Method and system for authentication of RF device
The method for authenticating a first device, the method including receiving RF signals from the first device, acquiring feature data from the received RF signals, fitting a model for authenticating the first device using the acquired feature data, storing in memory the fitted model in a second device, and, by the second device receiving a new RF signal from a device to authenticate, acquiring feature data from the new RF signal, and determining whether or not the device to authenticate is the first device by transmitting the feature data, acquired from the new RF signal, as input to the fitted model, the feature data acquired from a RF signal includes values of only one of an in-phase component and a quadrature component of the RF signal over time.
1 . A computer-implemented method for authentication of a first device, the method comprising:
A) receiving a plurality of RF signals from the first device;
B) acquiring feature data from the received RF signals;
C) fitting a model for authenticating said first device using the acquired feature data;
D) storing in memory the fitted model in a second device;
E) by the second device, receiving a new RF signal from a device to authenticate;
F) by the second device, acquiring feature data from the new RF signal; and
G) by the second device, determining whether or not the device to authenticate is the first device by transmitting the feature data, acquired from the new RF signal, as input to the fitted model, wherein said authentication is based on only one component of said RF signal acquired over time selected from a group consisting of an in-phase component and a quadrature component so as to reduce amount of data to be processed while providing enhanced device authentication accuracy.
2 . The method according to claim 1 , wherein the model is a machine learning model, and fitting the model includes training the machine learning model using the feature data as input training data.
3 . The method according to claim 2 , wherein the machine learning model is trained in a semi-supervised manner by using input training data that exclusively includes feature data acquired from RF signals transmitted from the first device.
4 . The method according to claim 1 , wherein the model is an anomaly detection model.
5 . The method according to claim 1 , wherein, after storing of the fitted model in memory, the model is updated by the second device based on feature data acquired from a new RF signal received from the first device.
6 . The method according to claim 1 , further comprising normalizing the feature data before providing the normalized feature data to the model in at least one of C) and G).
7 . The method according to claim 1 , further comprising filtering the feature data before providing the filtered feature data to the model in at least one of C) and G).
8 . The method according to claim 1 , wherein A) to C) are performed by the second device or by a third device, different from the second device.
9 . The method according to claim 1 , wherein:
A) to C) are also performed for a plurality of RF signals transmitted by the second device to fit a model for authenticating said second device, and
D) to G) are also performed by the first device for authentication of the second device using said fitted model, so that the first and second devices perform a mutual authentication.
10 . The method according to claim 1 , wherein A) to C) are performed for a plurality of first devices and a plurality of fitted models, corresponding to the plurality of first devices respectively, and the plurality of fitted models are stored in the memory in the second device.
11 . The method according to claim 10 , further comprising establishing a communication between the device to authenticate and the second device, wherein the device to authenticate transmits an identifier to the second device; and
selecting, by the second device, the fitted model corresponding to the device to authenticate from among the plurality of fitted models stored in memory based on the identifier of said device to authenticate, the selected fitted model being used in G).
12 . A system for authentication of a first device, the system comprising:
a second device; and
a third device, wherein
the third device is configured to:
receive a plurality of RF signals from the first device,
acquire feature data from the received RF signals, and
fit a model for authenticating said first device using the acquired feature data,
wherein the second device is configured to:
store in memory the fitted model,
receive a new RF signal from a device to authenticate,
acquire feature data from the new RF signal, and
determine whether or not the device to authenticate is the first device by transmitting the feature data, acquired from the new RF signal, as input to the fitted model, wherein said authentication is based on only one component of said RF signal acquired over time selected from a group consisting of an in-phase component and a quadrature component so as to reduce amount of data to be processed while providing enhanced device authentication accuracy.
13 . A non-transitory computer-readable storage medium including computer executable instructions, wherein the instructions, when executed by a computer, cause the computer to perform the method according to claim 1 .
14 . A first device for authentication of a second device, the first device comprising
circuitry configured to:
receive a plurality of RF signals from the second device,
acquire feature data from the received RF signals,
fit a model for authenticating said second device using the acquired feature data,
store in memory the fitted model,
receive a new RF signal from a device to authenticate,
acquire feature data from the new RF signal, and
determine whether or not the device to authenticate is the second device by transmitting the feature data, acquired from the new RF signal, as input to the fitted model, wherein said authentication is based on only one component of said RF signal acquired over time selected from a group consisting of an in-phase component and a quadrature component so as to reduce amount of data to be processed while providing enhanced device authentication accuracy.
15 . The first device according to claim 14 , wherein:
the model is a machine learning model, and
the circuitry is configured to fit the model by training the machine learning model using the feature data as input training data.
16 . The first device according to claim 15 , wherein the machine learning model is trained in a semi-supervised manner by using input training data that exclusively includes feature data acquired from RF signals transmitted from the second device.
17 . The first device according to claim 14 , wherein the model is an anomaly detection model.
18 . The first device according to claim 14 , wherein, after storing of the fitted model in memory, the circuitry is configured to update the model based on feature data acquired from a new RF signal received from the second device.
19 . The first device according to claim 14 , wherein the circuitry is configured to normalize the feature data before providing the normalized feature data to the model.
20 . The first device according to claim 14 , wherein the circuitry is configured to filter the feature data before providing the filtered feature data to the model.