Unmanned aerial vehicle air-ground channel authentication method based on PRACH signal

Abstract

An unmanned aerial vehicle air-ground channel authentication method based on a PRACH signal belongs to the field of communication, and comprises the following steps: user equipment sends a lead code through the PRACH signal; after the base station receives the MSG1, the signals form vectors, a signal covariance matrix is calculated, eigenvalue decomposition is carried out on the covariance matrix, spatial feature estimation is carried out by using a MUSIC algorithm, signal average receiving power is calculated, eigenvectors are constructed and input into a machine learning algorithm, online identification is carried out, and a judgment result is output; if the user equipment is judged to be the user equipment in the authorized area, sending RAR to the unmanned aerial vehicle terminal; if the user equipment in the non-authorized area is judged, the MSG2 is not sent, the first step is directly executed, the power of the MSG1 is improved by the user equipment, and the PRACH signal is continuously sent by the unmanned aerial vehicle terminal until the maximum lead code sending frequency is exceeded, so that access failure is caused. The system response speed and the security defense capability are improved, the cost is low, and the calculation overhead is small.

Description

Technical Field

[0001] The present invention belongs to the field of communication technologies, and particularly relates to a method for authenticating an unmanned aerial vehicle (UAV) air-to-ground channel based on a Physical Random Access Channel (PRACH) signal. Background Art

[0002] The rapid popularization of unmanned aerial vehicles (UAVs) in civilian and industrial fields has brought new challenges to airspace security. Unauthorized UAV intrusions pose potential threats to public safety, critical infrastructure, and even national security (A. Fotouhi, H. Qiang, M. Ding, M. Hassan, L. G. Giordano, A. Garcia Rodriguez, and J. Yuan, “Survey on UAV cellular communications: Practical aspects, standardization advancements, regulation, and security challenges,” IEEE Communications surveys & tutorials, vol. 21, no. 4, pp. 3417–3442, Mar. 2019.). To address these security issues, researchers have proposed various authentication mechanisms, including authenticating non-networked UAVs through wireless protocols, radar-based detection of low-speed airborne targets, and verification in air-ground integrated networks.Terminal access authentication mainly includes password authentication, key negotiation protocols, and physical layer authentication technologies that utilize channel characteristics and radio frequency (RF) fingerprinting. References include (Y. Wang, W. Zhang, X. Wang, W. Guo, M. K. Khan, and P. Fan, “Improving the security of LTE-R for high-speed railway: from the access authentication view,” IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 2, pp. 1332–1346, Oct. 2020.), (N. Xie and S. Zhang, “Blind authentication at the physical layer under time-varying fading channels,” IEEE Journal on Selected Areas in Communications, vol. 36, no. 7, pp. 1465–1479, Apr. 2018.), and (L. Xiao, X. Wan, and Z. Han, “Phy-layer authentication with multiple landmarks with reduced overhead,” IEEE Transactions on Wireless Communications, vol. 17, no. 3, pp. 1676–1687, Dec. 2017.). Currently, although various physical layer methods have been proposed, such as location-based authentication, hardware-specific authentication, and hybrid schemes, research and experimental verification of physical layer authentication using standard uplink signals are still relatively scarce.

[0003] The existing research on physical layer authentication using standard uplink signals mainly includes the following aspects:

[0004] 1) Extract transient features using PRACH signals and classify different types of terminals using deep learning techniques; also study the extraction of PRACH signal differential constellation trajectory map features using multi-channel convolutional neural networks. Moreover, fuse with the phase data of DMRS channel state information and process the mixed feature matrix through neural networks, achieving good performance. For example, see the reference (Y. Liu, P. Zhang, J. Liu, Y. Shen, and X. Jiang, “Exploiting fine-grained channel / hardware features for phy-layer authentication in mmwave MIMO systems,” IEEE Transactions on Information Forensics and Security, vol. 18, pp. 4059–4074, Jun. 2023.).

[0005] 2) There is also research on extracting the differential constellation silicon-based map of DMRS from MSG3 and processing these features using a thousand-layer long short-term memory network to achieve cross-scenario classification and good accuracy at a specific signal-to-noise ratio (Y. Qiu, L. Peng, J. Zhang, M. Liu, H. Fu, and A. Hu, “Signal independent RFF identification for LTE mobile devices via ensemble deep learning,” in GLOBECOM 2022 - 2022 IEEE Global Communications Conference. IEEE, Jan. 2022, pp. 37–42.).

[0006] 3) In terms of SRS signals, a study proposed a radio frequency fingerprint identification method combining SRS channel estimation and discrete Fourier transform to achieve uplink signal authentication (H. Fu, H. Dong, J. Yin, and L. Peng, “Radiofrequency fingerprint identification for 5G mobile devices using DCTF and deep learning,” Entropy, vol. 26, no. 1, 2024. [Online]. Available: https: / / www.mdpi.com / 1099-4300 / 26 / 1 / 38.).

[0007] Almost all existing studies on physical layer authentication using standard uplink signals are for terrestrial communication scenarios, while there are few studies on three-dimensional UAV air-ground communication scenarios.

[0008] Currently, the authentication methods for UAVs are mainly divided into three types, namely:

[0009] The first type is mainly based on location features, relying on the spatial parameters between the UAV and the base station, such as angle, distance, and signal strength, etc. Information, by comparing the actual location of the UAV with the preset authorized area to judge its legitimacy.

[0010] The second type is mainly based on hardware features, using the inherent hardware characteristics of UAV radio equipment, such as radio frequency fingerprints and device timing differences. These characteristics stem from the tiny differences in the manufacturing process and can provide a unique identifier for each device, thus enabling accurate authentication.

[0011] The third type is mainly based on a hybrid authentication scheme. The hybrid scheme combines the advantages of the above two methods, taking into account both spatial location information and the inherent hardware characteristics of the device, in order to improve the accuracy and robustness of authentication and adapt to more complex and changeable application environments.

[0012] The existing studies on physical layer authentication using standard uplink signals mentioned above are both based on location features, and those based on hardware features and hybrid authentication schemes.

[0013] The above-mentioned authentication studies based on PRACH signals are all based on typical terrestrial communication environments, considering a small number of terminals, without considering complex three-dimensional space environments, and existing UAV authentication methods all occur in the middle and later stages of the access process, and it is difficult to guarantee the overhead resources for communication, and it will also bring a series of security problems. Summary of the Invention

[0014] When the current UAV terminal communicates with the base station, it needs to go through the PRACH channel for initial access. However, most existing UAV authentication methods rely on subsequent signaling phases (such as MSG3, RRC connection), with lagging identification and high consumption of overhead resources. To solve this problem, the present invention provides a UAV air-ground channel authentication method based on PRACH signals.

[0015] The present invention extracts features such as pitch angle, azimuth angle, and received power from the access request MSG1 signal initiated by the UAV terminal, and combines machine learning algorithms to identify whether it comes from an authorized airspace, so as to realize the early identification and interception of illegal UAV terminals. The present invention can make a judgment at the very front end of the random access process, directly reject illegal UAV terminals at the access request MSG2 stage, which not only saves communication resources, but also effectively prevents access-type DDoS attacks in air-ground communication, fills the gap in the existing protection mechanism at the initial stage of random access, and is applicable to the detection of illegal area UAV terminals in the 5G massive connection scenario.

[0016] The technical solution adopted by the present invention to solve the technical problem is as follows:

[0017] A method for authenticating an air-ground channel of a UAV based on a PRACH signal provided by the present invention includes the following steps:

[0018] Step 1: The user equipment sends a preamble based on the Zadoff–Chu sequence through the PRACH signal;

[0019] Step 2: After the base station receives the access request MSG1 of the UAV terminal, the following operations are performed:

[0020] S2.1: The base station forms the received signal into a vector;

[0021] S2.2: Calculate the covariance matrix of the received signal;

[0022] S2.3: Perform eigenvalue decomposition on the covariance matrix of the received signal;

[0023] S2.4: Use the MUSIC algorithm for spatial feature estimation;

[0024] S2.5: Calculate the average received power of the signal;

[0025] S2.6: Construct a feature vector;

[0026] S2.7: Input the feature vector into a machine learning algorithm for online identification and output a judgment result;

[0027] S2.8: If it is judged that the user equipment is in the authorized area, send an RAR to the UAV terminal; if it is judged that the user equipment is in the unauthorized area, do not send the access request MSG2 but directly go to step S1, the user equipment increases the power of the access request MSG1, and the UAV terminal continuously sends the PRACH signal until the maximum number of preamble transmissions is exceeded and the access fails; for special cases, when the unauthorized area user equipment and the authorized area user equipment send the same preamble sequence at the same time, the unauthorized area user equipment will also fail to access due to competition.

[0028] As a preferred embodiment, a method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a PRACH signal provided by the present invention further includes step three, where the user equipment sends an RRC connection request on the uplink shared channel allocated by the access request MSG2, carrying a temporary UE-ID randomly generated by the user equipment to identify its own ID. If multiple user equipments send access requests MSG3 on the same uplink shared channel due to preamble collision, the base station can only successfully decode the strongest signal among them, and other user equipments will retry random access after a timeout.

[0029] As a preferred embodiment, a method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a PRACH signal provided by the present invention further includes step four, where after successfully decoding the access request MSG3, the base station sends an RRC connection establishment message through the downlink shared channel and echoes the temporary UE-ID as a contention resolution flag to notify the corresponding user equipment that the access is successful.

[0030] As a more preferred embodiment, in step four, after receiving the confirmation message, the user equipment sends a HARQ-ACK and officially obtains a C-RNTI; the user equipment that does not receive an access request MSG4 matching its own UE-ID within the specified time will be regarded as a contention failure and will re-initiate the random access process.

[0031] As a preferred embodiment, the mathematical expression of the covariance matrix of the received signal is:

[0032]

[0033] where R represents the covariance matrix of the received signal, the received signal vector r[n] = {r1[n], r2[n], r3[n],... r M [n]}, 1 ≤ n ≤ N, M represents the number of antennas, N represents the number of sampling points received by each antenna, and the superscript H represents conjugate transpose.

[0034] As a preferred embodiment, the mathematical expression of the eigenvalue decomposition is:

[0035]

[0036] where R represents the covariance matrix of the received signal, E S represents the signal subspace, E N represents the noise subspace, Λ S represents the eigenvalue matrix corresponding to the signal subspace, Λ N represents the eigenvalue matrix corresponding to the noise subspace, and the superscript H represents conjugate transpose.

[0037] As a preferred implementation, the specific implementation process of the MUSIC algorithm includes constructing a steering vector and calculating the MUSIC spectrum; the mathematical expression of the steering vector is:

[0038]

[0039] The mathematical expression for calculating the MUSIC spectrum is:

[0040]

[0041] where θ represents the elevation angle; represents the azimuth angle; j represents the imaginary unit; λ represents the wavelength; p = {p1[n], p2[n], … p MN [n]} represents the set of three-dimensional spatial positions of all array elements of the antenna array; represents the direction vector; M represents the number of antennas; N represents the number of sampling points received by each antenna; the superscript H represents the conjugate transpose.

[0042] As a preferred implementation, the mathematical expression for the average received signal power is:

[0043]

[0044] where r m [n] represents the complex signal value received by the m-th antenna at the n-th sampling point; M represents the number of antennas; N represents the number of sampling points received by each antenna.

[0045] As a preferred implementation, the mathematical expression for the eigenvector is:

[0046]

[0047] where represents the average received signal power, θ est represents the estimated elevation angle, represents the estimated azimuth angle.

[0048] As a preferred implementation, the specific implementation process of step S2.7 is as follows:

[0049] S2.7.1: The UAV terminal flies within a specified area while collecting signals both inside and outside the legal area;

[0050] S2.7.2: Extract the eigenvectors from the collected signal samples and input them into the machine learning algorithm training model;

[0051] S2.7.3: Divide the training set and the validation set in a ratio of 8:2 and statistically calculate the accuracy of the validation set;

[0052] S2.7.4: Statistical model overhead;

[0053] S2.7.5: Select models with high accuracy, strong stability, and low overhead and deploy them to the base station.

[0054] The beneficial effects of the present invention are:

[0055] The present invention provides a drone air-to-ground channel authentication method based on PRACH signal, which innovatively makes authentication decisions in the first stage of random access (i.e., MSG1), fundamentally solving the problem that illegal drone terminals cannot be identified in the initial stage of random access, and improving the response speed and security defense capabilities of the system. In particular, in 5G networks, before sending an RRC connection request, the drone terminal first initiates an access request (i.e., MSG1) through the PRACH channel. If the drone terminal can be quickly authenticated at this stage, the system security will be effectively improved. To this end, the present invention extracts physical layer features such as pitch angle, azimuth angle, and received power from the MSG1 signal, and combines machine learning algorithms to determine whether the drone terminal is in the authorized airspace. After receiving the access request MSG1, the base station immediately determines whether to issue an access request MSG2 based on the physical layer features. If it is identified as illegal access, it refuses to respond, thereby achieving rapid authentication and access control in the initial stage of communication, and effectively preventing illegal drone terminals from accessing the network.

[0056] Compared with the prior art, the present invention has the following advantages:

[0057] (1) The present invention uses PRACH signals to extract features and conduct authentication research, and uses PRACH signals for authentication in three-dimensional space, filling the gap that it is currently not used for drone authentication analysis.

[0058] (2) The present invention utilizes the angle information of drone communication and the received signal strength to distinguish drone terminals in legal areas and illegal areas. The angle information is calculated using an angle estimation algorithm, and the received signal strength is calculated based on the received signal. The calculation and processing are embedded in the first stage of the random access process (i.e., MSG1). The present invention does not introduce additional computing overhead in the actual system, has low cost, and reduces computing overhead.

[0059] (3) The present invention modifies the traditional random access process and can detect drone terminals in illegal areas before the RRC connection request is sent. It can well ensure the normal communication of drone terminals in some special areas, and has low computing overhead, does not occupy network resources, and has a certain degree of anti-interference. The current technology has a great impact on power deception, and the angle information extracted by the present invention can resist power deception.

[0060] (4) The present invention aims to prevent an illegal area UAV terminal from accessing the base station and affecting the legal area UAV terminal. By training a simple machine learning model, a satisfactory balance between security level and computational overhead is achieved.

[0061] (5) The present invention can be applied to the UAV attack detection in the random access phase of the 5G NR network, with low cost and small computational overhead, and has a good effect at the large-scale event sites where UAV terminals are needed. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flowchart of a UAV air-to-ground channel authentication method based on the PRACH signal provided by the present invention.

[0063] Figure 2 It is a flowchart of step S2.

[0064] Figure 3 It is a schematic diagram of the three-dimensional feature distribution of the data in the legal area (INSIDE).

[0065] Figure 4 It is a schematic diagram of the three-dimensional feature distribution of the data in the illegal area (OUTSIDE).

[0066] Figure 5 It is a schematic diagram of the three-dimensional feature distribution of the legal / illegal area data. DETAILED DESCRIPTION OF THE INVENTION

[0067] The present invention will be further described in detail below with reference to the accompanying drawings.

[0068] A UAV air-to-ground channel authentication method based on the PRACH signal provided by the present invention is a new random access process scheme proposed to solve the access problem of UAV terminals in the illegal area. Based on the PRACH signal, signal features are extracted at the initial stage of random access, and a new random access process is proposed on the basis of the original random access process. By reconstructing the original random access process, the judgment of UAV terminals in the illegal area is added. Using the first signal sent by the UAV terminal, the channel characteristics and location characteristics are extracted through the angle estimation algorithm, and the machine learning algorithm is used for discrimination. The present invention uses the pitch angle, azimuth angle, and received power in the access request MSG1 signal initiated by the UAV terminal as physical layer features to distinguish UAV terminals in different areas, and can reject the access of user equipment in the random access process stage, with low resource overhead, low computational complexity, and strong real-time performance.

[0069] The present invention is mainly applicable to the four-step random access process, and is also applicable to the non-conflicting two-step random access process. The flowchart is based on 3GPP TS 38.300 V15.17.0 Figure 9.2.6-1.

[0070] See Figure 1 For the description, a method for authenticating the UAV air-ground channel based on the PRACH signal provided by the present invention has the following specific implementation process:

[0071] Step S1: The user equipment (UE) sends a preamble based on the Zadoff–Chu sequence through the PRACH signal, and the time slot, frequency domain resource, and cyclic shift used are configured by SIB2 (a base station broadcast channel (BCCH) system information block specified by 3GPP);

[0072] Step S2: As Figure 2 shown, after the base station receives the access request MSG1 initiated by the UAV terminal, the following operations are performed:

[0073] S2.1: The base station uses an 8*8 antenna array to receive the signal, and forms a received signal vector r[n] = {r1[n], r2[n], r3[n], … r M [n]}, 1 ≤ n ≤ N, M represents the number of antennas, N not only represents the number of sampling points received by each antenna, but also represents the number of snapshots used for covariance estimation. A snapshot represents the received signal combination of all antennas at the same time point. The duration symbol of the PRACH signal is determined according to the configuration of the cell base station. Here, N includes all sampling points of the PRACH signal at the current sampling rate.

[0074] S2.2: Calculate the covariance matrix of the received signal:

[0075]

[0076] Among them, R represents the covariance matrix of the received signal, and the superscript H represents the conjugate transpose.

[0077] S2.3: Perform eigenvalue decomposition on the covariance matrix R of the received signal to obtain the signal subspace E S and the noise subspace E N , and the mathematical expression of this process is as follows:

[0078]

[0079] Among them, Λ S represents the eigenvalue matrix corresponding to the signal subspace, and Λ N represents the eigenvalue matrix corresponding to the noise subspace.

[0080] S2.4: Use the MUSIC algorithm for spatial feature estimation. Generally speaking, it is to use the MUSIC algorithm to estimate from which direction the signal comes.

[0081] The specific implementation process of the MUSIC algorithm is as follows:

[0082] First, construct the steering vector, and its mathematical expression is as follows:

[0083]

[0084] Where, θ represents the pitch angle, and the theoretical value range is (0~90); represents the azimuth angle, and the theoretical value range is (-90~90); j represents the imaginary unit; λ represents the wavelength; p = {p1[n], p2[n], … p MN [n]} represents the set of three-dimensional spatial positions of all array elements of the antenna array, which is used to calculate the phase difference corresponding to each array element in the steering vector, that is, composed of the spatial coordinates of each array element; represents the direction vector, and its mathematical expression is as follows:

[0085]

[0086] Where, the superscript T represents the transpose operation.

[0087] Then, perform MUSIC spectrum calculation, and its mathematical expression is as follows:

[0088]

[0089] The steering vector describes the phase pattern of the signal coming in from the direction and what the signal looks like in the antenna array. If there is indeed a signal coming in from this direction , then the signal should be in the signal subspace, and the signal is orthogonal to the noise subspace. This is why the noise subspace is selected for angle estimation in the present invention. If the steering vector is orthogonal to the noise subspace, then the denominator will be close to 0, thus generating a peak. If the direction is incorrect, no peak will be generated.

[0090] The resolution of using the MUSIC algorithm for spatial feature estimation and the dimension of the covariance matrix all depend on the number of array elements, that is, M*N. The more array elements there are, the higher the dimension of the covariance matrix, the more angles can be distinguished, and the more accurate the estimation is.

[0091] S2.5: Calculate the average received power of the signal:

[0092]

[0093] Where, r m [n] represents the complex signal value received by the mth antenna at the nth sampling point.

[0094] S2.6: Construct the eigenvector:

[0095]

[0096] Among them, represents the signal average received power, and θ est represents the estimated elevation angle, represents the estimated azimuth angle.

[0097] The angles of the legal region will be concentrated in a cluster, and the signal average received power will also be concentrated in a cluster. However, the data in the illegal region is divergent. Therefore, it can be distinguished based on this feature vector f.

[0098] As Figure 3 shown, the data in the legal region will form a concentrated cluster, indicating that the signals sent by drones in the authorized airspace have similar spatial and power characteristics.

[0099] As Figure 4 shown, the data in the illegal region is distributed in a divergent manner, with a large degree of dispersion in spatial angles and received power, and no obvious cluster can be formed.

[0100] As Figure 5 shown, the legal region and the illegal region are shown on the same graph.

[0101] S2.7: Input the extracted feature vector into the selected machine learning algorithm for online identification; the specific implementation process is as follows:

[0102] S2.7.1: First, let the drone terminal fly within the specified area to collect signals inside and outside the legal region;

[0103] S2.7.2: Extract the feature vector from these signal samples according to the above step S2.6, and then label them and input them into the training models of the selected typical machine learning algorithms (such as KNN, SVM, XGBoost, and LightGBM, etc.);

[0104] S2.7.3: Divide the training set and the validation set in a ratio of 8:2, and count the accuracy of the validation set;

[0105] S2.7.4: While counting the accuracy of the validation set, also count the overhead of the model, which is mainly measured in two aspects: the size of the model and the time required to predict 100 signal samples;

[0106] S2.7.5: According to the needs, select a model with high accuracy, strong stability, and low overhead and deploy it to the base station.

[0107] S2.8: According to the judgment result output in the above step S2.7, if the judged user equipment is within the authorized area, after calculating the corresponding parameters, it will send a RAR (Random Access Response) to the UAV terminal, including RA-RNTI (Random Access-RNTI, used to identify the resource block used by the user to initiate random access), TA (Time Alignment), uplink grant, and temporary C-RNTI (Cell Radio Network Temporary Identifier). The user equipment uses TA to complete timing alignment and saves the C-RNTI.

[0108] If the judged user equipment is in the unauthorized area, it does not send the access request MSG2 and directly goes to step S1. The user equipment will increase the power of the access request MSG1. Then, according to 3GPP TS 38.321 (the core specification of the 5G NR MAC layer, which defines key mechanisms such as data scheduling, resource management, and signaling interaction), the UAV terminal will continuously send PRACH signals, and finally, the access fails because the maximum number of preamble transmissions preambleTransMax (given in the scheduling information of SIB1 sent by the network side) is exceeded.

[0109] Step S3: The user equipment sends a RRC connection request (RRC Connection Request) on the uplink shared channel (UL-SCH) allocated by the access request MSG2, carrying a temporarily generated UE-ID (Globally Unique Temporary UE Identity) by the user equipment to identify its own ID. If multiple user equipments send access request MSG3 using the same uplink shared channel resource simultaneously due to preamble collision, the base station can only successfully decode the strongest signal among them, and other user equipments will attempt random access after timeout.

[0110] Step S4: After successfully decoding the access request MSG3 of the user equipment, the base station sends an RRC connection setup message (RRC Connection Setup) through the downlink shared channel, and echoes the temporary UE-ID as a contention resolution flag to notify the corresponding user equipment that the access is successful. After receiving the confirmation message, the user equipment sends a HARQ-ACK (Hybrid Automatic Repeat Request Acknowledgement), and officially obtains the C-RNTI for subsequent control plane and user plane data communication; the user equipment that does not receive the access request MSG4 matching its own UE-ID within the specified time will be regarded as a contention failure and initiate the random access process again.

[0111] The present invention takes the physical layer PRACH signal (Physical Random Access Channel, used for the initial connection between the terminal and the network) as the starting point, extracts the spatial angle and energy characteristics, and combines the machine learning model to implement the airspace access control mechanism without encryption, effectively improving the anti-interference ability and access security for illegal access terminals, and taking into account the real-time performance and deployment complexity.

[0112] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a Physical Random Access Channel (PRACH) signal, characterized in that, It includes the following steps: Step 1: The user equipment sends a preamble based on the Zadoff–Chu sequence through the PRACH signal; Step 2: After receiving the access request MSG1 from the drone terminal, the base station performs the following operations: S2.1: Form a vector from the received signal; S2.2: Calculate the covariance matrix of the received signal; S2.3: Perform eigenvalue decomposition on the covariance matrix of the received signal; S2.4: Use the MUSIC algorithm for spatial feature estimation; S2.5: Calculate the average received power of the signal; S2.6: Construct eigenvectors; S2.7: Input the eigenvectors into a machine learning algorithm for online recognition and output a judgment result; S2.8: If it is judged that the user equipment is within the authorized area, send an RAR to the drone terminal; If it is judged that the user equipment is in the unauthorized area, do not send the access request MSG2 and directly go to step S1. The user equipment increases the power of the access request MSG1, and the drone terminal continuously sends the PRACH signal until the maximum number of preamble transmissions is exceeded, resulting in access failure; For special cases, user equipment in the unauthorized area and user equipment in the authorized area send the same preamble sequence simultaneously, and the user equipment in the unauthorized area will also fail to access due to competition.

2. The method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a PRACH signal according to claim 1, wherein It further includes step 3. The user equipment sends an RRC connection request on the uplink shared channel allocated by the access request MSG2, carrying a temporary UE-ID randomly generated by the user equipment to identify its own ID. If multiple user equipments send access requests MSG3 on the same uplink shared channel due to preamble collision, the base station can only successfully decode the strongest signal among them, and other user equipments will retry random access after timeout.

3. The method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a PRACH signal according to claim 2, wherein It further includes step 4. After successfully decoding the access request MSG3, the base station sends an RRC connection establishment message through the downlink shared channel and echoes the temporary UE-ID as a contention resolution flag to notify the corresponding user equipment of successful access.

4. The method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a Physical Random Access Channel (PRACH) signal according to claim 3, wherein, After receiving the confirmation message, the user equipment sends a HARQ-ACK and officially obtains a C-RNTI; User equipment that does not receive an access request MSG4 matching its own UE-ID within the specified time will be regarded as a contention failure and restart the random access process.

5. The method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a Physical Random Access Channel (PRACH) signal according to claim 1, wherein The mathematical expression of the covariance matrix of the received signal is: Among them, R represents the covariance matrix of the received signal, and the received signal vector r[n] = {r1[n], r2[n], r3[n], … r M [n]}, where 1 ≤ n ≤ N, M represents the number of antennas, N represents the number of sampling points received by each antenna, and the superscript H represents conjugate transpose.

6. The method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a Physical Random Access Channel (PRACH) signal according to claim 1, wherein The mathematical expression of the eigenvalue decomposition is: where, R represents the covariance matrix of the received signal, E S represents the signal subspace, E N represents the noise subspace, Λ S represents the eigenvalue matrix corresponding to the signal subspace, Λ N represents the eigenvalue matrix corresponding to the noise subspace, and the superscript H represents the conjugate transpose.

7. A method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a PRACH signal according to claim 1, characterized in that, The specific implementation process of the MUSIC algorithm includes constructing a steering vector and calculating the MUSIC spectrum; The mathematical expression of the steering vector is: The mathematical expression of the MUSIC spectrum calculation is: Among them, θ represents the pitch angle; represents the azimuth angle; j represents the imaginary unit; λ represents the wavelength; p = {p1[n], p2[n], … p MN [n]} represents the set of three-dimensional spatial positions of all array elements of the antenna array; represents the direction vector; M represents the number of antennas; N represents the number of sampling points received by each antenna; the superscript H represents the conjugate transpose.

8. The method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a Physical Random Access Channel (PRACH) signal according to claim 1, wherein The mathematical expression of the average received power of the signal is: where r m [n] represents the complex signal value received by the m-th antenna at the n-th sampling point; M represents the number of antennas; N represents the number of sampling points received by each antenna.

9. A method for authenticating an unmanned aerial vehicle (UAV) air-to-ground channel based on a Physical Random Access Channel (PRACH) signal, characterized in that, The mathematical expression of the eigenvector is: Among them, represents the signal average received power, and θ est represents the estimated elevation angle, represents the estimated azimuth angle.

10. The method for authenticating an unmanned aerial vehicle (UAV) air-ground channel based on a Physical Random Access Channel (PRACH) signal according to claim 1, wherein, The specific implementation process of step S2.7 is as follows: S2.7.1: The drone terminal flies within the specified area and collects signals inside and outside the legal area at the same time; S2.7.2: Extract eigenvectors from the collected signal samples and input them into the training model of the machine learning algorithm; S2.7.3: Divide the training set and the validation set in a ratio of 8:2 and count the accuracy of the validation set; S2.7.4: Count the model overhead; S2.7.5: Select a model with high accuracy, strong stability, and low overhead and deploy it to the base station.

Images