Random numbers generation using continuous-time chaos
View Patent ↗Novel random number generation methods and random number generators (RNG)s based on continuous-time chaotic oscillators are presented. Offset and frequency compensation loops are added to maximize the statistical quality of the output sequence and to be robust against parameter variations and attacks. We have verified both numerically and experimentally that, when the one-dimensional section was divided into regions according to distribution, the generated bit streams passed the tests used in both the FIPS-140-2 and the NIST 800-22 statistical test suites without post processing. Numerical and experimental results presented in this innovation not only verify the feasibility of the proposed circuits, but also encourage their use as the core of a high-performance IC RNG as well. In comparison with RNGs based on discrete-time chaotic maps, amplification of a noise source and jittered oscillator sampling, it is seen that RNGs based on continuous-time chaotic oscillators can offer much higher and constant data rates without post-processing. In conclusion, we can deduce that the proposed circuits can be realized in integrated circuits and the use of continuous-time chaos with the proposed innovations is very promising in generating random numbers with very high throughput.
1. A method for generating binary random bits (S (xor)i ) regionally according to distribution based on a non-autonomous, continuous-time chaotic oscillator, comprising the step of:
generating non-invertible random binary bits regionally according to distribution of one state which corresponds to a waveform of the continuous-time chaotic oscillator, the generating further comprising steps of:
a. determining non-invertible appropriate sections where the distribution of samples has two regions, by considering only the samples belonging to a state x 1 by way of:
i. determining appropriate parameters set of normalized quantities, where the samples of the state x from a one dimensional section obtained at a status transition of another state (x 2 , x 3 , . . . , or x n ) defined as x 2 . . . n (t)=x 2 . . . n (0) with dx 2 . . . n /dt>0 or dx 2 . . . n /dt<0, has two regions or,
ii. adjusting appropriate x 2 . . . n (0) values which corresponds to v 2 . . . n (0), where the samples of the state x 1 from the one dimensional section obtained at the status transition of another state (x 2 , x 3 , . . . , or x n ) defined as x 2 . . . n (t)=x 2 . . . n (0) with dx 2 . . . n /dt>0 or dx 2 . . . n /dt<0, has two regions or,
iii. determining appropriate parameters set of normalized quantities, where periodic samples of the state x 1 obtained at rising or falling edges of an external periodical pulse signal, that is at times t satisfying ωtmod2π=t o (ω is the frequency of the pulse signal), has two regions or,
iv. adjusting appropriate t o , where the periodic samples of the state x 1 obtained at the rising or falling edges of an external periodical pulse signal that is at times t satisfying ωtmod2π=t o (ω is the frequency of the pulse signal), has two regions or,
v. determining appropriate parameters set of normalized quantities, where the samples of the states x 1 from the one dimensional section obtained at the rising or falling edges of a periodical pulse signal used to drive the non-autonomous chaotic oscillator (at times t satisfying ωtmod2π=t o where ω is the frequency of the pulse signal and 0<t o <1 has two regions or,
vi. adjusting appropriate t o , where samples of the states x 1 from the one dimensional section obtained at the rising or falling edges of a periodical pulse signal used to drive the non-autonomous chaotic oscillator (at times t satisfying ωtmod2π=t o where ω is the frequency of the pulse signal and 0<t o <1) has two regions;
b. generating binary S (top)i and S (bottom)i from regional x 1i values obtained from the appropriate section defined in step a above for regional thresholds according to the following equation:
S (top)i =sgn ( x 1i −q top ) when x 1i ≧q middle
S (bottom)i =sgn ( x 1i −q bottom ) when x 1i <q middle
where sgn(.) is the signum function, q top and q bottom are the thresholds for top and bottom distributions respectively and wherein q top and q bottom are chosen as thresholds because they are determined to be the medians of the top and bottom distributions, and q middle is the boundary between the distributions;
c. realizing offset compensations for q top and q bottom thresholds defined in step b above by implementing a Monobit Test;
d. realizing frequency compensation for the sampling frequency of x 1 by implementing a Runs Test to avoid oversampling of x 1 ;
e. generating random binary data S (xor)i using the binary sequences S (top)i and S (bottom)i defined in step b above according to the following equation:
S (xor)i =S (top)i (XOR) S (bottom)i
where exclusive-or (XOR) operation is exploited to eliminate the bias in order not to decrease the throughput.
2. A method according to claim 1 , wherein: said state x 1 can also be another state x 2 , x 3 . . . or x n .
3. A method according to claim 1 , wherein: said Monobit Test is a monobit test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
4. A method according to claim 1 , wherein: said Runs Test is a runs test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
5. A method according to claim 1 for generating binary random bits (S (xor)i ) regionally according to distribution based on an autonomous continuous-time chaotic oscillator, comprising the steps of:
a. determining non-invertible appropriate sections where the distribution of samples has two regions, by considering only the samples belonging to a state x 1 , by way of:
i. determining appropriate parameters set of normalized quantities, where the samples of the state x 1 from a one dimensional section obtained at a status transition of another state (x 2 , x 3 , . . . or x n ) defined as x 2 . . . n (t)=x 2 . . . n (0) with dx 2 . . . n /dt>0 or dx 2 . . . n /dt<0, has two regions or,
ii. adjusting appropriate x 2 . . . n (0) values which corresponds to v 2 . . . n (0), where the samples of the state x 1 from the one dimensional section obtained at the status transition of another state (x 2 , x 3 , . . . or x n ) defined as x 2 . . . n (t)=x 2 . . . n (0) with dx 2 . . . n /dt>0 or dx 2 . . . n /dt<0, has two regions or,
iii. determining appropriate parameters set of normalized quantities, where periodic samples of the state x 1 , x 2 , . . . x n , obtained at rising or falling edges of an external periodical pulse signal, that is at times t satisfying ωtmod2π=t o (ω is the frequency of the pulse signal), has two regions or,
iv. adjusting appropriate t o , where the periodic samples of the state x 1 , x 2 , . . . x n , obtained at the rising or falling edges of an external periodical pulse signal, that is at times t satisfying ωtmod2π=t o (ω is the frequency of the pulse signal), has two regions;
b. generating random binary sequences S (top)i and S (bottom)i from regional x 1i values obtained from the appropriate section defined in claim 3 . a for regional thresholds according to the following equation:
S (top)i =sgn ( x 1i −q top ) when x 1i ≧q middle
S (bottom)i =sgn ( x 1i −q bottom ) when x 1i <q middle
where sgn(.) is the signum function, q top and q bottom are the thresholds for top and bottom distributions, respectively, and wherein q top and q bottom are chosen as thresholds because the are determined to be the medians of the top and bottom distributions, and q middle is the boundary between the distributions;
c. realizing offset compensations for q top and q bottom thresholds defined in claim 3 . b by implementing a Monobit Test;
d. realizing frequency compensation for the sampling frequency of x 1 by implementing a Runs Test to avoid oversampling of x 1 ;
e. generating random binary data S (xor)i by using the binary sequences S (top)i and S (bottom)i defined in claim 3 . b above according to the following equation:
S (xor)i =S (top)i (XOR) S (bottom)i
where exclusive-or (XOR) operation is exploited to eliminate the bias in order not to decrease the throughput.
6. A method according to claim 5 , wherein: said state x 1 can also be another state x 2 , x 3 . . . or x n .
7. An apparatus comprising a random bit generator which is based on an autonomous or a non-autonomous, continuous-time chaotic oscillator and which relies on generating non-invertible random binary bits regionally according to distribution of a waveform of the continuous-time-chaotic oscillator, comprising:
a. three comparators in order to generate regional binary sequences S (top)i and S (bottom)i by using V top , V bottom , and V middle thresholds from the waveform of the continuous-time chaotic oscillator v 1 that has two regions, according to the following equation:
S (top)i =sgn ( v 1i −V top ) when v 1i ≧V middle
S (bottom)i =sgn ( v 1i −V bottom ) when v 1i <V middle
b. periodical pulse signal generator and two D flip-flops (D flip-flop) in order to sample regional binary S (top)i and S (bottom)i periodically;
c. two mono-bit test blocks (Monobit Test) and two digital-to-analog converters (DAC) to generate and compensate thresholds V top and V bottom used for top and bottom distributions, respectively;
d. a runs test block (Runs Test) and a prescaler block (Prescaler) to compensate the sampling frequency of v 1 ; and
e. an exclusive-or (XOR) gate to eliminate the bias and generate random binary data S (xor)i using the binary sequences S (top)i and S (bottom)i according to the following equation:
S (xor)i =S (top)i (XOR) S (bottom)i .
8. An apparatus according to claim 7 , wherein: said Monobit Test is a monobit test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
9. An apparatus according to claim 8 , wherein: said Runs Test is a runs test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
10. An apparatus according to claim 7 , wherein: said Runs Test is a runs test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
11. An apparatus according to claim 7 for generating binary random bits regionally according to distribution based on a non-autonomous, continuous-time chaotic oscillator, including:
a. three comparators in order to generate regional binary sequences S (top)i and S (bottom)i by using V top , V bottom , and V middle thresholds from the waveform of the non-autonomous chaotic oscillator v 1 that has two regions;
b. periodical pulse signal generator (v p (t)) which is used to drive the non-autonomous chaotic oscillator; a delay block (Delay) to adjust appropriate t o , where t o is a time inside a period of v p (t) and two D flip-flops (D flip-flop) in order to sample regional binary sequences S (top)i and S (bottom)i ;
c. two mono-bit test blocks (Monobit Test) and two digital-to-analog converters (DAC) to generate and compensate thresholds V top and V bottom used for top and bottom distributions, respectively;
d. a runs test block (Runs Test) and a prescaler block (Prescaler) to compensate the sampling frequency of v 1 ; and
e. an exclusive-or (XOR) gate to eliminate the bias and generate random binary data S (xor)i by using the binary sequences S (top)i and S (bottom)i .
12. An apparatus according to claim 11 , wherein: said Monobit Test is a monobit test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
13. An apparatus according to claim 12 , wherein: said Runs Test is a runs test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
14. An apparatus according to claim 11 , wherein: said Runs Test is a runs test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
15. An apparatus for generating binary random bits regionally according to distribution of a waveform of an autonomous or a non-autonomous, continuous-time chaotic oscillator, comprising:
a. three comparators in order to generate regional binary sequences S (top)i and S (bottom)i by using V top , V bottom , and V middle thresholds from the waveform of the continuous-time chaotic oscillator v 1 that has two regions, according to the following equation:
S (top)i =sgn ( v 1i −V top ) when v 1i ≧V middle
S (bottom)i =sgn ( v 1i −V bottom ) when v 1i <V middle
b. two D flip-flops (D flip-flop) and another comparator in order to sample regional binary sequences S (top)i and S (bottom)i from a one dimensional section of v 1 obtained at a status transition of another waveform (v 2 , v 3 , . . . v n ) defined as v 2 . . . n (t)=v 2 . . . n (0);
c. two mono-bit test blocks (Monobit Test) and two digital-to-analog converters (DAC) to generate and compensate thresholds V top and V bottom used for top and bottom distributions, respectively;
d. a runs test block (Runs Test) and a prescaler block (Prescaler) to compensate the sampling frequency of v 1 ; and
e. an exclusive-or (XOR) gate to eliminate the bias and generate random binary data S (xor)i by using the binary sequences S (top)i and S (bottom)i according to the following equation:
S (xor)i =S(top)i (XOR) S (bottom)i .
16. An apparatus according to claim 15 , wherein: said Monobit Test is a monobit test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
17. An apparatus according to claim 16 , wherein: said Runs Test is a runs test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.
18. An apparatus according to claim 15 , wherein: said Runs Test is a runs test selected from a Statistical Test Suite of FIPS-140-1, FIPS-140-2 or NIST 800-22.