceased
  • 2017-11-01 CN CN201780067311.1A patent/CN109891750B/zh active Active
  • 2017-11-01 CN CN202310746884.4A patent/CN116719508A/zh active Pending
  • 2017-11-01 EP EP17867386.9A patent/EP3535848A4/en active Pending
• 2021
  • 2021-08-10 JP JP2021130790A patent/JP7256504B2/ja active Active

Patent Citations (3)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events