A Radio Frequency Fingerprint Extraction Method Against Channel Characteristic Interference

Abstract

The present invention discloses a method for extracting radio frequency fingerprints against channel characteristic interference, which includes the following steps: Among the two parties in wireless network communication, end A sends a signal known to both ends to end B; after receiving the signal, end B obtains the transmission characteristic signal from A to B and makes a response to end A. There are two optional response methods: (1) Remove the transmission characteristic signal from A to B from the known signal to obtain a type-I response signal and send it to end A; (2) Directly send the known signal as a type-II response signal to end A, and at the same time transmit the transmission characteristic signal from A to B to end A without distortion; End A equivalently extracts the features containing the fingerprint information of the device at end B in different ways according to the type of the received response signal, which can be used for the identity authentication of end B. Through the present invention, radio frequency fingerprints can be obtained by both communication parties through one interaction, are not affected by channel characteristics in a channel with good reciprocity, retain fingerprint details, and can be widely used for fingerprint extraction and identity authentication of radio frequency devices.

Description

Technical Field

[0001] The present invention relates to the technical field of radio frequency fingerprint extraction of physical layer signals, and particularly to a radio frequency fingerprint extraction method without a third party and resistant to channel characteristic interference. Background Art

[0002] With the rapid development of modern computer technology and the popularization of the new generation of wireless communication technology, people's daily life increasingly depends on wireless networks. However, due to the openness of the wireless network transmission medium, the mobility of terminals, and the instability of communication systems, the reliability and security of transmission are seriously threatened, resulting in illegal access to network resources, threats to people's communication privacy, and intrusion into computers. Therefore, research on security aspects such as device identity authentication technology in wireless networks has gradually become a hot topic in recent years.

[0003] Currently, to solve the access authentication problem of terminal devices in wireless networks, the more common identification and access control technologies are implemented based on MAC addresses and digital certificates. However, MAC addresses are extremely easy to forge and tamper with; digital certificate technology is relatively complex to use in the system, requires strong computing power to achieve the distribution of keys or certificates, and itself also has the risk of being stolen and misappropriated. Therefore, traditional wireless network security access methods all have more or less defects. In recent years, more and more research has shown that the radio frequency characteristics of devices can be extracted from the signals transmitted by wireless communication systems. Radio frequency fingerprint is an inherent physical characteristic of wireless devices, generated by electronic components such as capacitors, inductors, and other digital logic circuits contained in the devices, which cannot be cloned and is difficult to forge and tamper with, and has been proven to have characteristics such as universality, uniqueness, short-term invariance, independence, and robustness.

[0004] To achieve secure access authentication of devices in wireless networks relying on radio frequency fingerprints, effective and refined technical means are required to extract stable and recognizable fingerprint information. Most of the early research on radio frequency fingerprints focused on transient signals, such as extracting instantaneous frequency, transient intensity, etc. as fingerprints based on transient responses. In recent years, considering the short duration of transient signals, high detection difficulty, and large influence of the channel, more research has begun to turn to the extraction of steady-state characteristics. A typical one is the steady-state fingerprint extraction method based on the preamble. However, most of the existing radio frequency fingerprint extraction schemes will lose fingerprint information when removing channel characteristics, resulting in a reduced discrimination degree of different fingerprint characteristics and bringing difficulties to subsequent device identity authentication. Therefore, it is necessary to redesign a radio frequency fingerprint extraction scheme that can resist channel characteristic interference and not lose fingerprint information when extracting fingerprints. Summary of the Invention

[0005] Technical problem: To overcome the deficiencies in the prior art, the present invention provides a method for extracting radio frequency fingerprints that resists interference from channel characteristics. This invention does not require a third party for signal acquisition and can obtain radio frequency fingerprints through a single interaction and a small amount of calculation between the two communication parties, without being restricted by signal formats. This invention is applied to channels with good reciprocity, can better resist interference from channel characteristics, does not lose fingerprint information, and can be widely used for fingerprint extraction and identity authentication of radio frequency devices.

[0006] Technical solution: To achieve the above objective, the present invention adopts a method for extracting radio frequency fingerprints that resists interference from channel characteristics, including the following steps:

[0007] Step 1, known signal transmission: For the two communication parties A and B in a wireless network, A sends the known signal x(n) to B, and the spectrum of this signal is X(z);

[0008] Step 2, signal reception: B obtains the received signal y B (n), and the spectrum of this signal is Y B (z);

[0009] Step 3, obtaining the transmission characteristic signal from A to B: B removes the known signal x(n) from the received signal y B (n) to obtain the transmission characteristic signal v AB (n), and the spectrum of this signal is V AB (z);

[0010] Step 4, response: B makes a response to A, and there are two response methods to choose from: B removes the transmission characteristic signal v AB (n) from the known signal x(n) to obtain the type I response signal x' B (n) and sends it; or B directly sends the known signal x(n) as the type II response signal and transmits the transmission characteristic signal v AB (n) from A to B to A without distortion;

[0011] Step 5, response signal reception: A receives the response signal sent by B. If the response signal sent by B is the type I response signal, then A obtains the signal y' A (n). If the response signal sent by B is the type II response signal, then A obtains the signal y A (n);

[0012] Step 6, feature extraction: A extracts the features containing the fingerprint information of B's device in the frequency domain according to the signal y' A (n) and x(n), or according to the signal y A (n), x(n) and v AB (n).

[0013] Further, in the present invention: in the said step 2, the received signal y B (n) is obtained after the transmitted signal x(n) passes through the A - end transmitter with transfer function H TA (z), the channel with transfer function H AB (z) and the B - end receiver with transfer function H RB (z) in sequence. The spectrum of the received signal satisfies:

[0014] Y B (z) = X(z)·H TA (z)·H AB (z)·H RB (z).

[0015] Further, in the present invention: in the said step 3, the method for obtaining the transmission characteristic signal from A to B is: in the frequency domain, divide the spectrum of the received signal by the spectrum of the known signal to obtain the spectrum of the transmission characteristic signal from A to B:

[0016]

[0017] H TA (z) is the transfer function of the A - end transmitter, H AB (z) is the transfer function of the channel from A to B, H RB (z) is the transfer function of the B - end receiver.

[0018] Further, in the present invention: in the said step 4, the method for the B - end to generate the type - I response signal is: in the frequency domain, first construct the spectrum X' B (z) of the type - I response signal by dividing the spectrum of the known signal by the spectrum of the transmission characteristic signal from A to B, that is:

[0019] X' B (z) = X(z) / V AB (z),

[0020] then obtain the time - domain waveform of the response signal through the inverse fast Fourier transform. V AB (z) is the spectrum of the transmission characteristic signal from A to B.

[0021] Further, in the present invention: in the said step 5, the response signal received at the A - end passes through the B - end transmitter with transfer function H TB (z), the channel with transfer function H BA (z) and the A - end receiver with transfer function H RA (z) in sequence. If the B - end sends a type - I response signal, the spectrum Y A (n) of the signal y' A obtained at the A - end satisfies:

[0022]

[0023] X' B (z) is the spectrum of the Class-I response signal, which is constructed from the spectrum X(z) of the known signal and the spectrum V AB (z) of the transmission characteristic signal from A to B; if the signal sent by the B-end is a Class-II response signal, then the signal y A (n) spectrum Y A (z) satisfies:

[0024] Y A (z) = X(z) · H TB (z) · H BA (z) · H RA (z).

[0025] Furthermore, in the present invention: in the said step 6, according to the signal y' A (n) and x(n) to extract the feature RFF B (z) containing the fingerprint information of the B-end device, the method is: calculate the amplitude quotient of the received response signal Y A '(z) and the known signal X(z) in the frequency domain, that is:

[0026]

[0027] H TA (z) is the transfer function of the A-end transmitter, H TB (z) is the transfer function of the B-end transmitter, H AB (z) is the transfer function of the A-to-B channel, H BA (z) is the transfer function of the B-to-A channel, H RA (z) is the transfer function of the A-end receiver, H RB (z) is the transfer function of the B-end receiver.

[0028] Furthermore, in the present invention: in the said step 6, according to the signal y A (n), x(n) and v AB (n) to extract the feature RFF B (z) containing the fingerprint information of the B-end device, the method is: in the frequency domain, first divide the spectrum Y A (n) of the received signal y A (z) by the spectrum X(z) of the known signal x(n) to obtain the spectrum of the B-to-A transmission characteristic signal:

[0029]

[0030] H TB (z) is the transfer function of the B-end transmitter, HBA (z) is the transfer function of the B-to-A channel, H RA (z) is the transfer function of the receiver at end A; then calculate V BA (z) and V AB (z) of the amplitude quotient, that is:

[0031]

[0032] V AB (z) is the spectrum of the A-to-B transmission characteristic signal v AB (n), using the transfer function H of the transmitter at end A TA (z), the transfer function H of the A-to-B channel AB (z) and the transfer function H of the receiver at end B RB (z) is expressed as:

[0033] V AB (z) = H TA (z) · H AB (z) · H RB (z).

[0034] Furthermore, in the present invention: the channel transfer functions H AB (z) and H BA (z) should satisfy channel reciprocity, that is:

[0035] H AB (z) ≈ H BA (z),

[0036] The extracted feature containing the fingerprint information of the device at end B is expressed as:

[0037]

[0038] Advantageous effects: Compared with the prior art, the present invention has the following advantages:

[0039] (1) Without the need for signal acquisition by a third party, the two communication parties can obtain the feature containing the device fingerprint information through only one interaction and a small amount of calculation;

[0040] (2) Under the condition of good channel reciprocity, the extracted feature removes the channel characteristics but does not lose the device fingerprint information, and the feature discrimination degree is high;

[0041] (3) Not affected by the signal format, fingerprints can be extracted for different types of transmitted signals in the communication system.

[0042] The method for extracting radio frequency fingerprints against channel characteristic interference proposed by the present invention can be used for fingerprint extraction of various radio frequency devices in a reciprocal channel and can be used for authenticating the identity of terminals in a wireless network. Brief Description of the Drawings

[0043] Figure 1 Schematic flow chart of the method for extracting radio frequency fingerprints against channel characteristic interference according to the present invention;

[0044] Figure 2 Time-domain waveform diagram of a long pilot symbol known to both communication parties in the Wi-Fi application scenario according to the present invention;

[0045] Figure 3 Frequency-domain waveform diagram of a long pilot symbol known to both communication parties in the Wi-Fi application scenario according to the present invention;

[0046] Figure 4 Frequency-domain waveform diagram of a long pilot symbol received by end B in the Wi-Fi application scenario according to the present invention;

[0047] Figure 5 Frequency-domain waveform diagram of the transmission characteristic signal from A to B calculated by end B in the Wi-Fi application scenario according to the present invention;

[0048] Figure 6 Time-domain waveform diagram of the type-I response signal sent by end B in the Wi-Fi application scenario according to the present invention;

[0049] Figure 7 Frequency-domain waveform diagram of the type-I response signal sent by end B in the Wi-Fi application scenario according to the present invention;

[0050] Figure 8 Frequency-domain waveform diagram of a long pilot symbol in the type-I response signal received by end A in the Wi-Fi application scenario according to the present invention;

[0051] Figure 9 Frequency-domain waveform diagram of a long pilot symbol in the type-II response signal received by end A in the Wi-Fi application scenario according to the present invention;

[0052] Figure 10 Frequency-domain waveform diagram of the transmission characteristic signal from B to A calculated by end A in the Wi-Fi application scenario according to the present invention;

[0053] Figure 11 Schematic diagram of the 52-dimensional features containing device fingerprint information extracted in the Wi-Fi application scenario according to the present invention. Detailed Description of the Invention

[0054] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings:

[0055] The present invention can be implemented in many different forms and should not be construed as limited to the embodiments described herein. On the contrary, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0056] As Figure 1 shown, it is a schematic diagram of the overall process of a radio frequency fingerprint extraction method for anti-interference of the primary channel characteristics proposed by the present invention. This method specifically includes the following steps.

[0057] Step 1, known signal transmission: For communication parties A and B in a wireless network, A sends a signal known to both parties to B. The time-domain sequence of this signal is {x(n)}, (n = 0, 1, 2,..., 64 - 1), and the spectrum of this signal is {X(k)}, (k = 0, 1, 2,..., 64 - 1).

[0058] Specifically, taking a Wi-Fi signal that conforms to the IEEE 802.11 protocol OFDM physical layer frame format as an example, the preamble signal of each frame is fixed. Therefore, when extracting the fingerprint of a Wi-Fi device, a 64-point ideal long preamble code symbol in the preamble signal can be used as the known signal for both communication parties. Figure 2 And Figure 3 respectively give the time-domain waveform diagram and frequency-domain waveform diagram of this symbol. As Figure 3 shown, an ideal long preamble code symbol contains 52 subcarriers. The time-domain waveform can be obtained by performing a 64-point inverse fast Fourier transform on the frequency-domain waveform.

[0059] Step 2, signal reception: B obtains the received signal {y B (n)}, (n = 0, 1, 2,..., 64 - 1), and the spectrum of this signal is {Y B (k)}, (k = 0, 1, 2,..., 64 - 1);

[0060] Specifically, the signal sent from A to B passes through the transmitter H TA (k) of A, the channel H AB (k), and the receiver H RB (k) of B in sequence. Taking a Wi-Fi signal as an example, in the received signal obtained by B, the spectrum of a 64-point long preamble code symbol should satisfy:

[0061] Y B (k) = X(k) · H TA (k) · H AB (k) · H RB (k), k = 0, 1, 2,..., 64 - 1

[0062] The frequency-domain waveform of this long preamble code symbol is as Figure 4 shown.

[0063] Step 3, obtain the transmission characteristic signal from A to B: At the B end, remove the signals known to both communication parties from the received signal to obtain the transmission characteristic signal {v AB (n)}, (n = 0, 1, 2,..., 64 - 1), and the spectrum of this signal is {V AB (k)}, (k = 0, 1, 2,..., 64 - 1);

[0064] Specifically, at the B end, divide the spectrum of the received long pilot symbol by the spectrum of the known ideal long pilot symbol to obtain the spectrum of a 64 - point transmission characteristic signal:

[0065]

[0066] The frequency - domain waveform of this transmission characteristic signal is as Figure 5 shown.

[0067] Step 4, reply: The B end makes a reply to the A end, and there are two reply methods to choose from: The B end removes the transmission characteristic signal from A to B from the known signal {x(n)}, (n = 0, 1, 2,..., 64 - 1) to obtain the type - I reply signal {x' B (n)}, (n = 0, 1, 2,..., 64 - 1) and sends it; or the B end directly sends the known signal {x(n)}, (n = 0, 1, 2,..., 64 - 1) as the type - II reply signal, and at the same time transmits the transmission characteristic signal from A to B to the A end without distortion;

[0068] Specifically, for Wi - Fi signals, the method for the B end to send a type - I reply signal to the A end is: First, construct the spectrum of a type - I reply symbol by dividing the spectrum of the ideal long pilot by the spectrum of the transmission characteristic signal from A to B, that is:

[0069] X' B (k) = X(k) / V AB (k), k = 0, 1, 2,..., 64 - 1

[0070] Then, obtain the time - domain waveform of this reply symbol through 64 - point inverse fast Fourier transform. Figure 6 And Figure 7 respectively give the time - domain waveform diagram and frequency - domain waveform diagram of this symbol.

[0071] Step 5, receive the reply signal: The A end receives the reply signal sent by the B end. If the B end sends a type - I reply signal, the A end obtains the signal {y' A (n)}, (0, 1, 2, 64 - 1), if the B end sends a type - II reply signal, the A end obtains the signal {y A (n)}, (n = 0, 1, 2,..., 64 - 1);

[0072] Specifically, the response signal sent by the B side passes through the transmitter H TB (k), the channel H BA (k), and the receiver H RA (k) in sequence. The received signal {y' A (n)} with (0, 1, 2, 64 - 1) and its spectrum {Y A '(k)} with (k = 0, 1, 2,..., 64 - 1) should satisfy:

[0073]

[0074] Similarly, the spectrum {Y A (k)} of the received signal {y A (n)} with (n = 0, 1, 2,..., 64 - 1) should satisfy:

[0075] Y A (k) = X(k) · H TB (k) · H BA (k) · H RA (k), k = 0, 1, 2,..., 64 - 1

[0076] For Wi-Fi signals, if the A side receives a type-I response signal, the frequency-domain waveform diagram of one 64-point long pilot symbol is as shown in Figure 8 ; if the A side receives a type-II response signal, the frequency-domain waveform diagram of one 64-point long pilot symbol is as shown in Figure 9 .

[0077] Step 6, feature extraction: The A side extracts features containing the fingerprint information of the B-side device based on the signals {y' A (n)} with (0, 1, 2, 64 - 1) and {x(n)} with (n = 0, 1, 2,..., 64 - 1), or based on the signals {y A (n)} with (n = 0, 1, 2,..., 64 - 1), {x(n)} with (n = 0, 1, 2,..., 64 - 1), and {v AB (n)} with (n = 0, 1, 2,..., 64 - 1) in the frequency domain.

[0078] Specifically, if the B side transmits a type-I response signal, the way for the A side to extract features containing the fingerprint information of the B-side device is: calculate the amplitude quotient of the received response signal and the known signal in the frequency domain, that is:

[0079]

[0080] If the Class II response signal is transmitted from end B, the method for end A to extract features is as follows: In the frequency domain, first divide the spectrum of the received signal by the spectrum of the known signal to obtain the spectrum of the transmission characteristic signal from B to A:

[0081]

[0082] Then calculate the amplitude quotient of the transmission characteristic signal from B to A and the transmission characteristic signal from A to B in the frequency domain, that is:

[0083]

[0084] Figure 10 The frequency-domain waveform diagram of the transmission characteristic signal from B to A extracted by end A in the Wi-Fi signal is given.

[0085] Furthermore, from the expression of the above RFF B (k), it can be seen that the two feature extraction methods are equivalent;

[0086] Furthermore, according to the channel reciprocity: H AB (k)≈H BA (k), the features containing the device fingerprint information extracted by end A can be expressed as:

[0087]

[0088] Figure 11 The features containing device fingerprint information extracted according to a 64-point long pilot symbol in the Wi-Fi signal are given. Since a long pilot symbol only contains 52 subcarriers, the dimension of this feature is 52 dimensions.

[0089] It should be noted that the above embodiments only represent some implementation manners of the present invention, and their descriptions should not be construed as limiting the scope of the patent of the present invention. It should be pointed out that for those of ordinary skill in the art, several improvements can be made without departing from the concept of the present invention, and these should all fall within the protection scope of the present invention.

Claims

1. A method for extracting radio frequency fingerprints against channel characteristic interference, characterized in that: including the following steps, Step 1, known signal transmission: For two communication parties A and B in a wireless network, party A sends a signal x(n) known to both parties to party B, and the spectrum of this signal is X(z); Step 2, signal reception: The B end obtains the received signal y B (n), and the spectrum of this signal is Y B (z); Step 3, obtain the transmission characteristic signal from A to B: At the B side, remove the signal x(n) known to both communication parties from the received signal y B (n) to obtain the transmission characteristic signal v AB (n), and the spectrum of this signal is V AB (z); Step 4, Response: End B makes a response to End A. There are two response methods to choose from: End B removes the transmission characteristic signal v AB (n) from the known signal x(n) to obtain a type-I response signal x' B (n) and sends it; or End B directly sends the known signal x(n) as a type-II response signal, and at the same time transmits the A-to-B transmission characteristic signal v AB (n) to End A without distortion; Step 5, response signal reception: The A end receives the response signal sent by the B end. If the response signal sent by the B end is a type I response signal, the A end obtains the signal y' A (n). If the response signal sent by the B end is a type II response signal, the A end obtains the signal y A (n); Step 6, Feature extraction: End A extracts the feature RFF A (n) and x(n), or extracts the feature RFF A (n), x(n) and v AB (n) in the frequency domain to obtain the feature RFF B (z) that contains the device fingerprint information of End B.

2. The method for extracting radio frequency fingerprints against channel characteristic interference according to claim 1, wherein: In the said step 2, the received signal y B (n) is obtained after the transmitted signal x(n) passes through the A-end transmitter with the transfer function H TA (z), the channel with the transfer function H AB (z), and the B-end receiver with the transfer function H RB (z) in sequence. The spectrum of the received signal satisfies: Y B (z) = X(z) · H TA (z) · H AB (z) · H RB (z).

3. The method for extracting radio frequency fingerprints against channel characteristic interference according to claim 1, wherein: In the said Step 3, the method for obtaining the transmission characteristic signal from A to B is: in the frequency domain, dividing the spectrum of the received signal by the spectrum of the known signal to obtain the spectrum of the transmission characteristic signal from A to B: H TA (z) is the transfer function of the transmitter at end A, H AB (z) is the transfer function of the channel from A to B, H RB (z) is the transfer function of the receiver at end B.

4. The method for extracting radio frequency fingerprints against channel characteristic interference according to claim 1, wherein: In step 4, the method for generating the type-I response signal at end B is as follows: in the frequency domain, first construct the spectrum X' B (z) of the type-I response signal by dividing the spectrum of the known signal by the spectrum of the transmission characteristic signal from A to B, i.e.: X' B (z) = X(z) / V AB (z), The time-domain waveform of the response signal is obtained by performing the inverse fast Fourier transform again, V AB (z) is the spectrum of the transmission characteristic signal from A to B.

5. The method for extracting radio frequency fingerprints against channel characteristic interference according to claim 1, wherein: In the said step 5, the response signal received at end A successively passes through the transmitter at end B with transfer function H TB (z), the channel with transfer function H BA (z), and the receiver at end A with transfer function H RA (z). If the response signal sent from end B is a type-I response signal, then the spectrum Y' A (n) of the signal y' A (z) obtained at end A satisfies: X' B (z) is the spectrum of the Class-I response signal, which is constructed from the spectrum X(z) of the known signal and the spectrum V AB (z) of the transmission characteristic signal from A to B; if the Class-II response signal is sent at the B end, then the signal y A (n) spectrum Y A (z) satisfies: Y A (z) = X(z)·H TB (z)·H BA (z)·H RA (z).

6. The method for extracting radio frequency fingerprints against channel characteristic interference according to claim 1, characterized in that: In step 6, according to the signal y' A (n) and x(n), extract the feature RFF B (z) containing the fingerprint information of the B-side device, and the method is as follows: calculate the amplitude quotient of the received response signal Y' A (z) and the known signal X(z) in the frequency domain, that is: H TA H(z) is the transfer function of the transmitter at end A TB H(z) is the transfer function of the transmitter at end B AB H(z) is the transfer function of the channel from A to B BA H(z) is the transfer function of the channel from B to A RA H(z) is the transfer function of the receiver at end A RB H(z) is the transfer function of the receiver at end B 7. The method for extracting radio frequency fingerprints against channel characteristic interference according to claim 1, characterized in that: In step 6, according to the signal y A (n), x(n) and v AB (n), the method for extracting the feature RFF B (z) containing the fingerprint information of the B-side device is as follows: in the frequency domain, first, according to the spectrum Y A (n) of the received signal y A (z) is divided by the spectrum X(z) of the known signal x(n) to obtain the spectrum of the transmission characteristic signal from B to A: H TB (z) is the transfer function of the transmitter at the B end, H BA (z) is the transfer function of the channel from B to A, H RA (z) is the transfer function of the receiver at the A end; then calculate V BA (z) and V AB (z) of the amplitude quotient, that is: V AB (z) is the transmission characteristic signal v from A to B AB (n) of the spectrum, using the transfer function H of the transmitter at end A TA (z), the transfer function H of the channel from A to B AB (z) and the transfer function H of the receiver at end B RB (z) is expressed as: V AB (z) = H TA (z) · H AB (z) · H RB (z).

8. The method for extracting radio frequency fingerprints against channel characteristic interference according to any one of claims 6 and 7, characterized in that: The channel transfer function H AB (z) and H BA (z) should satisfy channel reciprocity, i.e.: H AB (z) ≈ H BA (z), The feature containing the device fingerprint information of party B extracted is expressed as:

Images