Covert communication method and system based on Internet of Vehicles

Abstract

The invention discloses a covert communication method and system based on the Internet of Vehicles, and the method comprises the steps: determining a communication receiver object which comprises a vehicle and a background service center; if the object of the communication receiver is a vehicle, identity authentication and session key negotiation are carried out through vehicle speed difference sampling, and secret information is encrypted and then embedded into vehicle speed data or multimedia data to carry out covert communication; and if the object of the communication receiver is a background service center, the collected image data is used as a carrier to carry a secret image, and the secret image is embedded into the vehicle-mounted image data through hidden communication based on an image-hiding-image neural network or hidden communication based on an image or a video frame, so that hidden communication is realized. The security of the secret information can be ensured, the image data can be safely transmitted, and the security of a sender is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle networking, and in particular to a covert communication method and system based on vehicle networking. Background Art

[0002] Safely transmitting secret messages is an important task in combat and secret communication under information conditions. Seeking new steganographic methods has gradually become a new key difficulty that needs to be solved. With the advancement of technology, steganography can be achieved through methods based on carrier selection or carrier synthesis. However, the current image steganography technology only considers the concealment of the carrier and the secret carrier, but does not consider the concealment of the sending behavior.

[0003] At the same time, most of the current communication networks, such as videos and portals, are usually bound to personal identities, making it difficult to achieve covert communication. This makes it easy for the reviewer to take corresponding measures against specific types of communication networks, making it difficult to ensure the safety of special agents.

[0004] Covert communication technology is mainly used to support sensitive intelligence transmission, confidential communication, watermark or security protocol construction in uncontrolled situations. The sender of covert communication uses the identity of various daily industry practitioners as a cover, and uses the redundancy of communication content on daily communication platforms to carry secret information and send it to targeted targets. In theory, any daily communication behavior and platform are potential carriers of covert communication, which provides a lot of prospects and possibilities for the development of covert communication technology. The more common the communication platform, the more suitable it is for providing cover for covert communication behavior. Historically, covert communication was mainly based on steganography technology based on chemical majors. With the popularization of digital multimedia, digital images, videos, audio, text and other carriers have become mainstream. Generative multimedia data in the era of artificial intelligence has also brought new directions for covert communication. At present, the security basis of covert communication lies in keeping the secret carrier indistinguishable from the original carrier, so as to achieve the role of covering communication behavior. In contrast, steganalysis technology is committed to analyzing the abnormality of steganographic behavior and the abnormality of the carrier after steganography, so as to discover the existence of covert communication. The game-like confrontation between the two shows that the current development of covert communication technology shows a trend of diversification, specialization, efficiency and high security. Due to the diversity and complexity of the carriers used in covert communication, it is very difficult to model natural carriers and secret carriers. For example, in traditional image steganography technology, the modeling dimensions of natural images are often tens of thousands of dimensions, which brings a huge computational burden to the selection and security protection of steganographic features, and also makes it difficult for the secret carrier model to take into account both the embedding capacity and the carrier distortion. Therefore, for covert communication based on a certain carrier, achieving concealment has always been a difficult problem.

[0005] When introducing a new steganographic carrier, two factors are usually considered: one is whether the application environment of the carrier is widely used and can provide cover for covert communication; the other is the conditions and practical difficulty of achieving concealment, whether there are conditions or possibilities to ensure concealment. At present, several representative new types of covert communications are explored based on the above two considerations.

[0006] Generative steganography is one of the most representative techniques in recent years. Its carriers are multimedia derived from different generative models, such as generative adversarial networks (GANs), diffusion networks, and other deep generators. Generative carriers contain different types of multimedia, such as generated images and generated videos. Generative covert communication schemes can realize the automatic selection of embedded features, directly generate secret images, or merge secret messages with carrier images. However, the above methods have the following limitations: (a) there is a lack of unified evaluation criteria for the generated media, and there are deficiencies in security demonstration; (b) there are probabilistic fluctuations in the extraction accuracy, and the accuracy of communication loss is not enough; (c) an auxiliary network is required during extraction, which will generate additional communication overhead.

[0007] Covert communication based on the blockchain platform has the advantage of hiding the identities of the sender and receiver, and does not suffer from the compression distortion problem in traditional social communication channels, which can ensure the accuracy of data extraction. However, compared with other carriers, the communication efficiency of blockchain is relatively low.

[0008] Covert communications based on subthreshold channels (uniform resource locators, unmanned systems, and maritime communication systems) can provide reliable guarantees in terms of security and concealment (Evade Detection, EC for short). However, they are suitable for specific scenarios, have limited users, and are not widely used.

[0009] With the rapid development of smart cities and the Internet of Things (IoT) technology, the Internet of Vehicles (IoV) has gradually become a common communication medium as a typical application. Daily communications between vehicles and between vehicles and the backend can be confidential. The rapid development of network technology and software and hardware technology are invisibly promoting the maturity of IoT technology. IoT technology has been widely used in various fields. IoT technology is a technology that connects objects such as daily necessities, household appliances, and sensor devices to the Internet and shares data. Through IoT technology, people can remotely control through smart devices to achieve automated and intelligent management and operation. Combining IoT technology with information transmission is also one of the current hot issues.

[0010] There are many existing algorithms for confidential communication based on vehicle-mounted Internet networks, but the characteristics of such algorithms are that they use encryption or covert communication technology to ensure confidentiality and integrity during vehicle-mounted information communication. The main research status is as follows: Suresh JS uses a third-party trust platform module embedded in the vehicle to protect vehicle-mounted network information data from a hardware perspective. In vehicle-mounted networks, there are few studies on the use of information hiding technology to achieve privacy data security protection. De Fuentes et al. embedded the user's identity information into the subliminal channel and used the LSB hiding algorithm to ensure the security of the information. However, the LSB steganography algorithm has a low steganographic capacity, and the receiver cannot recover the original information. TaheriS changes the sending interval between service and control data packets in the vehicle network protocol and represents the secret information with the length of the sending interval. Although this scheme can ensure the embedding of a large amount of data, it will change the timing of the protocol to a certain extent, affecting the stability of the entire protocol, thereby bringing security risks.

[0011] However, according to existing public literature, covert communication and encryption technologies currently mainly serve the security of vehicle-mounted networks, but there is no method for covert communication based on vehicle-mounted networks. Realizing covert communication under vehicle-mounted networks has certain technical advantages in terms of behavioral security. Because there are a large number of vehicle-mounted networks, the communication volume is complex and difficult to be sniffed by attackers. The main difficulty lies in the method of covert communication, scene construction and guarantee of communication volume. Summary of the invention

[0012] In order to solve the above technical problems, the present invention proposes a covert communication method and system based on the vehicle-mounted Internet network, which utilizes the vehicle's driving status and communication network to achieve covert transmission of secret information to the outside.

[0013] On the one hand, to achieve the above-mentioned purpose, the present invention provides a covert communication method based on Internet of Vehicles, comprising:

[0014] Determine a communication recipient object, wherein the communication recipient object includes a vehicle and a background service center;

[0015] If the communication recipient is a vehicle, identity authentication is performed through vehicle speed difference sampling, a session key is negotiated, and the secret information is encrypted and embedded into the vehicle speed data or multimedia data for covert communication;

[0016] If the communication recipient is a background service center, the collected image data is used as a carrier to carry the secret image, and the secret image is embedded in the vehicle-mounted image data through covert communication based on a map-hidden-map neural network or covert communication based on images or video frames to achieve covert communication.

[0017] Preferably, identity authentication is performed by sampling the vehicle speed difference as follows:

[0018] Within the driving safety constraint conditions, the speed of the driving vehicles is sampled at the same time intervals to obtain a driving speed difference sequence, and the driving speed difference sequence is used to confirm whether the driving vehicles can communicate with each other. If communication is possible, the message transmission key is calculated.

[0019] Preferably, the driving safety constraint condition is:

[0020] The sender's vehicle speed is recorded as v1, the receiver's vehicle speed is recorded as v2, the sender's vehicle acceleration is recorded as v1'=Δv1 / Δt, t is the driving time; v1"=Δv1' / Δt, v1" indicates the change amplitude of v1', the larger the absolute value of v1" is, the more complex the driving environment of the vehicle is;

[0021] When the acceleration v1' of the sender's vehicle and the acceleration v2' of the receiver's vehicle are both less than the safety threshold v sec ', and the acceleration change amplitude v1" of the sender's vehicle and the acceleration change amplitude v2" of the receiver's vehicle are both less than the safety threshold v sec ", it indicates that the vehicle is within the driving safety constraint conditions.

[0022] Preferably, confirming whether the moving vehicles can communicate with each other is as follows:

[0023] Set the negotiation password between the sender and the receiver to psd∈{0,1} l , where l is the password length agreed upon by both parties;

[0024] The vehicle speed is obtained once at the same interval, sampled l times, and the speed difference sequence is obtained by calculation, which is recorded as psd'={psd1',psd2',...,psd l '} l ;

[0025] If the driving speed difference sequence psd' is exactly the same as the negotiated password psd between the sender and the receiver, the two parties enter the confidential communication stage, otherwise the communication is abandoned;

[0026] Among them, the driving speed difference Δv=|v1-v2|, when Δv=0, the transmitted information bit is psd i '=0;Δv=e(m / s), e>0, e is the basic unit of information transmission speed; when Δv=e, the transmitted information bit psd i '=1, that is, Δv=round(|v1-v2|), and the round(.) function means rounding the input to an approximate result.

[0027] Preferably, the secret information is encrypted and embedded into the vehicle speed data for covert communication as follows:

[0028] Using multi-base encoding, when the embedded data is a b-base number, the embedded information of the hidden information is l s is the message length, and the sampling interval is t sam The starting point of the sampling time is the first hour after the driving safety constraint is met and the command is connected, and then the information obtained by the i-th sampling is obtained, which is recorded as s i ,s i =round(|v1-v2| / e).

[0029] Preferably, the secret information is encrypted and embedded into multimedia data for covert communication as follows:

[0030] The embedded information s of the covert communication is embedded in a picture, audio or video to obtain a secret picture, secret audio or secret video, and sent to a receiving vehicle. The receiving vehicle uses an extraction algorithm to extract s and uses a key to decrypt to obtain the secret information.

[0031] Preferably, the communication process of the background service center is:

[0032] The collected image data is used as a carrier to steganographically record the secret image, and the image is processed with the image-hidden neural network and the vehicle-mounted image to generate a secret image. The secret image is then sent to the automobile service center, and the automobile service center extracts the secret information through the secret image.

[0033] Preferably, the process of generating a secret image by processing the image-hidden neural network and the vehicle-mounted image is as follows:

[0034] Match the size of the carrier image and the secret information, and input the embedding function to generate the secret image, specifically:

[0035] X Ste =StegoNET(X sec ,X co )

[0036] Where, X sec is the secret image, X co is the carrier image, X Ste is the secret image, and StegoNET(.) is the operation function.

[0037] Preferably, the process of extracting the secret image through the secret image is:

[0038] The secret image and any random vector are input into the reverse recovery network, preprocessed by discrete wavelet transform to obtain a feature map, and then the feature map is sent to the reversible block of the network. After several layers of reversible block transform, a new feature map is generated, and finally the restored image and the restored carrier image are obtained by inverse wavelet transform.

[0039] On the other hand, in order to achieve the above-mentioned purpose, the present invention also provides a covert communication system based on the Internet of Vehicles, comprising:

[0040] The hidden communication module between vehicles and the communication module between vehicles and the backend service center,

[0041] The covert communication module between vehicles is used to perform identity authentication through speed difference sampling, negotiate session keys, encrypt secret information and embed it into vehicle speed data or multimedia data for covert communication;

[0042] The communication module of the background service center is used to use the collected image data as a carrier to carry secret information, and embed the secret information into the vehicle-mounted image data through covert communication based on image-hidden-image neural network or covert communication based on image or video frame to realize covert communication.

[0043] Compared with the prior art, the present invention has the following advantages and technical effects:

[0044] The present invention improves the concealment and security of image transmission behavior from two perspectives: the steganographic algorithm and the sending behavior, ensures the security of the secret information itself, and safely transmits image data, making it difficult for the enemy to discover and detect the hidden secret information, while ensuring the security of the sender.

[0045] In terms of steganography algorithm, the present invention uses a densely connected reversible neural network to achieve image hiding and recovery. The carrier image and the secret image are transformed into a secret image through a forward hiding reversible transformation, and the secret image and the random variable are transformed into a restored image through a reverse recovery reversible transformation, which realizes the reversibility of the algorithm and can accurately achieve the steganography and recovery of image information at a high rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0047] Figure 1 A flow chart of a covert communication method based on an in-vehicle Internet network according to an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of a secret information transmission process according to an embodiment of the present invention;

[0049] Figure 3 A schematic diagram of an image steganography model based on a reversible network according to an embodiment of the present invention;

[0050] Figure 4 A schematic diagram of a secret image steganography process according to an embodiment of the present invention;

[0051] Figure 5 A schematic diagram of an image restoration process according to an embodiment of the present invention;

[0052] Figure 6 A schematic diagram of speed simulation according to an embodiment of the present invention;

[0053] Figure 7 It is a schematic diagram of acceleration simulation of an embodiment of the present invention;

[0054] Figure 8 It is a schematic diagram of simulating the acceleration change rate according to an embodiment of the present invention;

[0055] Fig. 9 This is a schematic diagram of the distribution of the extracted information when it is 0, 1, and 2 in an embodiment of the present invention;

[0056] Fig.10 This is a schematic diagram of the simulation of the speed fluctuation of two vehicles according to an embodiment of the present invention;

[0057] Fig.11 This is a schematic diagram of the working principle of the SSH protocol according to an embodiment of the present invention;

[0058] Fig.12 This is a schematic diagram of an image transmission code based on SSH according to an embodiment of the present invention;

[0059] Fig.13 This is a schematic diagram of the EV "Clairvoyance" function of an embodiment of the present invention;

[0060] Fig.14 A schematic diagram showing the comparison between a carrier image and a secret image according to an embodiment of the present invention;

[0061] Fig.15 A schematic diagram of the training and steganographic time corresponding to different models in an embodiment of the present invention;

[0062] Fig.16 A schematic diagram of a covert communication system architecture based on an in-vehicle Internet network according to an embodiment of the present invention;

[0063] Fig.17 1 and 2 are test images of an embodiment of the present invention, wherein (a) is test image one, (b) is test image two, (c) is test image three, (d) is test image four, and (e) is test image five. DETAILED DESCRIPTION

[0064] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0065] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0066] First, the covert communication in this embodiment is explained:

[0067] Covert communication is a communication method that allows a process to transmit information in a form that violates the system security policy. Covert communication is not a communication channel intended to transmit information, but it is widely used for the secure transmission of network information data.

[0068] This embodiment proposes a covert communication method based on the Internet of Vehicles, which is divided into two modes. Mode 1 is the communication between vehicles, and mode 2 is the communication between the vehicle-based Internet platform and the background service center. For mode 1, the communication platform that can be relied on in this embodiment mainly includes two aspects. One is the vehicle's own driving recorder, which can always sense the speed of surrounding vehicles and its own vehicle. The second is the vehicle-mounted communication software, that is, multimedia data can be transmitted between vehicles based on Android or Hongmeng system.

[0069] Mode 1 is the communication between vehicles, which performs identity authentication through speed difference sampling, negotiates session keys, and then encrypts secret information and embeds it into vehicle speed data or multimedia data for covert communication. Mode 2 is the communication between the Internet of Vehicles and the background service center, which uses the image data collected by the vehicle's driving recorder as a carrier to carry secret information, and embeds secret information into the vehicle-mounted image data through covert communication based on image-hidden neural network or covert communication based on image or video frame to achieve covert communication.

[0070] The basic functions of the Internet of Vehicles: 1. Both parties can obtain the speed information of nearby vehicles in real time. For example, the ADAS advanced driving assistance function of the 360 ​​smart cloud mirror series can display the driving speed of nearby vehicles; 2. The on-board systems of both parties have instant messaging software and support mutual communication such as multimedia data.

[0071] A covert communication method based on Internet of Vehicles, such as Figure 1 ,include:

[0072] The hidden communication process between vehicles and the communication process between vehicles and the background service center;

[0073] The covert communication process between vehicles is to perform identity authentication through speed difference sampling, negotiate session keys, encrypt secret information and embed it into vehicle speed data or multimedia data for covert communication;

[0074] The communication process of the background service center is to use the collected image data as a carrier to carry secret information, and embed the secret information into the vehicle-mounted image data through covert communication based on image-hidden-image neural network or covert communication based on image or video frame to realize covert communication.

[0075] Furthermore, the covert communication process mainly includes two processes: one is password docking and verification; the other is covert communication.

[0076] Specifically, identity authentication is performed through speed difference sampling:

[0077] Within the driving safety constraint conditions, the speed of the driving vehicles is sampled at the same time intervals to obtain a driving speed difference sequence, and the driving speed difference sequence is used to confirm whether the driving vehicles can communicate with each other. If communication is possible, the message transmission key is calculated.

[0078] The driving safety constraints are:

[0079] The sender's vehicle speed is recorded as v1, the receiver's vehicle speed is recorded as v2, the sender's vehicle acceleration is recorded as v1'=Δv1 / Δt, t is the driving time; v1"=Δv1' / Δt, v1" indicates the change amplitude of v1', the larger the absolute value of v1" is, the more complex the driving environment of the vehicle is;

[0080] To ensure safety, both parties need to perform password matching or communication during stable driving, that is, when the acceleration v1' of the sender's vehicle and the acceleration v2' of the receiver's vehicle are both less than the safety threshold v sec ', and the acceleration change amplitude v1" of the sender's vehicle and the acceleration change amplitude v2" of the receiver's vehicle are both less than the safety threshold v sec ", it indicates that the vehicle is within the driving safety constraint conditions.

[0081] Further, whether the moving vehicles can communicate with each other is confirmed as follows:

[0082] Set the negotiation password between the sender and the receiver to psd∈{0,1} l , where l is the password length agreed upon by both parties;

[0083] The vehicle speed is obtained once at the same interval, sampled l times, and the speed difference sequence is obtained by calculation, which is recorded as psd'={psd1',psd2',...,psd l '} l ;

[0084] In this embodiment, the sampling speed difference is the interval time t sam Get a result once. This interval ensures that both parties can adjust the speed. For example, t sam The sampling time can be 1-2 seconds. The starting point of the sampling time can be the first hour that meets the driving safety constraint.

[0085] If the driving speed difference sequence psd' is exactly the same as the negotiated password psd between the sender and the receiver, the two parties enter the confidential communication stage, otherwise the communication is abandoned;

[0086] In this embodiment, the communication key in the communication process is K:

[0087] K = ZUC(psd||psd)(1);

[0088] Among them, ZUC() is the national standard encryption ZUC algorithm.

[0089] Specifically, when the driving speed difference Δv=|v1-v2| and Δv=0, the transmitted information bit is psd i '=0;Δv=e(m / s), e>0, e is the basic unit of information transmission speed; when Δv=e, the transmitted information bit psd i '=1; Since the speed may not be accurately matched, an approximate calculation can be performed near 0 or e, that is, Δv=round(|v1-v2|), and the round(.) function represents the approximate result of rounding the input.

[0090] Furthermore, the secret information is encrypted and embedded into the vehicle speed data for covert communication:

[0091] Using multi-base encoding, when the embedded data is a b-base number, the embedded information of the hidden information is l s is the message length, and the sampling interval is t sam The starting point of the sampling time is the first hour after the driving safety constraint is met and the command is connected. The information obtained by the i-th sampling is recorded as s i ,s i =round(|v1-v2| / e).

[0092] Specifically, let the communication content be m, first encrypt m to obtain the embedded information s:

[0093] s=ENC(m,K)(2);

[0094] In the formula, ENC(.) indicates that the secret message m is encrypted using a symmetric encryption algorithm, and the key is K.

[0095] In this embodiment, if the two parties do not have the in-vehicle interconnection platform support for transmitting multimedia data, the information s can continue to be transmitted based on the speed difference between the two parties. The process is as follows:

[0096] When performing speed difference communication, the sender and receiver still need to transmit information through the speed difference Δv. In order to improve communication efficiency, multi-base encoding can be used. When the embedded data is a b-base number, l s is the message length. The sampling interval is t sam The starting point of the sampling time can be the first hour after the driving safety constraints are met and the command is connected. The information obtained by the i-th sampling is recorded as s i :

[0097] s i =round(|v1-v2| / e) (3).

[0098] The following is an analysis of the main parameters of speed difference sampling in this embodiment:

[0099] 1) The basic unit of speed for information transmission (m / s) is e. The value of e is proportional to the accuracy of the information sent based on the speed difference, that is, the larger the value of e, the higher the accuracy of information transmission.

[0100] To ensure the accuracy of vehicle speed monitoring, the speed difference should not be too large, otherwise it will easily cause the distance between the two vehicles to be too large. The upper limit of the vehicle speed fluctuation value can be set to 10e.

[0101] 2) The relationship between information transmission and speed difference. The sender transmits information by adjusting its own speed to change the speed difference with the receiver. According to formula (3), when the speed difference is close to 0, the information sent is 0; when the speed difference is close to e, the information sent is e; when the speed difference is close to ke, the information sent is k.

[0102] 3) Receiver speed constraint: Considering the impact of speed measurement accuracy and error on information transmission accuracy, the receiver vehicle speed fluctuation range should be controlled within the stable speed v2±0.5e.

[0103] Furthermore, the secret information is encrypted and embedded into the multimedia data for covert communication:

[0104] The embedded information s of covert communication is embedded into pictures, audio, and video to obtain secret pictures, secret audio, and secret video, and sent to the receiving vehicle. The receiving vehicle uses an extraction algorithm to extract s and uses a key to decrypt to obtain the secret information.

[0105] Specifically, for the situation where an instant messaging software connection can be established between vehicles to achieve workshop interconnection, a multimedia-based covert communication technology can be used to achieve covert communication. Currently, the covert communication technology based on multimedia (video, audio, image, etc.) is relatively mature. In this embodiment, the embedded information of the covert communication is s. At this time, relying on the multimedia sharing software supported by the interconnection system of the vehicles of both parties, s can be embedded in pictures, audio, and video to obtain secret pictures, secret audio, and secret video, and sent to the receiving vehicle. The receiving vehicle uses an extraction algorithm to extract s, and then uses the key K to decrypt to obtain the secret information m. In addition, the introduction of multimedia covert communication technology can better ensure that the two communicating vehicles will not arouse the suspicion of attackers, thereby ensuring the confidentiality and security of secret message transmission.

[0106] Furthermore, the communication process of the background service center is:

[0107] The collected image data is used as a carrier to steganographically record the secret information, and the hidden-image neural network and the vehicle-mounted image are processed to generate a secret image, which is then sent to the automobile service center, and the automobile service center extracts the secret information through the secret image.

[0108] The collected image data is used as a carrier to steganographically record the secret image, and the image is processed with the image-hidden neural network and the vehicle-mounted image to generate a secret image. The secret image is then sent to the automobile service center, and the automobile service center extracts the secret information through the secret image.

[0109] Specifically, in this embodiment, the communication method between the background and the moving car mainly includes cellular network (such as 4G / 5G), Wi-Fi, etc. The vehicle establishes a connection with the server of the service center through the built-in communication module. The main purpose of its communication includes the following 6 aspects of service content:

[0110] 1. Repair and maintenance services: When a smart car breaks down, the owner can send fault information to the service center through the vehicle's smart system. The service center's technicians can make a preliminary judgment on the cause of the fault based on the fault code and the vehicle's feedback. For example, if the vehicle's power system fails, the service center can prepare the relevant repair tools and parts in advance to prepare for the vehicle to come in for repair.

[0111] 2. Roadside assistance service: in the event of an accident or breakdown, the owner can call the rear service center for assistance with one click. The service center can quickly locate the vehicle (through the vehicle's built-in GPS and other positioning systems) and arrange for the nearest rescue vehicle to go.

[0112] 3. Provide towing service. According to the degree of damage to the vehicle and the on-site conditions, choose the appropriate towing method to tow the vehicle to a nearby repair station.

[0113] 4. Software upgrade service: The software systems of smart cars (such as autonomous driving assistance systems, in-vehicle infotainment systems, etc.) need to be constantly updated to fix vulnerabilities, improve performance, and add new features. The rear service center can use OTA (Over-The-Air) technology to send software update packages to vehicles in motion or when the vehicle is stationary.

[0114] 5. User consulting service. If the owner has any questions about the vehicle's functions, configurations, etc., he can contact the customer service staff at the rear service center at any time. The customer service staff can provide answers to the owner through telephone, online customer service, etc. For example, if the owner does not understand how to turn on or off the vehicle's intelligent driving mode, the service center can provide detailed operating instructions.

[0115] 6. Data monitoring and analysis services: The rear service center can collect vehicle data in real time or regularly, such as the vehicle's driving speed, battery status (for electric vehicles), and the working status of parts. By analyzing these data, the service center can discover potential problems in advance. For example, by analyzing the battery's charge and discharge data, the battery's health status can be predicted, and the owner can be reminded whether the battery needs to be replaced.

[0116] Functions 5 and 6 above indicate that the vehicle and the background can communicate actively and passively, which provides the basic communication conditions for the covert communication process.

[0117] The Internet of Vehicles has the following characteristics: the vehicle takes real-time photos, and the collected road condition images are sent to the automobile service center with the help of the communication module on the vehicle. In this embodiment, the road condition image taken by the driving recorder and transmitted back to the rear is used as the carrier image, and the secret information (or secret image) is steganographic to generate a secret image. The car sends the secret image to the automobile service center with the help of the Internet of Vehicles. The receiver of the secret information receives the information at the automobile service center, and obtains the sent secret image by batch screening of images. The current steganography technology based on image or video frame data is relatively mature and can provide technical support for this embodiment. At the same time, considering the situation that the sender of covert communication collects secret information during driving, in this case, the secret information communication exists in the form of images. This embodiment mainly introduces a reversible information network with images hidden in images as a covert implementation example of mode 2.

[0118] Furthermore, the process of generating a secret image by processing the image-hidden neural network and the vehicle-mounted image is as follows:

[0119] Match the size of the carrier image and the secret information, and input the embedding function to generate the secret image, specifically:

[0120] X Ste =StegoNET(X sec ,X co );

[0121] Where, X sec is the secret image, X co is the carrier image, X Ste is the secret image, and StegoNET(.) is the operation function.

[0122] Specifically, Figure 2 The figure shows a covert communication technology scenario based on the image-hidden-image neural network. The secret image is generated by the image-hidden-image neural network and the vehicle image to generate a secret image that is indistinguishable from the vehicle image, and then sent to the backend service center through the vehicle's built-in communication software. The service center completes the recovery of the secret image and obtains the secret information.

[0123] The information feedback process is mainly divided into three steps: information preprocessing, information embedding, and information extraction.

[0124] The process of information preprocessing is:

[0125] The secret image is denoted by X sec , the carrier image is denoted as X co ;

[0126] In the information preprocessing process, a lightweight image steganography scheme based on a reversible neural network is used to steganographically hide the secret image. The processing process mainly involves matching X sec With X co The dimensions of , and used as input to the embedding function.

[0127] Information embedding:

[0128] The function StegoNET(.) can embed the secret image into the carrier image to synthesize the secret image X Ste .X Ste With X co There is consistency in visual and statistical analysis, so it can provide cover.

[0129] X Ste =StegoNET(X sec ,X co )(4).

[0130] Information extraction:

[0131] Note that the function ExtNET(.) can extract the secret image from the encrypted image:.

[0132] X sec=ExtNET(X Ste )(5).

[0133] As another implementation method, this embodiment mainly generates a secret image by using a secret image and a carrier image through a forward steganalysis network, and obtains a secret image by using a reverse recovery network with a secret image and a randomly generated variable. Figure 3 shown.

[0134] The input of the forward network includes X sec and X co . Preprocessed by discrete wavelet transform (DWT). Wavelet transform divides the image into low-frequency and high-frequency wavelet sub-bands to better integrate the secret information into the carrier image. The good reconstruction characteristics of wavelets help reduce information loss and improve image hiding performance. After wavelet transform, the size of the feature map changes. Using wavelet transform, the computational cost is reduced and the training progress is accelerated. Since wavelet transform is bidirectional, it will not affect the end-to-end training of the network model.

[0135] Furthermore, the process of extracting the secret image through the secret image is as follows:

[0136] The secret image and any random vector are input into the reverse recovery network, preprocessed by discrete wavelet transform to obtain a feature map, and then the feature map is sent to the reversible block of the network. After several layers of reversible block transform, a new feature map is generated, and finally the restored image and the restored carrier image are obtained through inverse wavelet transform processing.

[0137] Specifically, the steganographic process and the recovery process share the same network, and the network parameters can also be shared. co , the secret image is X sec , the secret image is X Ste , the loss information is r, and the introduced random variable is Z.

[0138] In this embodiment, there are 8 reversible blocks with the same structure. For the Lth hidden block in the steganographic process, the input is X L co and X L sec , the output is X L+1 co and X L+1 sec , as shown in formulas (6)-(7):

[0139]

[0140] Where σ is the activation function.

[0141] The output obtained after 8 reversible block transforms is then inversely transformed to obtain the encrypted image and loss information.

[0142] The Lth display block in the recovery process, the input is X l+1 Ste and Z l+1 , the output is X l Ste and Z l , as shown in formulas (8)-(9):

[0143]

[0144] The final output is then transformed through inverse wavelet transform to obtain the recovered secret image.

[0145] Considering the reversibility of the network, f(·), r(·) and y(·) can use any function, and this embodiment is implemented by dense connection. This connection method is conducive to the propagation of information and gradients, can effectively alleviate the gradient disappearance problem, and improve the learning ability and performance of the network. In the experiment, it was found that the use of 7 layers of convolution is the most effective dense connection method.

[0146] The feature map is sent to the reversible block of the network, and a new feature map is generated after 8 layers of reversible block transformation. Then, it is processed by inverse wavelet transform (IWT) to obtain the secret image and loss information. In this process, due to the large capacity of embedded information, some information may be lost during the secret image extraction process, and the embedding of the secret image may affect the carrier image, resulting in information loss. Therefore, the loss information consists of the lost information and the damaged carrier information. Figure 4 As shown, the embedded function is hidden in a daily function of a common in-vehicle visualization system, thereby ensuring the concealment of the communication function software.

[0147] As another implementation, a random variable needs to be introduced in the recovery process. This variable is randomly selected from a Gaussian distribution, but it must be ensured to match the distribution of the lost information. In contrast to the steganographic process, the input of the recovery network is the secret image and the random vector. They are first preprocessed by discrete wavelet transform (DWT), and then the feature map is fed into the reversible block of the network. After 8 layers of reversible block transform, a new feature map is generated, and finally processed by inverse wavelet transform (IWT) to obtain the restored image and the restored carrier image.

[0148] The distortion function of this embodiment mainly consists of three parts: hiding loss, low-frequency wavelet loss and recovery loss. The hiding loss is to ensure that the secret image embedded with secret information is indistinguishable from the carrier image. The low-frequency wavelet loss is to reduce the risk of being detected. Compared with information in the low-frequency area, information in the high frequency area is less likely to be detected. The recovery loss is to make the secret image after steganography and recovery as consistent as possible with the original secret image. The total loss function is the weighted sum of the hiding loss, low-frequency wavelet loss and recovery loss. Figure 5 Schematic diagram of the image restoration process.

[0149] This embodiment also provides a covert communication system based on the Internet of Vehicles, such as Fig.16 ,include:

[0150] The hidden communication module between vehicles and the communication module between vehicles and the backend service center,

[0151] The covert communication module between vehicles is used to perform identity authentication through speed difference sampling, negotiate session keys, encrypt secret information and embed it into vehicle speed data or multimedia data for covert communication;

[0152] The communication module of the background service center is used to use the collected image data as a carrier to carry secret information, and embed the secret information into the vehicle-mounted image data through covert communication based on image-hidden-image neural network or covert communication based on image or video frame to realize covert communication.

[0153] In order to more clearly express the technical solution of the present invention, the functional simulation and result verification process of the above method are provided below:

[0154] Mode 1: Covert communication process between vehicles

[0155] The simulation diagram of communication status judgment based on speed is as follows Figure 6-Figure 8 The figure contains 4 stages. Time 1-4 is the acceleration stage: Stage 1 is t = 0-3, at this time the vehicle is in the acceleration stage, although the encryption Figure 7 The acceleration of stage 1 is small, but the distance between the two vehicles is large, which is not suitable for communication matching. Stage 2 is t=3-9. The speed fluctuates in this stage, but it is basically stable. Communication can be carried out according to the speed difference between the two vehicles. Stage 3 is t=9-13. The speed fluctuates in this stage, and the fluctuation range is large, indicating that the road conditions are complicated and it is not easy to carry out safe and reliable communication. Stage 4 is t=13-15. The speed drops rapidly in this stage, indicating that the two vehicles are separated and no longer communicate.

[0156] Information encoding based on speed difference

[0157] The preset secret data is a ternary number, that is, the information comes from {0, 1, 2}. By simulating the speed of the two cars, Fig. 9 As shown in the figure, the difference between the speeds of the two vehicles during the fluctuation process is the transmitted information. By comparing the speed difference sampling information of 100-bit data, the extraction accuracy is 100%.

[0158] from Figure 8 According to formula (3), the result is as follows: Fig.10 shown.

[0159] In-vehicle sensor interconnection and image transmission:

[0160] VNC Viewer Remote Connection

[0161] Raspberry Pi is a microcomputer the size of a credit card, so it is also called a "card computer". In this embodiment, the Raspberry Pi 4th generation B model is taken as an example. Its CPU adopts a quad-core Cortex-A72 64-bit BroadcomBCM2711, with a main frequency of 1.5GHz and a memory of 4GB. Compared with the previous generation of Raspberry Pi, in terms of processor, the clock frequency of the quad-core CPU is higher, the memory is larger, and the network speed and wireless network support are more comprehensive. It can not only handle more complex and efficient tasks, but also quickly connect to the network and other devices, with higher performance and wider application prospects.

[0162] First, burn the Raspberry Pi system. Use the official burning tool Raspberry Pi Imager, which is easy to operate and convenient to select the system. Secondly, make various settings. In the operating system selection, choose Raspberry Pi OS (64-bit), which is highly compatible with the Raspberry Pi and has more unrestricted operation. Finally, turn on VNC to achieve remote desktop connection.

[0163] Communication based on the SSH protocol refers to establishing an encrypted and secure communication channel. SSH is a secure network protocol based on the application layer developed by IETF. It is a protocol used to securely log in and execute commands remotely on the network. It provides an encrypted communication channel for remote connections and can prevent eavesdropping and information leakage. The following is a brief overview of the working principle of the SSH protocol. Fig.11 :

[0164] ① Establish a connection: The client initiates a connection request and requests to connect to the SSH server of the remote host. If the SSH server authentication is successful, the SSH session will be established. To ensure data security, the communication between the client and the server will be encrypted. ② Negotiate keys: After the connection is established, key negotiation begins between the client and the server. ③ Authentication: Once the session key is negotiated, the client will send authentication information to the server. Usually includes a username and password, or authentication using a public key and private key pair. ④ Establish a session: Once the identity is established, data can be transmitted securely. All data transmitted through the SSH connection will be encrypted to prevent eavesdropping or tampering. ⑤ Data transmission: After the SSH session is established, the client and server can listen to the specified port (usually 22). ⑥ Terminate the connection: When the communication ends, both parties can choose to close the connection.

[0165] Paramiko is an SSH client library written in Python, which makes it easy to implement SSH connections and operations in Python. Paramiko provides a set of APIs that can be used to establish SSH connections, execute remote commands, transfer files, and other operations. This example uses the Paramiko library to establish an SSH client instance, and then uses the SSH username and password to log in to the remote host, such as Fig.12 shown.

[0166] The goal of the SSH protocol is to provide a secure, encrypted remote connection method to protect user login information and the security of transmitted data. It ensures the security of remote communication through processes such as key negotiation, encrypted communication, and identity authentication.

[0167] Performance Analysis:

[0168] The concealment analysis is:

[0169] Concealment is the core of achieving secure communication. This embodiment will test from two aspects: behavior concealment and content concealment, and comprehensively analyze whether the sending behavior is natural or not and the quality of the image.

[0170] Behavior concealment analysis is:

[0171] In the existing covert communication application process, most of them only consider the concealment in the image steganography process, such as how to make the secret image more natural and how to better resist the detection of steganalysis tools. The concealment of the sending behavior also needs to be improved. In the process of secret message transmission, the concealment of the information sender is directly related to the success of secure communication. The concealment of the sending behavior can not only ensure the smooth transmission of secret information, but also ensure the personal safety of intelligence personnel.

[0172] Combined with the analysis of the vehicle network process, the existing communication behavior is no abnormality compared with the original behavior. The secret image is hidden in the road condition image sent by the vehicle itself at all times, and is uploaded to the car service center together with the road condition image. Our intelligence personnel screened the image at the car service center and obtained the secret information. During the transmission process, the secret image transmission is similar to the existing vehicle network image transmission process, achieving the purpose of covert transmission.

[0173] Content concealment analysis is as follows:

[0174] In the selection of carrier images, images similar to the road conditions are selected according to the characteristics of the vehicle itself. This embodiment takes the BYD Han EV as an example. The vehicle has a built-in "clairvoyance" function, such as Fig.13 , which can provide 360-degree image information about the vehicle's interior and exterior in real time and transmit it to the mobile phone APP. In addition, this model also comes with a driving recorder, which automatically enters the camera state after powering on, and can transmit information when connected to the same WIFI. Selecting a carrier image similar to the road conditions can be integrated with the image information uploaded by the vehicle in real time, further enhancing the concealment of confidential information.

[0175] In the image steganography process, the indistinguishability of the image is crucial. In terms of content concealment, this embodiment analyzes the image quality from two perspectives: the human eye and the professional steganalysis tool.

[0176] From the perspective of the naked eye, a large number of encrypted images are compared with the carrier images. Take one group as an example, Fig.14 , it can be seen that it is visually indistinguishable.

[0177] From the perspective of professional steganalysis, this embodiment uses the steganalysis tool SRNet to analyze the steganographic scheme. The parameter settings are as follows: use the Adam optimizer, the number of rounds is 100, the number of samples input at one time in machine learning is 10, the learning rate is 0.001, and the input image size is 256 pixels. The experimental results are shown in Table 1. It can be seen that compared with the traditional LSB steganography algorithm, the detection degree of this steganography scheme is closer to 50% and the concealment is stronger. Although the reduction of the number of reversible blocks in this method has some impact on the image quality, the image quality is improved due to the increase in the number of convolution blocks of the reversible functions f(·), r(·) and y(·) in each reversible block.

[0178] Table 1

[0179] Steganography Methods Accuracy 4bit-LSB 99.63％ Lightweight reversible neural network steganography scheme 57.64％

[0180] Efficiency Analysis

[0181] This embodiment analyzes the secret image embedding efficiency and the transmission efficiency when the image is automatically transmitted. In terms of embedding efficiency, the method of this embodiment is compared with the HiNet steganography scheme, and the comparison is made in terms of steganography time and training time. In terms of transmission efficiency, five different road condition images are taken as examples, and the image transmission rates of each are respectively calculated through the time function.

[0182] Embedding efficiency

[0183] In order to verify the embedding efficiency of the lightweight reversible neural network image steganography scheme, it is compared with the reversible network-based image hiding scheme HiNet. The two network models are trained for 50 rounds, and the steganography time and training time of each network model are calculated respectively.

[0184] Through experiments, it was found that Fig.15 As shown, the steganographic time and training time of the method in this embodiment are significantly less than HiNet, the steganographic time is shortened by more than 20%, and the training time is shortened by more than 40%. The reduction in the number of reversible blocks leads to a reduction in the burden of the network model, thereby improving the embedding efficiency.

[0185] Sending efficiency

[0186] The average time it takes for a Raspberry Pi to transfer an image to a computer depends on the transfer speed and the size of the image. Use the time library to time and count the image transfer time. Fig.17 (a)- Fig.17 Taking the five test images in (e) as an example, the time from connecting to the network to successfully transmitting each image is calculated by running the code, and the average transmission time is 1.336s. The overall transmission time is within a reasonable range and is not easily detected by the enemy (as shown in Table 2).

[0187] Table 2

[0188] Transferring images Transmission time (ms) Test image 1 2230 Test image 2 1760 Test image three 520 Test image 4 650 Test image five 1520

[0189] The actual transmission time may be affected by factors such as network load and transmission protocol, and may fluctuate to a certain extent.

[0190] This embodiment improves the concealment and security of image transmission behavior from the two perspectives of the steganographic algorithm and the sending behavior, ensures the security of the secret information itself, and securely transmits image data, making it difficult for the enemy to discover and detect the hidden secret information, and can ensure the safety of the sender, which meets the information requirements in the field of military communications.

[0191] In terms of the steganographic algorithm, this embodiment uses a densely connected reversible neural network to achieve image hiding and recovery. The carrier image and the secret image are transformed into a secret image through a forward hidden reversible transformation, and the secret image and the random variable are transformed into a restored image through a reverse recovery reversible transformation, thus achieving the reversibility of the algorithm. From the test results, the method of this embodiment can accurately achieve the steganography and recovery of image information at a high rate.

[0192] In terms of sending behavior, this embodiment is based on the Internet of Things technology and relies on the Internet of Vehicles system. It takes advantage of the fact that the vehicle itself always sends image information to the car service center. By automatically transmitting the image after connecting to a specific WIFI on a fixed road section, the secret image is successfully transmitted to our receiving end. This scenario not only protects the transmission of secret information, but also reduces the risk of exposure of special agents. In addition, this scenario is highly flexible and can reduce regularity and better hide behavior by changing the WIFI setting section.

[0193] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A covert communication method based on Internet of Vehicles, characterized in that: include: Determine a communication recipient object, wherein the communication recipient object includes a vehicle and a background service center; If the communication recipient is a vehicle, identity authentication is performed through vehicle speed difference sampling, a session key is negotiated, and the secret information is encrypted and embedded into the vehicle speed data or multimedia data for covert communication; If the communication recipient is a background service center, the collected image data is used as a carrier to carry the secret image, and the secret image is embedded in the vehicle-mounted image data through covert communication based on a map-hidden-map neural network or covert communication based on images or video frames to achieve covert communication.

2. The method according to claim 1, characterized in that The identity authentication is performed by sampling the vehicle speed difference: Within the driving safety constraint conditions, the speed of the driving vehicles is sampled at the same time intervals to obtain a driving speed difference sequence, and the driving speed difference sequence is used to confirm whether the driving vehicles can communicate with each other. If communication is possible, the message transmission key is calculated.

3. The method according to claim 2, characterized in that The driving safety constraints are: The sender's vehicle speed is recorded as v1, the receiver's vehicle speed is recorded as v2, the sender's vehicle acceleration is recorded as v1'=Δv1 / Δt, t is the driving time; v1"=Δv1' / Δt, v1" indicates the change amplitude of v1', the larger the absolute value of v1" is, the more complex the driving environment of the vehicle is; When the acceleration v1' of the sender's vehicle and the acceleration v2' of the receiver's vehicle are both less than the safety threshold v sec ', and the acceleration change amplitude v1" of the sender's vehicle and the acceleration change amplitude v2" of the receiver's vehicle are both less than the safety threshold v sec ", it indicates that the vehicle is within the driving safety constraint conditions.

4. The method according to claim 3, characterized in that Confirm whether the vehicles can communicate with each other: Set the negotiation password between the sender and the receiver to psd∈{0,1} l , where l is the password length agreed upon by both parties; The vehicle speed is obtained once at the same interval, sampled l times, and the speed difference sequence is obtained by calculation, which is recorded as psd'={psd1',psd2',...,psd l '} l ; If the driving speed difference sequence psd' is exactly the same as the negotiated password psd between the sender and the receiver, the two parties enter the confidential communication stage, otherwise the communication is abandoned; Among them, the driving speed difference Δv=|v1-v2|, when Δv=0, the transmitted information bit is psd i '=0;Δv=e(m / s), e>0, e is the basic unit of information transmission speed; when Δv=e, the transmitted information bit psd i '=1, that is, Δv=round(|v1-v2|), and the round(.) function means rounding the input to an approximate result.

5. The method according to claim 4, characterized in that The secret information is encrypted and embedded into the vehicle speed data for covert communication: Using multi-base encoding, when the embedded data is a b-base number, the embedded information of the hidden information is l s is the message length, and the sampling interval is t sam The starting point of the sampling time is the first hour after the driving safety constraint is met and the command is connected, and then the information obtained by the i-th sampling is obtained, which is recorded as s i ,s i =round(|v1-v2| / e).

6. The method according to claim 5, characterized in that The secret information is encrypted and embedded into multimedia data for covert communication: The embedded information s of the covert communication is embedded in a picture, audio or video to obtain a secret picture, secret audio or secret video, and sent to a receiving vehicle. The receiving vehicle uses an extraction algorithm to extract s and uses a key to decrypt to obtain the secret information.

7. The method according to claim 1, characterized in that The communication process of the background service center is as follows: The collected image data is used as a carrier to steganographically record the secret image, and the image is processed with the image-hidden neural network and the vehicle-mounted image to generate a secret image. The secret image is then sent to the automobile service center, and the automobile service center extracts the secret information through the secret image.

8. The method according to claim 7, characterized in that The process of generating a secret image by processing the image-hiding neural network and the vehicle-mounted image is as follows: Match the size of the carrier image and the secret information, and input the embedding function to generate the secret image, specifically: X Ste =StegoNET(X sec ,X co ) Where, X sec is the secret image, X co is the carrier image, X Ste is the secret image, and StegoNET(.) is the operation function.

9. The method according to claim 8, characterized in that The process of extracting the secret image through the secret image is as follows: The secret image and any random vector are input into the reverse recovery network, preprocessed by discrete wavelet transform to obtain a feature map, and then the feature map is sent to the reversible block of the network. After several layers of reversible block transform, a new feature map is generated, and finally the restored image and the restored carrier image are obtained by inverse wavelet transform.

10. A covert communication system based on Internet of Vehicles, characterized in that: include: The hidden communication module between vehicles and the communication module between vehicles and the backend service center, The covert communication module between vehicles is used to perform identity authentication through speed difference sampling, negotiate session keys, encrypt secret information and embed it into vehicle speed data or multimedia data for covert communication; The communication module of the background service center is used to use the collected image data as a carrier to carry secret information, and embed the secret information into the vehicle-mounted image data through covert communication based on image-hidden-image neural network or covert communication based on image or video frame to realize covert communication.

Images