CN111830321B - A method of detection and identification of UAV based on radio frequency fingerprint - Google Patents

Abstract

The invention discloses a drone detection and identification method based on radio frequency fingerprints, belonging to the technical field of drone detection, comprising a signal receiving module, a signal processing module and a drone classification and identification module, and is suitable for drone detection and classification In the field of identification, the main idea is: use the RF front-end to receive wireless signals at various frequencies in the 2.4GHz band; determine whether the amplitude of the received signal is stable and greater than a preset threshold to detect suspected signals, and calculate whether the autocorrelation function of the suspected signal is periodic. To complete the detection of the UAV; if so, extract some features in the UAV signal as the "fingerprint" of the UAV device, and use the classification and identification algorithm to classify and identify the UAV.

Description

A method of detection and identification of UAV based on radio frequency fingerprint

technical field

The invention belongs to the field of UAV detection and identification, and specifically provides a UAV detection and identification method.

Background technique

With the widespread use of civil drones in today's society, while bringing a lot of convenience to human society, it will also bring some unexpected situations. Uncontrolled drones will lead to accidental injuries to pedestrians or buildings. Therefore, there is a need for a mechanism that can detect drones and identify their identities, so as to distinguish legal and illegal drones and improve control efficiency.

At present, the main UAV detection technologies include radio spectrum monitoring, radar detection, acoustic identification and visible light/infrared detection. The radio spectrum detection technology mainly uses radio monitoring equipment to detect the wireless signal of the target UAV, and through comprehensive analysis and processing, finds and identifies the target. UAV radar detection technology uses radar scanning technology to detect UAVs according to the phenomenon of reflected waves generated when electromagnetic waves pass through different transmission media. When the drone is flying, its motor work and rotor vibration will produce a certain degree of noise, and the noise generated by each drone is unique and can be used as the "audio fingerprint" of the drone. UAV sonic identification technology uses "audio fingerprints" to discover and detect UAVs. Visible light/infrared detection is to use visible light or thermal infrared reflection of the target, and use a combination of over-the-horizon, high-definition visible light cameras and infrared thermal imager sensors to detect drones. A comparison of the advantages and disadvantages of using the above technologies to detect and identify UAVs is shown in Figure 1. From Figure 1, it can be seen that the equipment of traditional UAV detection technology is relatively complex and the recognition accuracy is not high.

The UAV detection and identification technology based on the radio frequency fingerprint proposed by the present invention can not only solve the complex equipment problems in the UAV detection system in the past, but also has high identification accuracy. The so-called radio frequency fingerprint refers to the tiny random features on the radio frequency circuit in the process of manufacturing wireless communication equipment under the premise of ensuring the product is qualified. These features have uniqueness, universality, robustness and short-term invariance, similar to biological fingerprints that can uniquely identify an individual, these features are called radio frequency fingerprints. Since UAVs use wireless communication for information exchange, the detection and identification of UAVs based on RF fingerprint technology has great feasibility, and can effectively identify the identity of UAVs, which can improve the efficiency of UAV management and control. .

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a method for detecting and identifying an unmanned aerial vehicle based on a radio frequency fingerprint, which effectively solves the problems of complex and low identification efficiency of the current unmanned aerial vehicle detection and identification equipment system.

The invention provides the following technical solutions:

A method for detecting and identifying drones based on radio frequency fingerprints, the technical solution includes:

The signal receiving module collects wireless signals through the RF front-end, performs a series of preprocessing on the received signals and stores them.

The signal processing module comprehensively analyzes and processes the stored signals to realize the detection of UAVs.

The UAV identification module selects the features of the detected UAV signals and extracts the signal features, uses the extracted signal features as the "fingerprint" of the UAV, and uses the UAV identification algorithm to realize the classification and identification of the UAV. .

A method for detecting and identifying drones based on radio frequency fingerprints, comprising the following steps:

Step 1: Receive wireless signals through the radio frequency front-end, specifically: use a 2.4GHz omnidirectional antenna to receive wireless signals, amplify the wireless signals received by the antenna through a 2.4GHz low-noise high-frequency amplifier, and then send them to a filter for filtering, and filter the filtered signals. The signal is down-converted by the mixer to output relatively stable frequency spectrum information of the intermediate frequency signal.

Step 2: Process based on the signal obtained in Step 1. If the signal amplitude on a certain frequency band of the received signal is stably greater than the preset threshold σ, it is determined that there is a signal entering and the signal is recorded as a suspected signal, and the suspected signal is determined. The communication signal of the UAV is a periodic signal, and its autocorrelation function is still periodic, while the autocorrelation function of the noise interference signal is not periodic. Therefore, the suspected signal is determined by calculating whether the autocorrelation function of the suspected signal is periodic. Whether it is a drone signal, so as to realize the detection of the drone.

Step 3, based on the UAV signal obtained in step 2, perform data collection, which is divided into an offline stage and an online stage, and the offline stage: use the signal feature extraction algorithm to extract the UAV signal features, as the "fingerprint" of the UAV equipment, Finally, the extracted "fingerprint" data and the label representing the UAV category are stored as training data. In the online stage, the signal features of the UAV signal obtained in step 2 are extracted and stored as test data.

Step 4: Based on the training data and test data of the UAV signal obtained in step 3, the UAV classification and identification algorithm is used to realize the classification and identification of the UAV.

Preferably, the UAV signal feature extraction algorithm includes the following steps:

S1: Modulate the signal x(t)=x _I (t)+x _Q (t) as the input signal through the I/Q quadrature modulator to obtain the output signal s(t), where X _I (t) is in-phase component, X _Q (t) is the quadrature component. Find the complex number of s(t) and simplify it to get S _B (t)=αx(t)+βx ^* (t) where α=cosθ+jεsinθ, β=εcosθ+jsinθ, ε and θ are quadrature modulators respectively gain mismatch and phase mismatch parameters.

对r(t)取复共轭得到r^*(t)＝β^*x(t)+α^*x^*(t)，定义

计算Y(t)的自相关矩阵R_Y＝E[Y(t)Y(t)^H]，对R_Y进行化简可得

其中σ_x ²是x(t)的能量，σ_s ²是S_B(t)的能量。S2: Analysis of S _B (t) in S1. Define the received signal

Take the complex conjugate of r(t) to get r ^* (t)=β ^* x(t)+α ^* x ^* (t), the definition

Calculate the autocorrelation matrix of Y(t) R _Y =E[Y(t)Y(t) ^H ], and simplify R _Y to get

where σ _x ² is the energy of x(t) and σ _s ² is the energy of S _B (t).

代入α和β进行化简得

由化简结果可知，

只与ε和θ有关，对于不同的无人机设备，由I/Q失配产生的相位失配θ和增益失配ε是不同的，因此不同无人机的

值不同，那么就可以将

的值作为无人机设备的“指纹”。S3: Based on the autocorrelation matrix R _Y obtained in S2, define

Substitute α and β to simplify

From the simplified result, it can be seen that

It is only related to ε and θ. For different UAV devices, the phase mismatch θ and gain mismatch ε caused by I/Q mismatch are different, so the

different values, then the

The value of is used as the "fingerprint" of the drone device.

Preferably, the K-nearest neighbor algorithm of the UAV classification and identification algorithm includes the following steps:

S1: After obtaining the training data S and the data to be tested T of the UAV, load S and T into the algorithm, and specify the value of K as k.

S2: Based on the data S and T of S1, normalize the data to obtain new training data S' and new test data T', the normalization formula is X'=(X-minX)/(maxX-minX), where X is the original data, X' is the new data normalized by X, and maxX and minX are the maximum and minimum values in X, respectively.

S3: Based on the normalized new data S' and T' in S2, select the data t in the data to be tested T', and calculate the Euclidean distance d _i from t to the data _si in S', where i=(1,2 ,...n), where n is the total number of data in S'.

S4: Based on the Euclidean distance d _i obtained in S3, sort d _i in ascending order to obtain a distance set D=(d ₁ , d ₂ ,..., d _n ), where n is the total number of data in S'.

S5: Based on the distance D in S4, select the first k Euclidean distances in D to correspond to the data s" in the training data S' respectively.

S6: Based on the data s" in S5, count the number of categories in s", and use the label with the highest frequency as the category of the test data T, then the label is the category of the UAV to be tested, that is, the UAV is completed. identification.

beneficial effect

The beneficial effects of the present invention are as follows: the present invention firstly uses a radio frequency front-end platform to receive signals, specifically: using a 2.4GHz omnidirectional antenna to receive wireless signals at different frequencies from 2.412GHz to 2.472GHz, and preprocess the received signals; Whether the amplitude of the received signal is stable and greater than the preset threshold σ can detect the suspected signal, and analyze whether the autocorrelation function of the suspected signal is periodic to realize the detection of the UAV; after the UAV is detected successfully, use the signal feature extraction algorithm to extract the unmanned aerial vehicle. The UAV signal features are used as the "fingerprint" of the UAV equipment; finally, the UAV classification and identification algorithm is used to realize the classification and identification of the UAV. Among them, the signal features extracted in the present invention have uniqueness and short-term invariance, and can distinguish different UAV devices well. The UAV classification and identification algorithm uses the K nearest neighbor algorithm. The Euclidean distance is used in the algorithm to calculate the distance between the unknown point and the known point. By setting different K values, the algorithm has the highest recognition rate, which solves the problem based on the existing The problems of complex system and low identification efficiency in man-machine detection and identification equipment.

Description of drawings

Fig. 1 is the comparative schematic diagram of the advantages and disadvantages of the commonly used UAV detection technology in the present invention;

Fig. 2 is the schematic flow chart of the method for detecting and identifying the UAV in the present invention;

3 is a schematic flowchart of the classification and identification algorithm of UAV in the present invention;

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings:

The invention provides a method for detecting and identifying an unmanned aerial vehicle based on a radio frequency fingerprint, which is applied to an unmanned aerial vehicle detection and identification system. Specifically, the unmanned aerial vehicle detection and identification system includes: a signal receiving module, a signal processing module and an unmanned aerial vehicle. Identify the module

As shown in Figure 2, a method for detecting and identifying UAVs based on radio frequency fingerprints, the method specifically includes the following steps:

Step 1: Receive wireless signals through the radio frequency front end. Specifically: use a 2.4GHz omnidirectional antenna to receive wireless signals, amplify the wireless signals received by the antenna through a 2.4GHz low-noise high-frequency amplifier, and then send them to a filter for filtering, and pass the filtered signal through a mixer for down-conversion processing. Output relatively stable IF signal spectrum information.

Step 2: Process based on the signal obtained in step 1. If the signal amplitude on a certain frequency band of the received signal is stably greater than the preset threshold σ, it is determined that a signal has entered and the signal is recorded as a suspected signal. It is a periodic signal, and its autocorrelation function is still periodic, while the autocorrelation function of interference signals such as noise is not periodic. Therefore, it can be determined whether the suspected signal is unmanned by calculating whether the autocorrelation function of the suspected signal is periodic. UAV signal, so as to realize the detection of UAV.

r^*(t)为r(t)的复共轭,计算Y(t)的自相关函数

其中σ_x ²是理想信号x(t)的能量，σ_s ²是r(t)的能量。定义信号特征

将

的值作为该无人机设备的射频“指纹”，将采集的训练数据和测试数据分别存储。In step 3, data collection is performed based on the UAV signals obtained in steps 1-2, which are divided into training data collection in the online stage and test data collection in the offline stage. The specific steps are: the received signal is r(t), let

r ^* (t) is the complex conjugate of r(t), calculate the autocorrelation function of Y(t)

where σ _x ² is the energy of the ideal signal x(t) and σ _s ² is the energy of r(t). Define Signal Characteristics

Will

The value of is used as the RF "fingerprint" of the UAV equipment, and the collected training data and test data are stored separately.

Step 4, based on the UAV signal training data and test data obtained in step 3, use the K nearest neighbor algorithm to classify and identify the UAV. The flowchart of the K-nearest neighbor algorithm is shown in Figure 3. The specific steps are: (1) Normalize the data set to obtain new training data S' and new test data T', and the normalization formula is X'=( X-minX)/(maxX-minX), where X is the original data, X' is the new data after normalization, and maxX and minX are the maximum and minimum values in X, respectively. (2) Select the data t in the data to be tested T', and calculate the Euclidean distance d _i from t to the data _si in S', where i=(1,2,...n), and n is the total number of data in S' number. (3) Sort d _i in ascending order to obtain the distance set D=(d ₁ , d ₂ ,...,d _n ), where n is the total number of data in S'. (4) Select the first k Euclidean distances in D corresponding to the data points s" in the training set S'. (5) Count the number of categories in s", and use the category label with the highest frequency as the category of the test data T , the label is the category of the drone to be tested, that is, the identification of the drone is completed.

Claims (2)

1. a drone detection and identification method based on radio frequency fingerprint, is characterized in that, comprises: Step 1: Receive wireless signals through the radio frequency front-end, specifically: use a 2.4GHz omnidirectional antenna to receive wireless signals, amplify the wireless signals received by the antenna through a 2.4GHz low-noise high-frequency amplifier, and then send them to a filter for filtering, and filter the filtered signals. The signal is down-converted by the mixer to output relatively stable frequency spectrum information of the intermediate frequency signal; Step 2: Process based on the signal obtained in Step 1. If the signal amplitude on a certain frequency band of the received signal is stably greater than the preset threshold σ, it is determined that there is a signal entering and the signal is recorded as a suspected signal, and the suspected signal is determined. The communication signal of the UAV is a periodic signal, and its autocorrelation function is still periodic, while the autocorrelation function of the noise interference signal is not periodic. Therefore, the suspected signal is determined by calculating whether the autocorrelation function of the suspected signal is periodic. Whether it is a drone signal, so as to realize the detection of the drone; Step 3, based on the UAV signal obtained in step 2, perform data collection, which is divided into an offline stage and an online stage, and the offline stage: use the signal feature extraction algorithm to extract the UAV signal features, as the "fingerprint" of the UAV equipment, Finally, the extracted "fingerprint" data and the label representing the UAV category are stored as training data. In the online stage: the signal characteristics of the UAV signal obtained in step 2 are extracted as test data and stored; the step 3 Specifically: S1: Analyze the suspected signal described in step 2. The suspected signal is r(t)=αx(t)+βx ^* (t), where α=cosθ+jεsinθ, β=εcosθ+jsinθ, x(t) is an ideal signal, ε and θ are the gain mismatch and phase mismatch parameters of the quadrature modulator, respectively, taking the complex conjugate of r(t) to obtain r ^* (t)=β ^* x(t)+α ^* x ^* (t); S2: Analyze r(t) and r ^* (t) in S1, define

Calculate the autocorrelation matrix of Y(t) R _Y =E[Y(t)Y(t) ^H ] and simplify to get

where σ _x ² is the energy of x(t) and σ _s ² is the energy of r(t); S3: Define features based on the autocorrelation matrix R _Y obtained in S2

Substitute α and β to simplify

From the simplified result, it can be seen that

It is only related to ε and θ. For different UAV devices, the phase mismatch θ and gain mismatch ε caused by I/Q mismatch are different, so the

different values, then the

The value acts as the "fingerprint" of the drone device; In step 4, based on the training data and test data of the UAV obtained in step 3, the UAV classification and identification algorithm is used to realize the classification and identification of the UAV. 2. a kind of drone detection and identification method based on radio frequency fingerprint according to claim 1, is characterized in that, described step 4 is specifically: After obtaining the training data and test data of the UAV, the K-nearest neighbor algorithm is used to classify and identify the UAV.

Images