Authentication based on a challenge and response using a physically unclonable function and a machine learning engine
View Patent ↗The present disclosure relates to implementations of physically unclonable functions (PUFs) for cryptographic and authentication purposes. Specifically, the disclosure describes implementations of machine learning engines (MLEs) in conjunction with PUFs generating outputs having multiple states.
1. A method of authentication, comprising the steps of:
receiving by a machine learning engine a challenge from a secure server, wherein the challenge comprises a first plurality of states;
generating by a physically unclonable function (PUF) a response, wherein the response comprises a second plurality of states;
generating outputs from the PUF comprising physical parameters that determine the operation of the PUF, wherein each output is a measurement of a cell in a plurality of cells in the PUF;
creating a multi-state output of the PUF by sorting the outputs from the PUF into a plurality of different states arranged to capture a profile of the physical parameters of the PUF;
communicating the response and the multi-state output from the PUF to the machine learning engine; and
processing by the machine learning engine the challenge, the response and the multi-state output of the PUF to generate a positive authentication or a negative authentication.
2. The method of claim 1 , wherein the multi-state output of the PUF was created and stored in a multi-state memory when the challenge was generated.
3. The method of claim 1 , wherein the multi-state output of the PUF was created when the response was generated.
4. The method of claim 1 , wherein each output, in the generating outputs from the PUF, is a measurement of a cell in the PUF comprises the steps of:
sweeping an input voltage to each cell in the plurality of cells in the PUF;
using the sensing element to measure the input voltage for each cell in the plurality of cells in the PUF when each cell switches from storing a logical value of “0” to a logical value of “1;” and
quantizing and mapping the input voltage for each cell in the plurality of cells in the PUF to create a multi-state output from the PUF.
5. The method of claim 1 , further comprising the steps of:
sensing a first parameter of a surroundings of the PUF when the challenge was generated;
sensing a second parameter of the surroundings of the PUF when the response was generated; and
processing by a processor the challenge, the response, the multi-state output of the PUF, the first parameter of the surroundings of the PUF when the challenge was generated and the second parameter of the surroundings of the PUF when the response was generated to generate the positive authentication.
6. The method of claim 5 , wherein the first parameter is a first temperature measurement of the surroundings of the PUF when the challenge was generated and the second parameter is a second temperature measurement of the surroundings of the PUF when the response was generated.
7. A system of authentication, comprising: a machine learning engine comprising a processor and a physically unclonable function (PUF) comprising an array of memory cells, wherein the machine learning engine and the PUF are configured to:
receive by the machine learning engine a challenge from a secure server, wherein the challenge comprises a first plurality of states;
generate by the PUF a response, wherein the response comprises a second plurality of states;
generate outputs from the PUF comprising physical parameters that determine the operation of the PUF, wherein each output is a measurement of a cell in a plurality of cells in the PUF;
create a multi-state output of the PUF by sorting the outputs from the PUF into a plurality of different states arranged to capture a profile of the physical parameters of the PUF;
communicate the response and the multi-state output from the PUF to the machine learning engine; and
process by the machine learning engine the challenge, the response and the multi-state output of the PUF to generate a positive authentication or a negative authentication.
8. The system of claim 7 , wherein the multi-state output of the PUF was created and stored in a multi-state memory when the challenge was generated.
9. The system of claim 7 , wherein the multi-state output of the PUF was created when the response was generated.
10. The system of claim 7 , wherein each output, in the generate outputs from the PUF, is a measurement of a cell in the PUF, wherein the machine learning engine and the PUF are further configured to:
sweep an input voltage to each cell in the plurality of cells in the PUF;
use the sensing element to measure the input voltage for each cell in the plurality of cells in the PUF when each cell switches from storing a logical value of “0” to a logical value of “1;” and
quantize and map the input voltage for each cell in the plurality of cells in the PUF to create a multi-state output from the PUF.
11. The system of claim 7 , wherein the machine learning engine and the PUF are further configured to:
sense a first parameter of a surroundings of the PUF when the challenge was generated;
sense a second parameter of the surroundings of the PUF when the response was generated; and
process by a processor the challenge, the response, the multi-state output of the PUF, the first parameter of the surroundings of the PUF when the challenge was generated and the second parameter of the surroundings of the PUF when the response was generated to generate the positive authentication.
12. The system of claim 11 , wherein the first parameter is a first temperature measurement of the surroundings of the PUF when the challenge was generated and the second parameter is a second temperature measurement of the surroundings of the PUF when the response was generated.