Image data secure sending method, image data secure receiving method and system

Abstract

The invention provides an image data secure sending method, an image data secure receiving method and an image data secure sending system. The image data secure sending method comprises the following steps: identifying first image data related to content description and second image data unrelated to the content description from to-be-sent image data; performing image coding on the first image data based on the first coding rate to obtain first to-be-encrypted data, and performing image coding on the second image data based on the second coding rate to obtain second to-be-encrypted data; encrypting the first to-be-encrypted data and the second to-be-encrypted data based on the channel state of the transmission channel to obtain a first coding result and a second coding result; and sending the first coding result and the second coding result to an image receiving end through a transmission channel. According to the embodiment of the invention, the method can effectively guarantee the retention of key semantic information, reduces the amount of information required for transmission, and prevents an attacker from cracking through a violent means.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to an image data security method, an image data security receiving method, and a system. Background Art

[0002] In existing communication systems, fixed channel coding schemes are generally used. Since the fixed channel coding schemes require limited verification data during encoding, there is a risk of being cracked by attackers. Summary of the invention

[0003] In view of this, the present application proposes a method for securely sending image data, a method for securely receiving image data, and an apparatus to solve the problem that the channel coding scheme of the existing communication system may be cracked by attackers due to limited verification data.

[0004] The first embodiment of the present application provides a method for securely sending image data, including:

[0005] Identifying, from the image data to be sent, first image data related to the content description and second image data not related to the content description;

[0006] Performing image encoding on the first image data based on a first encoding rate to obtain first data to be encrypted, and performing image encoding on the second image data based on a second encoding rate to obtain second data to be encrypted; the first encoding rate is less than the second encoding rate;

[0007] Encrypting the first to-be-encrypted data and the second to-be-encrypted data respectively based on a channel state of a transmission channel to obtain a first encoding result and a second encoding result;

[0008] The first encoding result and the second encoding result are sent to the image receiving end through the transmission channel.

[0009] The disclosed embodiment adopts image encoding methods with different coding rates for different images, thereby ensuring the retention of key semantic information and reducing the amount of information required for transmission, which can effectively save bandwidth and reduce communication delays. Preferably, the disclosed embodiment encrypts the first data to be encrypted and the second data to be encrypted respectively based on the channel state of the transmission channel to obtain the first encoding result and the second encoding result, which can prevent attackers from cracking through brute force means such as exhaustive enumeration, thereby ensuring physical layer security during communication.

[0010] In the embodiment of the present application, after identifying the first image data related to the content description from the image data to be sent, the method further includes:

[0011] If the signal-to-noise ratio of the transmission channel is greater than a first preset threshold, enlarging the image area of ​​the first image data by a first preset ratio;

[0012] If the signal-to-noise ratio of the transmission channel is less than a second preset threshold, the image area of ​​the first image data is reduced by a second preset ratio; and the second preset threshold is less than the first preset threshold.

[0013] In an embodiment of the present application, the first to-be-encrypted data and the second to-be-encrypted data are encrypted based on the channel state of the transmission channel to obtain a first encoding result and a second encoding result, including:

[0014] quantizing a channel state of the transmission channel to obtain first state quantization data;

[0015] Using the first state quantization data as a first encryption key, and generating first verification data according to the first encryption key;

[0016] dividing the first data to be encrypted into a plurality of first image data packets, and performing channel coding on each first image data packet using the first verification data to obtain a first coding result;

[0017] The second data to be encrypted is divided into a plurality of second image data packets, and each second image data packet is channel-encoded using the first verification data to obtain a second encoding result.

[0018] In an embodiment of the present application, the method further includes:

[0019] Perform error correction code encoding on the first state quantization data to obtain second verification data, and send the second verification data to the image receiving end.

[0020] In the embodiment of the present application, after performing image encoding on the first image data based on the first encoding rate to obtain the first data to be encrypted, the method further includes:

[0021] The first encryption key is embedded into the first data to be encrypted as a digital watermark.

[0022] The disclosed embodiment can effectively prevent attackers from performing replay attacks or forgeries by embedding the first encryption key as a digital watermark, thereby effectively protecting the integrity and security of the data.

[0023] The second aspect of the present application provides a method for securely receiving image data, including:

[0024] Receiving a first encoding result and a second encoding result;

[0025] Decrypting the first encoding result and the second encoding result respectively based on first verification data to obtain first decrypted data and second decrypted data; the first verification data is determined according to a channel state of a transmission channel;

[0026] Performing image decoding on the first decrypted data based on a first decoding rate to obtain first image data related to content description, and performing image decoding on the second decrypted data based on a second decoding rate to obtain second image data unrelated to content description; the first decoding rate corresponds to a first encoding rate, and the second decoding rate corresponds to a second encoding rate;

[0027] The first image data and the second image data are fused to obtain original image data.

[0028] In the embodiment of the present application, before receiving the first encoding result and the second encoding result, the method further includes:

[0029] quantizing a channel state of a transmission channel to obtain second state quantization data;

[0030] receiving second verification data from the image sending end, and performing error correction on the second state quantization data according to the second verification data to obtain first state quantization data;

[0031] The first state quantization data is used as a second encryption key, and first verification data is generated according to the second encryption key; the second encryption key is the same as the first encryption key.

[0032] In an embodiment of the present application, the method further includes:

[0033] extracting a digital watermark from the first decrypted data;

[0034] If the digital watermark is inconsistent with the preset digital watermark, an alarm message is issued; the alarm message indicates that the image data has been tampered with or damaged during the transmission process.

[0035] The third aspect of the present application provides an image data transmission system, including:

[0036] The image sending end identifies first image data related to content description and second image data unrelated to the content description from the image data to be sent; performs image encoding on the first image data based on a first encoding rate to obtain first data to be encrypted, and performs image encoding on the second image data based on a second encoding rate to obtain second data to be encrypted; the first encoding rate is less than the second encoding rate; encrypts the first data to be encrypted and the second data to be encrypted respectively based on the channel state of the transmission channel to obtain a first encoding result and a second encoding result; and sends the first encoding result and the second encoding result to the image receiving end through the transmission channel;

[0037] An image receiving end receives a first coding result and a second coding result; the first coding result and the second coding result are respectively decrypted based on first verification data to obtain first decrypted data and second decrypted data; the first verification data is determined according to the channel state of the transmission channel; the first decrypted data is image-decoded based on a first decoding rate to obtain first image data related to the content description, and the second decrypted data is image-decoded based on a second decoding rate to obtain second image data unrelated to the content description; the first decoding rate corresponds to a first coding rate, and the second decoding rate corresponds to a second coding rate; the first image data and the second image data are merged to obtain original image data.

[0038] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] By reading the detailed description of the preferred embodiment below, various other advantages and benefits will become clear to those of ordinary skill in the art. The accompanying drawings are only used for the purpose of illustrating the preferred embodiment and are not considered to be limitations of the present application. In addition, the same reference symbols are used to represent the same components throughout the accompanying drawings.

[0040] In the attached picture:

[0041] Figure 1 A schematic diagram showing a flow chart of a method for securely sending image data provided by an embodiment of the present application;

[0042] Figure 2 A schematic diagram of a process of identifying first image data related to content description from image data to be sent provided by an embodiment of the present application is shown;

[0043] Figure 3 A schematic diagram of a process for dynamically adjusting the recognition image area according to the channel state provided by an embodiment of the present application is shown;

[0044] Figure 4 A schematic diagram of a process for generating an encryption key based on a channel state provided in an embodiment of the present application is shown;

[0045] Figure 5 A schematic diagram of a process for generating first verification data according to a first encryption key provided in an embodiment of the present application is shown;

[0046] Figure 6 A schematic diagram of a process of embedding an encryption key as a digital watermark into an image provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0047] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0048] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by technicians in the field to which this application belongs.

[0049] The following describes the technical scenarios involved in the embodiments of the present application.

[0050] From the first generation (1G) to the fifth generation (5G), the goals of communication systems have changed dramatically, from analog audio signal transmission to high-speed and low-latency multimedia services. In particular, in 5G, various advanced wireless communication technologies have been applied, such as non-orthogonal multiple access (NOMA), massive multiple-input multiple-output (MIMO), millimeter wave communication, etc. Although 5G can meet most of the requirements of different services with low latency and high data rates, existing technologies may not be able to support many smart applications in super-6G communications. In 6G networks, services such as connected life, brain-computer interaction, virtual reality (VR), augmented reality (AR), and mixed reality (MR) will be supported. The technical requirements of these services are much higher than those of 5G networks, such as 1 to 10Gb / s / m^3 service density, 1Tb / s uplink and downlink data rate, and 0.1ms latency.

[0051] Existing communication technologies have already approached the Shannon physical layer capacity limit, so semantic communication has been proposed as an intelligent communication solution that focuses on the meaning of transmitted messages rather than precise bit stream transmission, and plays an important role in the transition of traditional communication technologies to the future. However, due to the fragility of neural networks and the openness of wireless channels, semantic communication requires additional technical means to ensure the secure transmission of information compared to traditional communication.

[0052] Existing semantic communication systems use a unified compression scheme for data, which cannot achieve high compression rate and retain key semantic information at the same time. To address this problem, the present invention uses a large model to extract key semantic information of the image, and adopts different degrees of compression rate for the key and non-key parts of the image, thereby reducing the amount of data required for transmission while ensuring the retention of key semantic information. This architecture will combine advanced large models with semantic communication technology to ensure data accuracy and compression rate.

[0053] Existing communications use a fixed channel coding scheme, and the required check matrix for coding is limited, which poses a risk of being cracked by attackers. To address this problem, the present invention uses the channel state during communication to dynamically generate a physical layer key, and simultaneously uses a neural network and the physical layer key to dynamically generate the check matrix required for coding. Under the premise of ensuring the quality of communication, it prevents attackers from brute-forcing the physical layer information and protects the transmission security of information.

[0054] The existing data watermark is fixed and unchanging, and once intercepted by an attacker, there is a risk of being forged. To address this problem, the present invention uses the physical layer key as a dynamic data watermark to ensure that the attacker cannot forge the data through brute force cracking, replay attacks, etc., and can effectively protect the security and integrity of the data.

[0055] Specifically, first, the present invention uses a large model to extract the key parts of the image, and can change the area of ​​the key parts according to the quality of the channel. For the key parts, a smaller compression rate is adopted to ensure the retention of key information of the image; for the non-critical parts, a larger compression rate is adopted to ensure the reduction of the amount of transmitted information. Secondly, using the channel estimation results, a neural network-assisted dynamic QC-LDPC coding method is adopted to enhance the security of channel coding. The dynamic coding scheme dynamically adjusts the matrix required for QC-LDPC coding by introducing channel state information, so that the coding process has a high degree of randomness and unpredictability. Finally, in order to prevent the data from being illegally tampered with during transmission, the physical layer key is used to embed watermarks into the data. By embedding watermarks into the data, the receiver can verify the integrity and authenticity of the data, prevent attackers from tampering with and forging the data, and ensure the reliability of the transmitted information.

[0056] According to an embodiment of the present application, an embodiment of a method for securely sending image data is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0057] Embodiment 1:

[0058] In this embodiment, a method for securely sending image data is provided, and the method is applied to an image sending end; Figure 1 is a flowchart of a method for securely sending image data according to an embodiment of the present application. Figure 1 As shown, the process includes the following steps:

[0059] Step S101: identifying first image data related to content description and second image data irrelevant to the content description from image data to be sent.

[0060] In the embodiment of the present disclosure, the content description can be understood as text information, and the first image data can be understood as part of the image data to be sent that is related to the description content of the text information, for example Figure 2 As shown, the large model can identify the first image data w1 and the second image data w2 from the image data w0 to be sent according to the text information "a man holding a notebook". The specific implementation method of the large model is not specifically limited here.

[0061] In some specific embodiments, after step S101, the method further includes:

[0062] If the signal-to-noise ratio of the transmission channel is greater than a first preset threshold, enlarging the image area of ​​the first image data by a first preset ratio;

[0063] If the signal-to-noise ratio of the transmission channel is less than a second preset threshold, the image area of ​​the first image data is reduced by a second preset ratio.

[0064] Among them, the second preset threshold is smaller than the first preset threshold, and the first preset threshold and the second preset threshold can be set according to actual conditions and are not specifically limited here.

[0065] In the embodiments of the present disclosure, for example Figure 3 As shown, the area of ​​the key part (i.e., the first image data) can be dynamically adjusted according to the channel state (e.g., signal-to-noise ratio) of the transmission channel. Specifically, when the signal-to-noise ratio is high, the area of ​​the key part identified by the large model is enlarged to retain more semantic information in the image; when the signal-to-noise ratio is low, the area of ​​the key part is reduced, effectively reducing the amount of data required for transmission and ensuring that the data can successfully reach the receiving end.

[0066] Step S102, performing image encoding on the first image data based on a first encoding rate to obtain first data to be encrypted, and performing image encoding on the second image data based on a second encoding rate to obtain second data to be encrypted; the first encoding rate is lower than the second encoding rate.

[0067] In the disclosed embodiment, the key part (i.e., the first image data) and the non-key part (i.e., the second image data) of the image data to be transmitted are encoded by encoders with different encoding rates, for example: Figure 3 As shown in the encoder A in: For the key part of the picture, it is necessary to effectively ensure that its semantic information is retained as much as possible, so a low compression rate encoder is used to effectively ensure its information; for example Figure 2 As shown in Encoder B in Figure 1: For non-critical parts, a high compression rate encoder is used to effectively reduce the amount of data it occupies, reduce the amount of information transmitted, and ensure low bandwidth occupancy and low latency during communication. At the same time, a unified decoder is deployed at the receiving end of the data, which can effectively restore the data encoded by the encoder.

[0068] The disclosed embodiment adopts encoders with different coding rates to perform image encoding on the critical part (i.e., the first image data mentioned above) and the non-critical part (i.e., the second image data mentioned above) of the image data to be sent, thereby effectively ensuring the retention of key semantic information and reducing the amount of information required for transmission, thereby effectively saving bandwidth and reducing communication delays.

[0069] Step S103: Encrypt the first data to be encrypted and the second data to be encrypted respectively based on the channel state of the transmission channel to obtain a first encoding result and a second encoding result.

[0070] In some specific embodiments, the above step S103 further includes steps S1031 to S1034:

[0071] Step S1031: quantize the channel state of the transmission channel to obtain first state quantization data.

[0072] In the embodiments of the present disclosure, the channel state includes but is not limited to signal scattering information, environmental attenuation information, distance attenuation information, channel gain matrix, instantaneous CSI and statistical CSI. Among them, signal scattering refers to the scattering of the signal during the propagation process, environmental attenuation refers to the attenuation of the signal due to environmental factors, and distance attenuation refers to the attenuation of the signal due to the increase in propagation distance. The channel gain matrix describes the channel properties of each subchannel, that is, the value of each element in the channel gain matrix H. Instantaneous CSI means that the current channel state is known, and statistical CSI represents the statistical characteristics of the channel, such as the type of fading distribution, average channel gain, spatial correlation, etc. The specific implementation method of quantization is not specifically limited in the present invention.

[0073] In the embodiment of the present disclosure, the image transmitting end quantizes the channel state of the transmission channel to obtain first state quantization data, for example Figure 4 As shown in the figure, the “1” on the side of “Alice” in the figure.

[0074] In some specific embodiments, after step S1031, the method further includes:

[0075] Perform error correction code encoding on the first state quantization data to obtain second verification data, and send the second verification data to the image receiving end.

[0076] In the disclosed embodiment, the second check data can be understood as redundant bits obtained by encoding the first state quantized data with an error correction code. The redundant bits can be used by the image receiving end to check the received data, and detect errors by using the relationship between the redundant bits and the data bits established during encoding.

[0077] For example Figure 4 As shown: both the image sending end Alice and the image receiving end Bob perform quantization according to the channel state of the transmission channel to obtain the first state quantization data Bit A and the second state quantization data Bit A'; after quantizing to obtain the second state quantization data Bit A', the image receiving end Bob performs error correction code encoding to obtain the second error correction code encoding result "Bit A'+BitB'", and after quantizing to obtain the first state quantization data BitsA, the image sending end Alice performs error correction code encoding to obtain the first error correction code encoding result "BitsA+BitsB", and sends the redundant bit BitsB therein to the image receiving end Bob, so that the image receiving end Bob processes "Bit A'+Bit B'" according to the redundant bit BitsB to obtain "Bit A'+Bit B", and then obtains the second state quantization data BitsA consistent with the first state quantization data as the second encryption key.

[0078] Step S1032: using the first state quantization data as a first encryption key, and generating first verification data according to the first encryption key.

[0079] In the embodiment of the present disclosure, the first check data refers to the coding matrix of QC-LDPC, that is, the check matrix; the check matrix is ​​generated by the first encryption key, and the specific generation process is as follows:

[0080] For example Figure 5 As shown: the first encryption key BitsA is input into the neural network output basis matrix, and the QC-LDPC encoding matrix H is generated by the unit matrix I and the basis matrix.

[0081] In the disclosed embodiment, the neural network is trained using the first encryption key and the loss function to ensure randomness while ensuring that the communication quality is not affected. Due to the dynamic change of the base matrix, the check matrix during encoding also changes dynamically with the channel state, effectively preventing attackers from brute force cracking and ensuring the physical layer security of the data.

[0082] Step S1033: divide the first data to be encrypted into a plurality of first image data packets, and perform channel coding on each first image data packet using the first verification data to obtain a first coding result.

[0083] In the embodiment of the present disclosure, the first check data refers to a coding matrix of QC-LDPC, that is, a check matrix. Channel coding is performed on each first image data packet using the check matrix.

[0084] Step S1034: divide the second data to be encrypted into a plurality of second image data packets, and perform channel coding on each second image data packet using the first verification data to obtain a second coding result.

[0085] In the embodiment of the present disclosure, the first check data refers to a coding matrix of QC-LDPC, that is, a check matrix. Channel coding is performed on each second image data packet using the check matrix.

[0086] Step S104: sending the first encoding result and the second encoding result to an image receiving end through the transmission channel.

[0087] In some specific embodiments, after the above step S102, the method further includes:

[0088] The first encryption key is embedded into the first data to be encrypted as a digital watermark.

[0089] In the embodiments of the present disclosure, Figure 6 As shown: the image transmitter embeds the first encryption key BitsA as a digital watermark into the first encrypted data through a watermark embedding encoder. The first encryption key BitsA is generated according to the channel state of the transmission channel. Due to the randomness of the channel, the dynamic watermark is difficult to be forged and impersonated by an attacker.

[0090] Embodiment 2:

[0091] In this embodiment, a method for securely receiving image data is provided, which is applied to an image receiving end and comprises the following steps:

[0092] Step S201, receiving a first encoding result and a second encoding result.

[0093] Step S202: decrypt the first encoding result and the second encoding result based on the first verification data to obtain first decrypted data and second decrypted data.

[0094] In the disclosed embodiment, the first coding result is channel-decoded by the first verification data to obtain the corresponding first image data packet, and then the plurality of first image data packets are merged to obtain the first decrypted data. The process of decrypting the second coding result based on the first verification data is similar, and will not be described in detail here.

[0095] The first verification data is determined according to the channel status of the transmission channel.

[0096] In some specific embodiments, the image receiving end generates the first verification data in advance according to the channel state of the transmission channel. The specific process is as follows: Figure 4 As shown: the image receiving end Bob quantizes the second state quantization data Bit A' according to the channel state of the transmission channel; after quantizing the second state quantization data Bit A', the image receiving end Bob performs error correction code encoding to obtain the second error correction code encoding result "Bit A'+Bit B'"; receives the second verification data from the image sending end, that is, the redundant bit BitsB; then processes "Bit A'+Bit B'" according to the redundant bit BitsB to obtain "Bit A+Bit B'", and then obtains the second state quantization data BitsA that is consistent with the first state quantization data as the second encryption key; finally, generates the first verification data according to the second encryption key (the specific process can refer to step S1032 in the above embodiment 1).

[0097] Step S203: performing image decoding on the first decrypted data based on a first decoding rate to obtain first image data related to the content description, and performing image decoding on the second decrypted data based on a second decoding rate to obtain second image data unrelated to the content description.

[0098] In the disclosed embodiment, the first decoding rate corresponds to the first encoding rate, and the second decoding rate corresponds to the second encoding rate.

[0099] Step S204: merge the first image data and the second image data to obtain original image data.

[0100] In some specific embodiments, extracting a digital watermark in the first decrypted data;

[0101] If the digital watermark is inconsistent with the preset digital watermark, an alarm message is issued; the alarm message indicates that the image data has been tampered with or damaged during the transmission process.

[0102] In the embodiments of the present disclosure, for example Figure 6As shown: the digital watermark in the first decrypted data is extracted by the watermark embedding decoder, and the extracted digital watermark is compared with the local watermark. If they are the same, it means that the picture is safe and has not been attacked by attackers; if they are different, it means that there is a problem with the picture.

[0103] The disclosed embodiment can effectively prevent attackers from performing replay attacks or forgeries by embedding an encryption key dynamically generated based on the channel state as a watermark, thereby effectively protecting the integrity and security of the data.

[0104] Embodiment 3:

[0105] In this embodiment, a system for transmitting image data is provided, the system comprising:

[0106] The image sending end identifies first image data related to content description and second image data unrelated to the content description from the image data to be sent; performs image encoding on the first image data based on a first encoding rate to obtain first data to be encrypted, and performs image encoding on the second image data based on a second encoding rate to obtain second data to be encrypted; the first encoding rate is less than the second encoding rate; encrypts the first data to be encrypted and the second data to be encrypted respectively based on the channel state of the transmission channel to obtain a first encoding result and a second encoding result; and sends the first encoding result and the second encoding result to the image receiving end through the transmission channel;

[0107] An image receiving end receives a first coding result and a second coding result; the first coding result and the second coding result are respectively decrypted based on first verification data to obtain first decrypted data and second decrypted data; the first verification data is determined according to the channel state of the transmission channel; the first decrypted data is image-decoded based on a first decoding rate to obtain first image data related to the content description, and the second decrypted data is image-decoded based on a second decoding rate to obtain second image data unrelated to the content description; the first decoding rate corresponds to a first coding rate, and the second decoding rate corresponds to a second coding rate; the first image data and the second image data are merged to obtain original image data.

[0108] Embodiment 4:

[0109] The disclosed embodiments provide a 6G-oriented adaptive semantic extraction and dynamic channel coding security architecture, which aims to provide an efficient and secure communication environment to ensure data integrity and data security of the source channel. The various modules work together. The architecture will combine the current advanced large language models, neural networks, physical layer keys and other technologies to ensure data security and low latency of communication. The following is a description of the 6G-oriented adaptive semantic extraction and dynamic channel coding security architecture, including its basic components:

[0110] Adaptive semantic extraction component (i.e. Figure 2 The blue component in

[0111] The large model is used to identify the key semantic information in the multimodal data. For the key semantic information, a lower compression rate is used to effectively ensure the integrity of the key part of the data; for the non-key semantic information, a higher compression rate is used to ensure the reduction of data volume, achieve a higher degree of compression, and effectively reduce communication delay. In addition, the area of ​​the key part identified by the large model can be changed according to the channel state. When the communication quality is good, the area of ​​the key part is increased to retain as much semantic information as possible; when the communication quality is poor, the area of ​​the key part is reduced to increase the compression rate as much as possible to ensure low delay during communication.

[0112] Dynamic channel coding component (i.e. Figure 2 Green components in

[0113] The dynamic physical layer key is generated by using the channel state, and the key is input into the neural network to obtain the check matrix required for dynamic channel coding. Due to the dynamic coding of the channel, the channel state and the check matrix also change accordingly, which effectively prevents attackers from brute force cracking and ensures the physical layer security of the data.

[0114] Dynamic channel watermarking component (i.e. Figure 2 The red component in

[0115] The channel state is used to generate a dynamic physical layer key, which is then embedded in the image as a watermark. Due to the randomness of the physical layer key, the watermark is constantly changing, which can effectively prevent watermark forgery / leakage and ensure data integrity and security.

[0116] By extracting key semantic information through a large model, the amount of communication data is effectively reduced, while ensuring that the semantic information is fully preserved, and the communication experiment is reduced. At the same time, the constantly changing characteristics of the channel are used to generate a dynamic physical layer key, which is used to generate a dynamic channel coding matrix, preventing attackers from brute-forcing it, effectively ensuring the physical layer security of the data. Finally, the dynamic channel watermark effectively prevents attackers from performing replay attacks or forgery attacks, ensuring the security and integrity of data at the source level. Based on this security architecture, through a variety of defense measures, semantic communication for 6G can more safely and reliably serve the diverse application scenarios of the future.

[0117] like Figure 3 As shown in the figure: In response to the ultra-low latency requirement of 6G, we combine the large model to extract the semantic information of the key parts. The following is the specific method of how the adaptive semantic extraction component works:

[0118] Key semantic information identification: For multimodal data (image + text), the information expressed in the text is the key part of the image. As shown in the figure above, the text content is "a man holding a notebook", and the key part of the image should be consistent with the text description. Input the image into the big model and use the text content as part of the prompt word. The complete prompt word is designed as: "According to the following description, mark the part of the image corresponding to: text content". Since the reasoning ability of the big model is much stronger than the traditional model, it can effectively represent the key semantic information part of the image.

[0119] Image encoding and decoding: For the key and non-key parts of the image, encoders with different compression rates are used. For the key parts of the image, its semantic information needs to be effectively preserved as much as possible, so an encoder with a low compression rate is used to effectively preserve its information; for the non-key parts, an encoder with a high compression rate is used to effectively reduce the amount of data it occupies, reduce the amount of information transmitted, and ensure low bandwidth occupancy and low latency during communication. At the same time, we deploy a unified decoder at the receiving end of the data, which can effectively restore the data encoded by the encoder.

[0120] Area Adaptation: The area of ​​the key part can be dynamically adjusted according to the quality of the channel, that is, the signal-to-noise ratio, etc. Specifically, when the signal-to-noise ratio is high (good channel), we enlarge the key part identified by the large model to retain more semantic information in the image; when the signal-to-noise ratio is low, we reduce the area of ​​the key part, effectively reducing the amount of data required for transmission and ensuring that the data can successfully reach the receiving end.

[0121] In general, this component effectively ensures the retention of key semantic information and reduces the amount of information required for transmission, which can effectively save bandwidth and reduce communication delays.

[0122] In order to ensure the physical layer security of 6G communication, we combine channel state, neural network and other technologies to generate a dynamic channel coding matrix. The following is the specific working method of the dynamic channel coding component:

[0123] Physical layer key generation:

[0124] The physical layer key generation process is as follows: Figure 4 As shown in the figure, the communicating parties Alice and Bob first quantize the channel state and obtain a string of similar data Bit A and Bit A'. Both parties encode the string of data with error correction code, and then Alice transmits the encoded redundant bits to Bob. Bob uses the redundant bits to correct the errors in his own data and obtains the data Bit A that is consistent with Alice. This data is the physical layer key. Since the channel is constantly changing dynamically, the channel state is also dynamically encoded, and the generated physical layer key is also dynamic and random, which can effectively prevent attackers from forging and cracking.

[0125] Basis matrix generation: For example Figure 5 As shown in Figure 1, the obtained dynamic physical layer key is input into a dedicated neural network to obtain a part of the basis matrix. The neural network is trained using a large number of physical layer keys and a reasonable loss function to ensure randomness while ensuring that the communication quality is not affected.

[0126] Generation of coding matrix: e.g. Figure 5 As shown in the figure, the coding matrix of QC-LDPC is generated by the unit matrix and the base matrix. Due to the dynamic change of the base matrix, the check matrix during coding also changes dynamically with the channel state, which effectively prevents attackers from brute force cracking and ensures the physical layer security of the data.

[0127] In general, this component generates a dynamic and random check matrix through dynamic channel status to prevent attackers from cracking the system through brute force means such as exhaustive enumeration, thus ensuring the physical layer security during communication.

[0128] For example Figure 6 As shown: In order to ensure data integrity and security during 6G communication, the physical layer key is embedded in the image as a watermark to prevent attackers from forging it. The following is the specific method of how the dynamic channel watermark component works:

[0129] Watermark embedding encoder: Using neural network, the dynamic physical layer key is used as the key part of watermark embedding into the image. Due to the randomness of the channel, the dynamic watermark is difficult to be forged and impersonated by attackers.

[0130] Watermark embedding decoder: Use a neural network to extract the watermark from the image and compare the extracted watermark with the local watermark. If they are the same, it means that the image is safe and has not been attacked by attackers; if they are different, it means that there is a problem with the image.

[0131] In general, embedding the dynamic physical layer key as a watermark can effectively prevent attackers from performing replay attacks or forgeries, and effectively protect the integrity and security of the data.

[0132] The three components work together to effectively reduce the amount of data transmitted while ensuring that the key parts of the data are not affected. At the same time, they protect the physical layer security and source layer security of the data, achieving end-to-end security protection and providing a feasible security structure for 6G scenarios such as ultra-large connections and ultra-low latency.

[0133] The embodiments of the present disclosure have the following technical effects:

[0134] An adaptive semantic extraction solution based on a large model, which combines the latest large model and semantic communication technology, dynamically identifies the key and non-key parts of the image according to the channel status, so as to simultaneously retain the key parts of the data and highly compress the data, effectively ensuring that the amount of transmitted data is reduced while ensuring that semantic information is not lost.

[0135] The dynamic channel coding scheme combines a neural network with a dynamic physical layer key to dynamically generate the check matrix required for channel coding (QC-LDPC) encoding. Under the premise of ensuring that the communication quality is not affected, it makes it impossible for attackers to predict the matrix during encoding, effectively ensuring the physical layer security of the data.

[0136] The dynamic data watermarking solution uses a dynamic physical layer key as a watermark to embed into the data, ensuring that attackers cannot forge the watermark through brute force cracking, replay attacks, and other attack methods, effectively ensuring the security and integrity of the data.

[0137] It should be noted that:

[0138] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known structures and technologies are not shown in detail so as not to obscure the understanding of this description.

[0139] Similarly, it should be understood that in order to streamline the present application and help understand one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present application, the various features of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be interpreted as reflecting the following schematic diagram: the claimed application requires more features than the features clearly stated in each claim. More specifically, as reflected in the claims below, the inventive aspects are less than all the features of the single embodiment disclosed above. Therefore, the claims following the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself serves as a separate embodiment of the present application.

[0140] In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present application and form different embodiments. For example, in the claims below, any one of the claimed embodiments may be used in any combination.

[0141] The above is only a preferred specific implementation 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 shall be based on the protection scope of the claims.

Claims

1. A method for securely sending image data, characterized in that: The method comprises: Identifying, from the image data to be sent, first image data related to the content description and second image data not related to the content description; Performing image encoding on the first image data based on a first encoding rate to obtain first data to be encrypted, and performing image encoding on the second image data based on a second encoding rate to obtain second data to be encrypted; the first encoding rate is less than the second encoding rate; Encrypting the first to-be-encrypted data and the second to-be-encrypted data respectively based on a channel state of a transmission channel to obtain a first encoding result and a second encoding result; The first encoding result and the second encoding result are sent to the image receiving end through the transmission channel.

2. The method according to claim 1, characterized in that After identifying first image data related to the content description from the image data to be sent, the method further includes: If the signal-to-noise ratio of the transmission channel is greater than a first preset threshold, enlarging the image area of ​​the first image data by a first preset ratio; If the signal-to-noise ratio of the transmission channel is less than a second preset threshold, the image area of ​​the first image data is reduced by a second preset ratio; and the second preset threshold is less than the first preset threshold.

3. The method according to claim 1 or 2, characterized in that: Encrypting the first to-be-encrypted data and the second to-be-encrypted data respectively based on the channel state of the transmission channel to obtain a first encoding result and a second encoding result, including: quantizing a channel state of the transmission channel to obtain first state quantization data; Using the first state quantization data as a first encryption key, and generating first verification data according to the first encryption key; dividing the first data to be encrypted into a plurality of first image data packets, and performing channel coding on each first image data packet using the first verification data to obtain a first coding result; The second data to be encrypted is divided into a plurality of second image data packets, and each second image data packet is channel-encoded using the first verification data to obtain a second encoding result.

4. The method according to claim 3, characterized in that The method further comprises: Perform error correction code encoding on the first state quantization data to obtain second verification data, and send the second verification data to the image receiving end.

5. The method according to claim 3, characterized in that: After performing image encoding on the first image data based on the first encoding rate to obtain first data to be encrypted, the method further includes: The first encryption key is embedded into the first data to be encrypted as a digital watermark.

6. A method for securely receiving image data, characterized in that: The method comprises: Receiving a first encoding result and a second encoding result; Decrypting the first encoding result and the second encoding result respectively based on first verification data to obtain first decrypted data and second decrypted data; the first verification data is determined according to a channel state of a transmission channel; Performing image decoding on the first decrypted data based on a first decoding rate to obtain first image data related to content description, and performing image decoding on the second decrypted data based on a second decoding rate to obtain second image data unrelated to content description; the first decoding rate corresponds to a first encoding rate, and the second decoding rate corresponds to a second encoding rate; The first image data and the second image data are fused to obtain original image data.

7. The method according to claim 6, characterized in that Before receiving the first encoding result and the second encoding result, the method further includes: quantizing a channel state of a transmission channel to obtain second state quantization data; receiving second verification data from the image sending end, and performing error correction on the second state quantization data according to the second verification data to obtain first state quantization data; The first state quantization data is used as a second encryption key, and first verification data is generated according to the second encryption key; the second encryption key is the same as the first encryption key.

8. The method according to claim 6, characterized in that The method further comprises: extracting a digital watermark from the first decrypted data; If the digital watermark is inconsistent with the preset digital watermark, an alarm message is issued; the alarm message indicates that the image data has been tampered with or damaged during the transmission process.

9. An image data transmission system, characterized in that: The system comprises: The image sending end identifies first image data related to content description and second image data unrelated to the content description from the image data to be sent; performs image encoding on the first image data based on a first encoding rate to obtain first data to be encrypted, and performs image encoding on the second image data based on a second encoding rate to obtain second data to be encrypted; the first encoding rate is less than the second encoding rate; encrypts the first data to be encrypted and the second data to be encrypted respectively based on the channel state of the transmission channel to obtain a first encoding result and a second encoding result; and sends the first encoding result and the second encoding result to the image receiving end through the transmission channel; An image receiving end receives a first coding result and a second coding result; the first coding result and the second coding result are respectively decrypted based on first verification data to obtain first decrypted data and second decrypted data; the first verification data is determined according to the channel state of the transmission channel; the first decrypted data is image-decoded based on a first decoding rate to obtain first image data related to the content description, and the second decrypted data is image-decoded based on a second decoding rate to obtain second image data unrelated to the content description; the first decoding rate corresponds to a first coding rate, and the second decoding rate corresponds to a second coding rate; the first image data and the second image data are merged to obtain original image data.

Images