Anti-leakage casual transmission method and system based on large language model

Abstract

The invention provides an anti-leakage casual transmission method and system based on a large language model. The anti-leakage casual transmission method based on the large language model comprises the following steps: a sender Sen processes a to-be-transmitted message by using a non-adaptive anti-leakage LR-OTkn protocol, and sends the processed to-be-transmitted message to a receiver in an encrypted form; and starting an adaptive anti-leakage LR-OTkn protocol to perform data transmission on the processed to-be-transmitted message when monitoring that a channel for transmitting the to-be-transmitted message is attacked in a transmission process of the processed to-be-transmitted message. The system comprises modules corresponding to the steps of the method.

Description

Technical Field

[0001] The invention proposes a large language model-based anti-leakage oblivious transmission method and system, belonging to the technical field of encrypted data transmission. Background Art

[0002] With the rapid development of artificial intelligence technology, large language models (LLMs) such as the GPT series have become core tools in the field of natural language processing. These models can generate high-quality natural language texts through training with massive data, and are widely used in many fields such as chatbots, content creation, and intelligent customer service. However, with the widespread application of LLMs, their data privacy and security issues have become increasingly prominent, especially the risk of data leakage has become a key factor restricting its further development. Large language models will be exposed to a large amount of sensitive data during the training process, including personal privacy information, commercial secrets, etc. Once these data are leaked, it will bring immeasurable losses to individuals and enterprises. For example, LLM applications such as ChatGPT have experienced user privacy data leakage incidents because vulnerabilities in the open source code base allowed some users to see the chat records of other users. Such incidents not only undermine user trust, but also cause widespread concern in the industry about the security of LLM data.

[0003] At present, the protection measures against LLM data leakage mainly include data encryption, access control, security audit, etc. However, these traditional methods are often powerless in the face of complex and changeable network attacks. In particular, when the transmission channel is attacked, the traditional non-adaptive anti-leakage protocol may not be able to effectively protect the integrity and confidentiality of the data. Summary of the invention

[0004] The present invention provides a method and system for inadvertent transmission against leakage based on a large language model, which is used to solve the problem of poor data transmission security in the prior art. The technical solutions adopted are as follows:

[0005] A method for inadvertent transmission based on a large language model to resist leakage, the method comprising:

[0006] The sender Sen uses non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form;

[0007] During the transmission of the processed message to be transmitted, if it is detected that the channel used to transmit the message to be transmitted is attacked, adaptive anti-leakage is initiated. The protocol performs data transmission on the processed message to be transmitted.

[0008] Furthermore, the sender Sen uses non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form, including:

[0009] The sender Sen calls the Setup algorithm to generate a public key and a master key pair (mpk, msk), and generates system parameters G(1 λ ); wherein the structures of the public key mpk and the master key msk are as follows:

[0010]

[0011] Where N = p1p2p3, and are all different λ-bit prime numbers;

[0012] The sender Sen sends the public key mpk to the receiver Rec;

[0013] After receiving the public key mpk sent by the sender Sen, the receiver Rec obtains k identities σ by executing a leakage-resistant blind extraction protocol with the sender Sen. i ; Wherein, the value of i is: 1≤i≤k, which is the corresponding identity private key;

[0014] The sender Sen encrypts all messages to be transmitted, generates encrypted ciphertext, and sends all ciphertext to the receiver Rec;

[0015] The receiver Rec uses the identity σ i , and the identity private key corresponding to 1≤i≤k decrypts the ciphertext to obtain the message value corresponding to the message to be transmitted, where the message value is

[0016] Furthermore, during the transmission of the processed message to be transmitted, when it is detected that the channel used to transmit the message to be transmitted is attacked, adaptive anti-leakage is initiated. The protocol performs data transmission on the processed message to be transmitted, including:

[0017] During the transmission of the processed message to be transmitted, real-time monitoring is performed on the channel used to transmit the message to be transmitted to determine whether the channel is attacked;

[0018] When the channel used to transmit the message to be transmitted encounters an attack, the current non-adaptive anti-leakage The data transmission process of the protocol for the processed message to be transmitted;

[0019] The current non-adaptive anti-leakage After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted.

[0020] Further, in the present non-adaptive anti-leakage After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted, including:

[0021] The sender Sen randomly selects W1,···,W n ← R G T ;

[0022] Using non-adaptive anti-leakage The encryption algorithm in the protocol is W1,···,W n To encrypt, for any i∈[1,n], let A i W i The corresponding ciphertext;

[0023] For all 1≤i≤n, the sender Sen calculates B i =H(W i )⊕M i , and set C i =(A i ,B i ), where H:G T →G T It is a collision-resistant Hash function;

[0024] The sender Sen sends the public key mpk and the ciphertext C1, ..., C n Send to the receiver Rec;

[0025] The receiver Rec receives the public key mpk and the ciphertext C1, ..., C n After that, the receiver Rec uses its index value σ i Execute the anti-leakage blind extraction protocol with the sender Sen to obtain the corresponding decryption key sk σi ;

[0026] After obtaining the decryption key, the receiver Rec obtains the message value corresponding to the transmitted message through decryption, where the message value is

[0027] Furthermore, the anti-leakage inadvertent transmission method also includes:

[0028] Continuously monitoring the transmission channel and collecting channel status information, wherein the channel status information includes signal strength, error rate, and packet loss rate;

[0029] Inputting the channel state information into a large language model, and analyzing the collected channel data through the natural language processing and analysis functions of the large language model LLM to obtain the difference value between the historical data and the current channel state information;

[0030] The channel security evaluation parameter is obtained by the difference between the historical data and the current channel state information, wherein the channel security evaluation parameter is obtained by the following formula:

[0031]

[0032] Where SCP represents the channel security evaluation parameter; P e , L e and S e Respectively represent the difference between the current error rate, packet loss rate and signal strength and the maximum error rate, maximum packet loss rate and minimum signal strength in the historical data; P m , L m and S y Respectively represent the maximum error rate, the maximum packet loss rate and the minimum signal strength in the historical data; x, y and z represent the first weight value, the second weight value and the third weight value respectively; and the first weight value, the second weight value and the third weight value are obtained by the following formula:

[0033]

[0034] Wherein, x, y and z represent the first weight value, the second weight value and the third weight value respectively; ΔP e Indicates the floating change rate of the current error rate compared to the median error rate in historical data; ΔL e Indicates the floating change rate of the current packet loss rate compared to the median packet loss rate in historical data; ΔS e Indicates the floating change rate of the current signal strength compared to the median signal strength in historical data;

[0035] When the security evaluation parameter is lower than a preset security parameter threshold, it is determined that the channel security quality is poor, and a channel security warning is issued and the encrypted transmission channel is replaced.

[0036] A large language model-based, anti-leakage oblivious transmission system, the large language model-based, anti-leakage oblivious transmission system comprising:

[0037] The first anti-leakage processing module is used by the sender Sen to use non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form;

[0038] The second anti-leakage processing module is used to start adaptive anti-leakage when it is detected that the channel used to transmit the message to be transmitted is attacked during the transmission of the processed message to be transmitted. The protocol performs data transmission on the processed message to be transmitted.

[0039] Furthermore, the first anti-leakage processing module includes:

[0040] The master key pair generation module is used by the sender Sen to call the Setup algorithm to generate the public key and master key pair (mpk, msk), and generate the system parameter G (1 λ ); wherein the structures of the public key mpk and the master key msk are as follows:

[0041]

[0042] Where N = p1p2p3, and are all different λ-bit prime numbers;

[0043] A first data sending module, used for the sender Sen to send the public key mpk to the receiver Rec;

[0044] The first blind extraction protocol module is used for the receiver Rec to obtain k identities σ by executing a leakage-resistant blind extraction protocol with the sender Sen after receiving the public key mpk sent by the sender Sen. i ; Wherein, the value of i is: 1≤i≤k, which is the corresponding identity private key;

[0045] The first encryption processing module is used by the sender Sen to encrypt all messages to be transmitted, generate encrypted ciphertexts, and send all ciphertexts to the receiver Rec;

[0046] The first decryption module is used by the receiver Rec to use the identity σ i , and the identity private key corresponding to 1≤i≤k decrypts the ciphertext to obtain the message value corresponding to the message to be transmitted, where the message value is

[0047] Furthermore, the second anti-leakage processing module includes:

[0048] A channel attack monitoring and judging module, used to monitor the channel used to transmit the processed message to be transmitted in real time during the transmission process, and judge whether the channel is attacked;

[0049] The data transmission process stopping module is used to stop the current non-adaptive anti-leakage when the channel used to transmit the message to be transmitted is attacked. The data transmission process of the protocol for the processed message to be transmitted;

[0050] The adaptive anti-leakage protocol startup module is used to After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted.

[0051] Furthermore, the adaptive anti-leakage protocol startup module includes:

[0052] Random selection module, used by the sender Sen to randomly select W1,···,W n ← R G T ;

[0053] The second encryption processing module is used to utilize non-adaptive anti-leakage The encryption algorithm in the protocol is W1,···,W n To encrypt, for any i∈[1,n], let A i W i The corresponding ciphertext;

[0054] The data processing module is used to calculate B for all 1≤i≤n, the sender Sen i =H(W i )⊕M i , and set C i =(A i ,B i ), where H:G T →G T It is a collision-resistant Hash function;

[0055] The second data sending module is used for the sender Sen to send the public key mpk and the ciphertext C1, ..., C n Send to the receiver Rec;

[0056] The second blind extraction protocol module is used for the receiver Rec to receive the public key mpk and the ciphertext C1, ..., C n After that, the receiver Rec uses its index value σ i Execute the anti-leakage blind extraction protocol with the sender Sen to obtain the corresponding decryption key sk σi ;

[0057] The second decryption module is used for the receiver Rec to obtain the message value corresponding to the transmission message by decryption after obtaining the decryption key, wherein the message value is

[0058] Furthermore, the anti-leakage inadvertent transmission system also includes:

[0059] A status information monitoring module, used to continuously monitor the transmission channel and collect channel status information, wherein the channel status information includes signal strength, error rate and packet loss rate;

[0060] A difference value acquisition module, used to input the channel state information into a large language model, and analyze the collected channel data through the natural language processing and analysis functions of the large language model LLM to obtain the difference value between the historical data and the current channel state information;

[0061] The channel security evaluation parameter acquisition module is used to obtain the channel security evaluation parameter through the difference between the historical data and the current channel state information, wherein the channel security evaluation parameter is obtained through the following formula:

[0062]

[0063] Where SCP represents the channel security evaluation parameter; P e , L e and S e Respectively represent the difference between the current error rate, packet loss rate and signal strength and the maximum error rate, maximum packet loss rate and minimum signal strength in the historical data; P m , L m and S y Respectively represent the maximum error rate, the maximum packet loss rate and the minimum signal strength in the historical data; x, y and z represent the first weight value, the second weight value and the third weight value respectively; and the first weight value, the second weight value and the third weight value are obtained by the following formula:

[0064]

[0065] Wherein, x, y and z represent the first weight value, the second weight value and the third weight value respectively; ΔP e Indicates the floating change rate of the current error rate compared to the median error rate in historical data; ΔL e Indicates the floating change rate of the current packet loss rate compared to the median packet loss rate in historical data; ΔS e Indicates the floating change rate of the current signal strength compared to the median signal strength in historical data;

[0066] The channel security determination module is used to determine that the channel security quality is poor when the security evaluation parameter is lower than a preset security parameter threshold, and to perform a channel security warning and change the encrypted transmission channel.

[0067] Beneficial effects of the present invention:

[0068] The present invention proposes a method and system for inadvertent transmission based on a large language model that is resistant to leakage. By combining non-adaptive and adaptive anti-leakage protocols, the method can fully protect the data security during the transmission process and effectively resist various network attacks. The introduction of the adaptive anti-leakage protocol enables the system to dynamically adjust the protection strategy according to the actual attack situation, thereby improving the flexibility and pertinence of the protection. Through the comprehensive use of multiple security mechanisms such as encryption and obfuscation, the risk of data being leaked during the transmission process is greatly reduced. Since this method can effectively ensure the security of data transmission, users can use applications and services based on large language models with greater confidence, thereby improving user experience and trust. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 is a flow chart of the method of the present invention;

[0070] Figure 2 is a system block diagram of the system of the present invention;

[0071] Figure 3 The non-adaptive anti-leakage of the present invention The algorithmic structure of the protocol;

[0072] Figure 4 The invention provides an adaptable anti-leakage The algorithm structure of the protocol. DETAILED DESCRIPTION

[0073] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0074] The embodiment of the present invention proposes a method for inadvertent transmission based on a large language model to resist leakage, such as Figure 1 , Figure 3 and Figure 4 As shown, the anti-leakage oblivious transmission method based on a large language model includes:

[0075] S1, the sender Sen uses non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form;

[0076] S2. When the processed message to be transmitted is detected during transmission, the channel used to transmit the message to be transmitted is attacked, and adaptive anti-leakage is initiated. The protocol performs data transmission on the processed message to be transmitted.

[0077] The working principle of the above technical solution is as follows: the sender Sen first uses a non-adaptive anti-leakage protocol to process the message to be transmitted, and protects the message through encryption or other security mechanisms (such as obfuscation, hiding, etc.) to ensure that the message content will not be easily leaked before transmission. The processed message is sent to the receiver in encrypted form. The encryption process uses a high-strength encryption algorithm to ensure that even if it is intercepted during transmission, the attacker cannot easily decrypt and obtain the original information.

[0078] During the message transmission process, the system monitors in real time whether the channel used to transmit the message is attacked. This is usually achieved through security monitoring equipment or software, which can detect and analyze abnormal traffic in the channel, data packet tampering and other signs. Once the channel is detected to be attacked, the system immediately starts the adaptive anti-leakage protocol. Unlike non-adaptive protocols, adaptive protocols can dynamically adjust the protection strategy according to the type and intensity of the attack to provide more effective security protection. Under the protection of the adaptive anti-leakage protocol, the processed message to be transmitted can continue to be safely transmitted to the recipient, and the confidentiality and integrity of the data can be maintained even in the face of attacks.

[0079] The effect of the above technical solution is: by combining non-adaptive and adaptive anti-leakage protocols, this method can fully protect the data security during transmission and effectively resist various network attacks. The introduction of adaptive anti-leakage protocols enables the system to dynamically adjust the protection strategy according to the actual attack situation, thereby improving the flexibility and pertinence of protection. Through the comprehensive use of multiple security mechanisms such as encryption and obfuscation, the risk of data leakage during transmission is greatly reduced. Since this method can effectively ensure the security of data transmission, users can use applications and services based on large language models with greater confidence, thereby improving user experience and trust.

[0080] In summary, this technical solution provides a more comprehensive, flexible and effective security protection measure for data transmission based on large language models by innovatively combining non-adaptive and adaptive anti-leakage protocols.

[0081] In one embodiment of the present invention, the sender Sen uses non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form, including:

[0082] S101, the sender Sen calls the Setup algorithm to generate a public key and a master key pair (mpk, msk), and generates system parameters G (1 λ ); wherein the structures of the public key mpk and the master key msk are as follows:

[0083]

[0084] Where N = p1p2p3, and are all different λ-bit prime numbers;

[0085] S102, the sender Sen sends the public key mpk to the receiver Rec;

[0086] S103, after receiving the public key mpk sent by the sender Sen, the receiver Rec obtains k identities σ by executing a leakage-resistant blind extraction protocol with the sender Sen i ; Wherein, the value of i is: 1≤i≤k, which is the corresponding identity private key;

[0087] S104, the sender Sen encrypts all messages to be transmitted, generates encrypted ciphertexts, and sends all ciphertexts to the receiver Rec;

[0088] S105, the receiver Rec uses the identity σ i , and the identity private key corresponding to 1≤i≤k decrypts the ciphertext to obtain the message value corresponding to the message to be transmitted, where the message value is

[0089] The working principle of the above technical solution is as follows: the sender Sen first executes the Setup algorithm to generate a pair of public key mpk and master key msk, as well as system parameters G(1 λ ). Here, the structure of the public key and the master key is designed with security in mind, especially N is obtained by multiplying three different λ-bit prime numbers, which increases the difficulty of cracking. The public key mpk is sent to the receiver Rec for subsequent encryption and decryption operations.

[0090] After receiving the public key mpk, the receiver Rec performs a leakage-resistant blind extraction protocol with the sender Sen. This protocol ensures that the receiver can obtain k identity private keys σi (where 1≤i≤k) from the sender without revealing its identity information. These identity private keys will be used for subsequent decryption operations.

[0091] The sender Sen encrypts all messages to be transmitted and generates encrypted ciphertext. The encryption process here relies on the public key mpk, ensuring that only the receiver holding the corresponding private key can decrypt it. The encrypted ciphertext is sent to the receiver Rec.

[0092] The receiver Rec uses the identity private key σi (where 1≤i≤k) obtained previously through the blind extraction protocol to decrypt the received ciphertext. Since each identity private key corresponds to a possible message value, the receiver can decrypt the message set {Mσ1,...,Mσk}.

[0093] The effect of the above technical solution is: by using a non-adaptive anti-leakage protocol, the method takes security measures in key generation, distribution, message encryption and decryption, etc., effectively preventing data leakage during transmission. The generation of public and private keys is based on complex mathematical structures (such as N multiplied by three different prime numbers), which increases the difficulty of cracking. The blind extraction protocol ensures that the recipient's identity information will not be leaked in the process of obtaining the private key, further enhancing security.

[0094] The receiver can decrypt the corresponding message value according to the different identity private keys σi they hold. This design provides a flexible access control mechanism. The sender can use different encryption strategies and identity private keys for different receivers or different message contents to achieve refined data management.

[0095] Although the encryption and decryption process may involve complex mathematical operations, this method can ensure the efficiency of data transmission and decryption by optimizing the algorithm design and implementation. The receiver can process the decryption operations of multiple ciphertexts in parallel, improving the overall processing efficiency.

[0096] This method enhances users' trust in the data transmission process by providing strong security protection. The efficient transmission and decryption process also improves the user experience, allowing users to use services and applications based on this technology more smoothly.

[0097] In one embodiment of the present invention, during the transmission of the processed message to be transmitted, when it is detected that the channel used to transmit the message to be transmitted is attacked, adaptive anti-leakage is initiated. The protocol performs data transmission on the processed message to be transmitted, including:

[0098] S201, during the transmission of the processed message to be transmitted, monitoring in real time the channel used to transmit the message to be transmitted to determine whether the channel is under attack;

[0099] S202: When the channel used to transmit the message to be transmitted encounters an attack, the current non-adaptive anti-leakage The data transmission process of the protocol for the processed message to be transmitted;

[0100] S203, in the current non-adaptive anti-leakage After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted.

[0101] The working principle of the above technical solution is: during the process of transmitting the processed message to be transmitted through the channel, the system will monitor the channel in real time. This usually involves analyzing multiple indicators such as the traffic flow, data packet content, abnormal behavior, etc. in the channel to determine whether the channel is under attack. Once an anomaly is detected, the system will further analyze whether these anomalies constitute a threat to the security of data transmission, that is, determine whether the channel is indeed under attack.

[0102] If the judgment result is that the channel is indeed attacked, the system will immediately stop the ongoing data transmission process based on the non-adaptive anti-leakage protocol. In order to prevent attackers from further stealing or tampering with data by exploiting the loopholes of the current protocol. After stopping the data transmission of the non-adaptive protocol, the system will quickly switch to the adaptive anti-leakage protocol. The adaptive protocol can dynamically adjust its protection strategy according to the real-time situation such as the type and intensity of the attack to more effectively resist the attack. Under the protection of the adaptive protocol, the processed message to be transmitted will restart the data transmission process until it is safely delivered to the recipient.

[0103] The effect of the above technical solution is: through real-time monitoring and dynamic switching of protocols, the method can quickly respond to changes in channel security status during data transmission and effectively resist sudden attacks. The introduction of the adaptive anti-leakage protocol enables the system to flexibly adjust the protection strategy according to actual conditions, improving the pertinence and effectiveness of defense.

[0104] When a channel is attacked, the non-adaptive protocol data transmission is stopped in time to prevent attackers from stealing or tampering with data by exploiting protocol vulnerabilities. After switching to the adaptive protocol, the system can continue to transmit data with higher security, thereby reducing the risk of data leakage.

[0105] Real-time monitoring and protocol switching mechanisms ensure the continuity and stability of the data transmission process, and can quickly restore transmission even in the event of an attack. This approach improves the overall reliability and availability of the data transmission system, reducing service interruptions and data loss caused by attacks. By providing strong security measures, this approach enhances user trust in the data transmission process. While ensuring data security, it also improves user experience and satisfaction.

[0106] In one embodiment of the present invention, the current non-adaptive anti-leakage After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted, including:

[0107] S2031, the sender Sen randomly selects W1, ···, W n ← R G T ;

[0108] S2032, using non-adaptive anti-leakage The encryption algorithm in the protocol is W1,···,W n To encrypt, for any i∈[1,n], let A i W i The corresponding ciphertext;

[0109] S2033. For all 1≤i≤n, the sender Sen calculates B i =H(W i )⊕M i , and set C i =(A i ,B i ), where H:G T →G T It is a collision-resistant Hash function;

[0110] S2034, the sender Sen sends the public key mpk and the ciphertext C1, ..., C n Send to the receiver Rec;

[0111] S2035. The receiver Rec receives the public key mpk and the ciphertext C1, ..., C n After that, the receiver Rec uses its index value σ i Execute the anti-leakage blind extraction protocol with the sender Sen to obtain the corresponding decryption key sk σi ;

[0112] S2036. After obtaining the decryption key, the receiver Rec obtains the message value corresponding to the transmission message by decryption, wherein the message value is

[0113] The working principle of the above technical solution is as follows: the sender Sen first randomly selects W1,...,Wn (where Wi belongs to the group GT) as the intermediate value. W1,...,Wn is encrypted using the encryption algorithm in the non-adaptive anti-leakage protocol to obtain the corresponding ciphertext A1,...,An. For each Wi, the sender Sen calculates Bi=H(Wi)⊕Mi, where H is a collision-resistant Hash function and Mi is the message value to be transmitted. In this way, Bi not only contains the Hash value information of Wi, but also hides Mi through the XOR operation. The sender Sen combines the ciphertext Ai and Bi of each Wi into Ci=(Ai,Bi) and prepares to send it. The sender Sen sends the public key mpk (which may be used in the subsequent decryption process) and all ciphertexts C1,...,Cn to the receiver Rec. After the receiver Rec receives the public key mpk and the ciphertext set, it uses its index value σi (obtained through the anti-leakage blind extraction protocol previously performed with the sender Sen) to perform the anti-leakage blind extraction protocol with the sender Sen again to obtain the corresponding decryption key skσi. The receiver Rec uses the decryption key skσi to decrypt the ciphertext Ai and obtain Wi. Then, the receiver Rec calculates H(Wi) and restores the original message value Mi through the XOR operation H(Wi)⊕Bi. Since the value range of σi is 1≤i≤k, the receiver Rec can finally obtain the message set {Mσ1,...,Mσk}.

[0114] The effect of the above technical solution is: by combining non-adaptive and adaptive anti-leakage protocol elements, the method provides multi-level security protection during data transmission. Even if the channel is attacked, it is difficult for the attacker to directly recover the original message from the ciphertext. The use of the collision-resistant Hash function H increases the difficulty of cracking, making it difficult for the attacker to deceive the system by forging the Hash value. The blind extraction protocol allows the receiver Rec to obtain the decryption key without exposing its identity information, which enhances the flexibility and security of key management. The receiver Rec can process the decryption operations of multiple ciphertexts in parallel, improving the overall processing efficiency. By hiding the message value by performing an XOR operation on the Hash value and the message value, the receiver Rec can quickly recover the original message after decryption without a complex calculation process. The method enhances the user's trust in the data transmission process by providing strong security measures. The efficient encryption and decryption process also improves the user experience and satisfaction.

[0115] The embodiment of the present invention proposes a leakage-resistant oblivious transmission method based on a large language model, and the leakage-resistant oblivious transmission method further includes:

[0116] Continuously monitoring the transmission channel and collecting channel status information, wherein the channel status information includes signal strength, error rate, and packet loss rate;

[0117] Inputting the channel state information into a large language model, and analyzing the collected channel data through the natural language processing and analysis functions of the large language model LLM to obtain the difference value between the historical data and the current channel state information;

[0118] The channel security evaluation parameter is obtained by the difference between the historical data and the current channel state information, wherein the channel security evaluation parameter is obtained by the following formula:

[0119]

[0120] Where SCP represents the channel security evaluation parameter; P e , L e and S e Respectively represent the difference between the current error rate, packet loss rate and signal strength and the maximum error rate, maximum packet loss rate and minimum signal strength in the historical data; P m , L m and S y Respectively represent the maximum error rate, the maximum packet loss rate and the minimum signal strength in the historical data; x, y and z represent the first weight value, the second weight value and the third weight value respectively; and the first weight value, the second weight value and the third weight value are obtained by the following formula:

[0121]

[0122] Wherein, x, y and z represent the first weight value, the second weight value and the third weight value respectively; ΔP e Indicates the floating change rate of the current error rate compared to the median error rate in historical data; ΔL e Indicates the floating change rate of the current packet loss rate compared to the median packet loss rate in historical data; ΔS e Indicates the floating change rate of the current signal strength compared to the median signal strength in historical data;

[0123] When the security evaluation parameter is lower than a preset security parameter threshold, it is determined that the channel security quality is poor, and a channel security warning is issued and the encrypted transmission channel is replaced.

[0124] The working principle of the above technical solution is as follows: the system continuously monitors the transmission channel and collects channel status information, including signal strength, error rate and packet loss rate. The collected channel status information is input into the large language model (LLM). The natural language processing and analysis functions of the LLM are used to analyze the difference between the historical data and the current channel status information. The channel security evaluation parameter (SCP) is obtained by calculating the difference between the current channel status information and the historical data. The calculation of SCP takes into account the difference in error rate, packet loss rate and signal strength, and is adjusted by a specific weight formula. The weight values ​​(x, y, z) are calculated based on the floating change rate (ΔPe, ΔLe, ΔSe) of the current channel status compared to the median value of the historical data. When the SCP is lower than the preset security parameter threshold, the channel security quality is judged to be poor. The system issues a channel security warning and switches to an encrypted transmission channel.

[0125] The technical effects of the above technical solution are: through continuous monitoring and data analysis, changes in channel status can be discovered in a timely manner, effectively preventing potential security threats. The transmission strategy is dynamically adjusted according to the channel security evaluation parameters, and when a decline in security quality is detected, a more secure encrypted transmission channel is promptly replaced. The natural language processing and analysis capabilities of large language models are used to realize intelligent analysis and processing of channel status data. Through timely early warning and channel replacement, transmission errors and data loss caused by channel problems are reduced, and the user experience is improved. In summary, this technical solution effectively improves the security and stability of the transmission channel through continuous monitoring, data analysis, dynamic adjustment and intelligent management, and provides users with more reliable and efficient transmission services.

[0126] The embodiment of the present invention proposes a leakage-resistant oblivious transmission system based on a large language model, such as Figure 2 , Figure 3 and Figure 4 As shown, the large language model-based anti-leakage oblivious transmission system includes:

[0127] The first anti-leakage processing module is used by the sender Sen to use non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form;

[0128] The second anti-leakage processing module is used to start adaptive anti-leakage when it is detected that the channel used to transmit the message to be transmitted is attacked during the transmission of the processed message to be transmitted. The protocol performs data transmission on the processed message to be transmitted.

[0129] The working principle of the above technical solution is as follows: the sender Sen first uses a non-adaptive anti-leakage protocol to process the message to be transmitted, and protects the message through encryption or other security mechanisms (such as obfuscation, hiding, etc.) to ensure that the message content will not be easily leaked before transmission. The processed message is sent to the receiver in encrypted form. The encryption process uses a high-strength encryption algorithm to ensure that even if it is intercepted during transmission, the attacker cannot easily decrypt and obtain the original information.

[0130] During the message transmission process, the system monitors in real time whether the channel used to transmit the message is attacked. This is usually achieved through security monitoring equipment or software, which can detect and analyze abnormal traffic in the channel, data packet tampering and other signs. Once the channel is detected to be attacked, the system immediately starts the adaptive anti-leakage protocol. Unlike non-adaptive protocols, adaptive protocols can dynamically adjust the protection strategy according to the type and intensity of the attack to provide more effective security protection. Under the protection of the adaptive anti-leakage protocol, the processed message to be transmitted can continue to be safely transmitted to the recipient, and the confidentiality and integrity of the data can be maintained even in the face of attacks.

[0131] The effect of the above technical solution is: by combining non-adaptive and adaptive anti-leakage protocols, this method can fully protect the data security during transmission and effectively resist various network attacks. The introduction of adaptive anti-leakage protocols enables the system to dynamically adjust the protection strategy according to the actual attack situation, thereby improving the flexibility and pertinence of protection. Through the comprehensive use of multiple security mechanisms such as encryption and obfuscation, the risk of data leakage during transmission is greatly reduced. Since this method can effectively ensure the security of data transmission, users can use applications and services based on large language models with greater confidence, thereby improving user experience and trust.

[0132] In summary, this technical solution provides a more comprehensive, flexible and effective security protection measure for data transmission based on large language models by innovatively combining non-adaptive and adaptive anti-leakage protocols.

[0133] In one embodiment of the present invention, the first anti-leakage processing module comprises:

[0134] The master key pair generation module is used by the sender Sen to call the Setup algorithm to generate the public key and master key pair (mpk, msk), and generate the system parameter G (1 λ ); wherein the structures of the public key mpk and the master key msk are as follows:

[0135]

[0136] Where N = p1p2p3, and are all different λ-bit prime numbers;

[0137] A first data sending module, used for the sender Sen to send the public key mpk to the receiver Rec;

[0138] The first blind extraction protocol module is used for the receiver Rec to obtain k identities σ by executing a leakage-resistant blind extraction protocol with the sender Sen after receiving the public key mpk sent by the sender Sen. i ; Wherein, the value of i is: 1≤i≤k, which is the corresponding identity private key;

[0139] The first encryption processing module is used by the sender Sen to encrypt all messages to be transmitted, generate encrypted ciphertexts, and send all ciphertexts to the receiver Rec;

[0140] The first decryption module is used by the receiver Rec to use the identity σ i , and the identity private key corresponding to 1≤i≤k decrypts the ciphertext to obtain the message value corresponding to the message to be transmitted, where the message value is

[0141] The working principle of the above technical solution is as follows: the sender Sen first executes the Setup algorithm to generate a pair of public key mpk and master key msk, as well as system parameters G(1 λ ). Here, the structure of the public key and the master key is designed with security in mind, especially N is obtained by multiplying three different λ-bit prime numbers, which increases the difficulty of cracking. The public key mpk is sent to the receiver Rec for subsequent encryption and decryption operations.

[0142] After receiving the public key mpk, the receiver Rec performs a leakage-resistant blind extraction protocol with the sender Sen. This protocol ensures that the receiver can obtain k identity private keys σi (where 1≤i≤k) from the sender without revealing its identity information. These identity private keys will be used for subsequent decryption operations.

[0143] The sender Sen encrypts all messages to be transmitted and generates encrypted ciphertext. The encryption process here relies on the public key mpk, ensuring that only the receiver holding the corresponding private key can decrypt it. The encrypted ciphertext is sent to the receiver Rec.

[0144] The receiver Rec uses the identity private key σi (where 1≤i≤k) obtained previously through the blind extraction protocol to decrypt the received ciphertext. Since each identity private key corresponds to a possible message value, the receiver can decrypt the message set {Mσ1,...,Mσk}.

[0145] The effect of the above technical solution is: by using a non-adaptive anti-leakage protocol, the method takes security measures in key generation, distribution, message encryption and decryption, etc., effectively preventing data leakage during transmission. The generation of public and private keys is based on complex mathematical structures (such as N multiplied by three different prime numbers), which increases the difficulty of cracking. The blind extraction protocol ensures that the recipient's identity information will not be leaked in the process of obtaining the private key, further enhancing security.

[0146] The receiver can decrypt the corresponding message value according to the different identity private keys σi they hold. This design provides a flexible access control mechanism. The sender can use different encryption strategies and identity private keys for different receivers or different message contents to achieve refined data management.

[0147] Although the encryption and decryption process may involve complex mathematical operations, this method can ensure the efficiency of data transmission and decryption by optimizing the algorithm design and implementation. The receiver can process the decryption operations of multiple ciphertexts in parallel, improving the overall processing efficiency.

[0148] This method enhances users' trust in the data transmission process by providing strong security protection. The efficient transmission and decryption process also improves the user experience, allowing users to use services and applications based on this technology more smoothly.

[0149] In one embodiment of the present invention, the second anti-leakage processing module comprises:

[0150] A channel attack monitoring and judging module, used to monitor the channel used to transmit the processed message to be transmitted in real time during the transmission process, and judge whether the channel is attacked;

[0151] The data transmission process stopping module is used to stop the current non-adaptive anti-leakage when the channel used to transmit the message to be transmitted is attacked. The data transmission process of the protocol for the processed message to be transmitted;

[0152] The adaptive anti-leakage protocol startup module is used to After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted.

[0153] The working principle of the above technical solution is: during the process of transmitting the processed message to be transmitted through the channel, the system will monitor the channel in real time. This usually involves analyzing multiple indicators such as the traffic flow, data packet content, abnormal behavior, etc. in the channel to determine whether the channel is under attack. Once an anomaly is detected, the system will further analyze whether these anomalies constitute a threat to the security of data transmission, that is, determine whether the channel is indeed under attack.

[0154] If the judgment result is that the channel is indeed attacked, the system will immediately stop the ongoing data transmission process based on the non-adaptive anti-leakage protocol. In order to prevent attackers from further stealing or tampering with data by exploiting the loopholes of the current protocol. After stopping the data transmission of the non-adaptive protocol, the system will quickly switch to the adaptive anti-leakage protocol. The adaptive protocol can dynamically adjust its protection strategy according to the real-time situation such as the type and intensity of the attack to more effectively resist the attack. Under the protection of the adaptive protocol, the processed message to be transmitted will restart the data transmission process until it is safely delivered to the recipient.

[0155] The effect of the above technical solution is: through real-time monitoring and dynamic switching of protocols, the method can quickly respond to changes in channel security status during data transmission and effectively resist sudden attacks. The introduction of the adaptive anti-leakage protocol enables the system to flexibly adjust the protection strategy according to actual conditions, improving the pertinence and effectiveness of defense.

[0156] When a channel is attacked, the non-adaptive protocol data transmission is stopped in time to prevent attackers from stealing or tampering with data by exploiting protocol vulnerabilities. After switching to the adaptive protocol, the system can continue to transmit data with higher security, thereby reducing the risk of data leakage.

[0157] Real-time monitoring and protocol switching mechanisms ensure the continuity and stability of the data transmission process, and can quickly restore transmission even in the event of an attack. This approach improves the overall reliability and availability of the data transmission system, reducing service interruptions and data loss caused by attacks. By providing strong security measures, this approach enhances user trust in the data transmission process. While ensuring data security, it also improves user experience and satisfaction.

[0158] In one embodiment of the present invention, the adaptive anti-leakage protocol starting module includes:

[0159] Random selection module, used by the sender Sen to randomly select W1,···,W n ← R G T ;

[0160] The second encryption processing module is used to utilize non-adaptive anti-leakage The encryption algorithm in the protocol is W1,···,W n To encrypt, for any i∈[1,n], let A i W i The corresponding ciphertext;

[0161] The data processing module is used to calculate B for all 1≤i≤n, the sender Sen i =H(W i )⊕M i , and set C i =(A i ,B i ), where H:G T →G T It is a collision-resistant Hash function;

[0162] The second data sending module is used for the sender Sen to send the public key mpk and the ciphertext C1, ..., C n Send to the receiver Rec;

[0163] The second blind extraction protocol module is used for the receiver Rec to receive the public key mpk and the ciphertext C1, ..., C n After that, the receiver Rec uses its index value σ i Execute the anti-leakage blind extraction protocol with the sender Sen to obtain the corresponding decryption key sk σi ;

[0164] The second decryption module is used for the receiver Rec to obtain the message value corresponding to the transmission message by decryption after obtaining the decryption key, wherein the message value is

[0165] The working principle of the above technical solution is as follows: the sender Sen first randomly selects W1,...,Wn (where Wi belongs to the group GT) as the intermediate value. W1,...,Wn is encrypted using the encryption algorithm in the non-adaptive anti-leakage protocol to obtain the corresponding ciphertext A1,...,An. For each Wi, the sender Sen calculates Bi=H(Wi)⊕Mi, where H is a collision-resistant Hash function and Mi is the message value to be transmitted. In this way, Bi not only contains the Hash value information of Wi, but also hides Mi through the XOR operation. The sender Sen combines the ciphertext Ai and Bi of each Wi into Ci=(Ai,Bi) and prepares to send it. The sender Sen sends the public key mpk (which may be used in the subsequent decryption process) and all ciphertexts C1,...,Cn to the receiver Rec. After the receiver Rec receives the public key mpk and the ciphertext set, it uses its index value σi (obtained through the anti-leakage blind extraction protocol previously performed with the sender Sen) to perform the anti-leakage blind extraction protocol with the sender Sen again to obtain the corresponding decryption key skσi. The receiver Rec uses the decryption key skσi to decrypt the ciphertext Ai and obtain Wi. Then, the receiver Rec calculates H(Wi) and restores the original message value Mi through the XOR operation H(Wi)⊕Bi. Since the value range of σi is 1≤i≤k, the receiver Rec can finally obtain the message set {Mσ1,...,Mσk}.

[0166] The effect of the above technical solution is: by combining non-adaptive and adaptive anti-leakage protocol elements, the method provides multi-level security protection during data transmission. Even if the channel is attacked, it is difficult for the attacker to directly recover the original message from the ciphertext. The use of the collision-resistant Hash function H increases the difficulty of cracking, making it difficult for the attacker to deceive the system by forging the Hash value. The blind extraction protocol allows the receiver Rec to obtain the decryption key without exposing its identity information, which enhances the flexibility and security of key management. The receiver Rec can process the decryption operations of multiple ciphertexts in parallel, improving the overall processing efficiency. By hiding the message value by performing an XOR operation on the Hash value and the message value, the receiver Rec can quickly recover the original message after decryption without a complex calculation process. The method enhances the user's trust in the data transmission process by providing strong security measures. The efficient encryption and decryption process also improves the user experience and satisfaction.

[0167] The embodiment of the present invention proposes a leakage-resistant oblivious transmission system based on a large language model, wherein the leakage-resistant oblivious transmission system further comprises:

[0168] A status information monitoring module, used to continuously monitor the transmission channel and collect channel status information, wherein the channel status information includes signal strength, error rate and packet loss rate;

[0169] A difference value acquisition module, used to input the channel state information into a large language model, and analyze the collected channel data through the natural language processing and analysis functions of the large language model LLM to obtain the difference value between the historical data and the current channel state information;

[0170] The channel security evaluation parameter acquisition module is used to obtain the channel security evaluation parameter through the difference between the historical data and the current channel state information, wherein the channel security evaluation parameter is obtained through the following formula:

[0171]

[0172] Where SCP represents the channel security evaluation parameter; P e , L e and S e Respectively represent the difference between the current error rate, packet loss rate and signal strength and the maximum error rate, maximum packet loss rate and minimum signal strength in the historical data; P m , L m and S y Respectively represent the maximum error rate, the maximum packet loss rate and the minimum signal strength in the historical data; x, y and z represent the first weight value, the second weight value and the third weight value respectively; and the first weight value, the second weight value and the third weight value are obtained by the following formula:

[0173]

[0174] Wherein, x, y and z represent the first weight value, the second weight value and the third weight value respectively; ΔP e Indicates the floating change rate of the current error rate compared to the median error rate in historical data; ΔL e Indicates the floating change rate of the current packet loss rate compared to the median packet loss rate in historical data; ΔS e Indicates the floating change rate of the current signal strength compared to the median signal strength in historical data;

[0175] The channel security determination module is used to determine that the channel security quality is poor when the security evaluation parameter is lower than a preset security parameter threshold, and to perform a channel security warning and change the encrypted transmission channel.

[0176] The working principle of the above technical solution is as follows: the system continuously monitors the transmission channel and collects channel status information, including signal strength, error rate and packet loss rate. The collected channel status information is input into the large language model (LLM). The natural language processing and analysis functions of the LLM are used to analyze the difference between the historical data and the current channel status information. The channel security evaluation parameter (SCP) is obtained by calculating the difference between the current channel status information and the historical data. The calculation of SCP takes into account the difference in error rate, packet loss rate and signal strength, and is adjusted by a specific weight formula. The weight values ​​(x, y, z) are calculated based on the floating change rate (ΔPe, ΔLe, ΔSe) of the current channel status compared to the median value of the historical data. When the SCP is lower than the preset security parameter threshold, the channel security quality is judged to be poor. The system issues a channel security warning and switches to an encrypted transmission channel.

[0177] The technical effects of the above technical solution are: through continuous monitoring and data analysis, changes in channel status can be discovered in a timely manner, effectively preventing potential security threats. The transmission strategy is dynamically adjusted according to the channel security evaluation parameters, and when a decline in security quality is detected, a more secure encrypted transmission channel is promptly replaced. The natural language processing and analysis capabilities of large language models are used to realize intelligent analysis and processing of channel status data. Through timely early warning and channel replacement, transmission errors and data loss caused by channel problems are reduced, and the user experience is improved. In summary, this technical solution effectively improves the security and stability of the transmission channel through continuous monitoring, data analysis, dynamic adjustment and intelligent management, and provides users with more reliable and efficient transmission services.

[0178] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for anti-leakage oblivious transmission based on a large language model, characterized in that: The large language model-based anti-leakage oblivious transmission method includes: The sender Sen uses non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form; During the transmission of the processed message to be transmitted, if it is detected that the channel used to transmit the message to be transmitted is attacked, adaptive anti-leakage is initiated. The protocol performs data transmission on the processed message to be transmitted.

2. The method for preventing leakage of inadvertent transmission based on a large language model according to claim 1, characterized in that: The sender Sen uses non-adaptive anti-leakage The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form, including: The sender Sen calls the Setup algorithm to generate a public key and a master key pair (mpk, msk), and generates system parameters G(1 λ ); wherein the structures of the public key mpk and the master key msk are as follows: Where N = p1p2p3, and are all different λ-bit prime numbers; The sender Sen sends the public key mpk to the receiver Rec; After receiving the public key mpk sent by the sender Sen, the receiver Rec obtains k identities σ by executing a leakage-resistant blind extraction protocol with the sender Sen. i ; Wherein, the value of i is: 1≤i≤k, which is the corresponding identity private key; The sender Sen encrypts all messages to be transmitted, generates encrypted ciphertext, and sends all ciphertext to the receiver Rec; The receiver Rec uses the identity σ i , and the identity private key corresponding to 1≤i≤k decrypts the ciphertext to obtain the message value corresponding to the message to be transmitted, where the message value is 3. The method for preventing leakage of inadvertent transmission based on a large language model according to claim 1, characterized in that: During the transmission of the processed message to be transmitted, if it is detected that the channel used to transmit the message to be transmitted is attacked, adaptive anti-leakage is initiated. The protocol performs data transmission on the processed message to be transmitted, including: During the transmission of the processed message to be transmitted, real-time monitoring is performed on the channel used to transmit the message to be transmitted to determine whether the channel is attacked; When the channel used to transmit the message to be transmitted encounters an attack, the current non-adaptive anti-leakage The data transmission process of the protocol for the processed message to be transmitted; The current non-adaptive anti-leakage After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted.

4. The method for preventing leakage of inadvertent transmission based on a large language model according to claim 3, characterized in that: The current non-adaptive anti-leakage After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted, including: The sender Sen randomly selects W1,···,W n ← R G T ; Using non-adaptive anti-leakage The encryption algorithm in the protocol is W1,···,W n To encrypt, for any i∈[1,n], let A i W i The corresponding ciphertext; For all 1≤i≤n, the sender Sen calculates B i =H(W i )⊕M i , and set C i =(A i ,B i ), where H:G T →G T It is a collision-resistant Hash function; The sender Sen sends the public key mpk and the ciphertext C1, ..., C n Send to the receiver Rec; The receiver Rec receives the public key mpk and the ciphertext C1, ..., C n After that, the receiver Rec uses its index value σ i Execute the anti-leakage blind extraction protocol with the sender Sen to obtain the corresponding decryption key sk σi ; After obtaining the decryption key, the receiver Rec obtains the message value corresponding to the transmitted message through decryption, where the message value is 5. The method for preventing leakage of inadvertent transmission based on a large language model according to claim 1, characterized in that: The anti-leakage inadvertent transmission method also includes: Continuously monitoring the transmission channel and collecting channel status information, wherein the channel status information includes signal strength, error rate, and packet loss rate; Inputting the channel state information into a large language model, and analyzing the collected channel data through the natural language processing and analysis functions of the large language model LLM to obtain the difference value between the historical data and the current channel state information; The channel security evaluation parameter is obtained by the difference between the historical data and the current channel state information, wherein the channel security evaluation parameter is obtained by the following formula: Where SCP represents the channel security evaluation parameter; P e , L e and S e Respectively represent the difference between the current error rate, packet loss rate and signal strength and the maximum error rate, maximum packet loss rate and minimum signal strength in the historical data; P m , L m and S y Respectively represent the maximum error rate, the maximum packet loss rate and the minimum signal strength in the historical data; x, y and z represent the first weight value, the second weight value and the third weight value respectively; and the first weight value, the second weight value and the third weight value are obtained by the following formula: Wherein, x, y and z represent the first weight value, the second weight value and the third weight value respectively; ΔP e Indicates the floating change rate of the current error rate compared to the median error rate in historical data; ΔL e Indicates the floating change rate of the current packet loss rate compared to the median packet loss rate in historical data; ΔS e Indicates the floating change rate of the current signal strength compared to the median signal strength in historical data; When the security evaluation parameter is lower than a preset security parameter threshold, it is determined that the channel security quality is poor, and a channel security warning is issued and the encrypted transmission channel is replaced.

6. A leakage-resistant oblivious transmission system based on a large language model, characterized in that: The large language model-based anti-leakage oblivious transmission system includes: The first anti-leakage processing module is used by the sender Sen to use the non-adaptive anti-leakage LR-OT k n The protocol processes the message to be transmitted and sends the processed message to be transmitted to the receiver in encrypted form; The second anti-leakage processing module is used to start adaptive anti-leakage LR-OT when it is detected that the channel used to transmit the message to be transmitted is attacked during the transmission of the processed message to be transmitted. k n The protocol performs data transmission on the processed message to be transmitted.

7. The large language model-based anti-leakage oblivious transmission system according to claim 6, characterized in that: The first anti-leakage processing module comprises: The master key pair generation module is used by the sender Sen to call the Setup algorithm to generate the public key and master key pair (mpk, msk), and generate the system parameter G (1 λ ); wherein the structures of the public key mpk and the master key msk are as follows: Where N = p1p2p3, and are all different λ-bit prime numbers; A first data sending module, used for the sender Sen to send the public key mpk to the receiver Rec; The first blind extraction protocol module is used for the receiver Rec to obtain k identities σ by executing a leakage-resistant blind extraction protocol with the sender Sen after receiving the public key mpk sent by the sender Sen. i ; Wherein, the value of i is: 1≤i≤k, which is the corresponding identity private key; The first encryption processing module is used by the sender Sen to encrypt all messages to be transmitted, generate encrypted ciphertexts, and send all ciphertexts to the receiver Rec; The first decryption module is used by the receiver Rec to use the identity σ i , and the identity private key corresponding to 1≤i≤k decrypts the ciphertext to obtain the message value corresponding to the message to be transmitted, where the message value is 8. The large language model-based anti-leakage oblivious transmission system according to claim 6, characterized in that: The second anti-leakage processing module comprises: A channel attack monitoring and judging module, used to monitor the channel used to transmit the processed message to be transmitted in real time during the transmission process, and judge whether the channel is attacked; The data transmission process stopping module is used to stop the current non-adaptive anti-leakage when the channel used to transmit the message to be transmitted is attacked. The data transmission process of the protocol for the processed message to be transmitted; The adaptive anti-leakage protocol startup module is used to After the protocol stops the data transmission process of the processed message to be transmitted, the adaptive anti-leakage is started. The protocol performs data transmission on the processed message to be transmitted.

9. The large language model-based anti-leakage oblivious transmission system according to claim 8, characterized in that: The adaptive anti-leakage protocol startup module includes: Random selection module, used by the sender Sen to randomly select W1,···,W n ← R G T ; The second encryption processing module is used to utilize non-adaptive anti-leakage The encryption algorithm in the protocol is W1,···,W n To encrypt, for any i∈[1,n], let A i W i The corresponding ciphertext; The data processing module is used to calculate B for all 1≤i≤n, the sender Sen i =H(W i )⊕M i , and set C i =(A i ,B i ), where H:G T →G T It is a collision-resistant Hash function; The second data sending module is used for the sender Sen to send the public key mpk and the ciphertext C1, ..., C n Send to the receiver Rec; The second blind extraction protocol module is used for the receiver Rec to receive the public key mpk and the ciphertext C1, ..., C n After that, the receiver Rec uses its index value σ i Execute the anti-leakage blind extraction protocol with the sender Sen to obtain the corresponding decryption key sk σi ; The second decryption module is used for the receiver Rec to obtain the message value corresponding to the transmission message by decryption after obtaining the decryption key, wherein the message value is 10. The large language model-based anti-leakage oblivious transmission system according to claim 6, characterized in that: The inadvertent transmission system further comprises: A status information monitoring module, used to continuously monitor the transmission channel and collect channel status information, wherein the channel status information includes signal strength, error rate and packet loss rate; A difference value acquisition module, used to input the channel state information into a large language model, and analyze the collected channel data through the natural language processing and analysis functions of the large language model LLM to obtain the difference value between the historical data and the current channel state information; The channel security evaluation parameter acquisition module is used to obtain the channel security evaluation parameter through the difference between the historical data and the current channel state information, wherein the channel security evaluation parameter is obtained through the following formula: Where SCP represents the channel security evaluation parameter; P e , L e and S e Respectively represent the difference between the current error rate, packet loss rate and signal strength and the maximum error rate, maximum packet loss rate and minimum signal strength in the historical data; P m , L m and S y Respectively represent the maximum error rate, the maximum packet loss rate and the minimum signal strength in the historical data; x, y and z represent the first weight value, the second weight value and the third weight value respectively; and the first weight value, the second weight value and the third weight value are obtained by the following formula: Wherein, x, y and z represent the first weight value, the second weight value and the third weight value respectively; ΔP e Indicates the floating change rate of the current error rate compared to the median error rate in historical data; ΔL e Indicates the floating change rate of the current packet loss rate compared to the median packet loss rate in historical data; ΔS e Indicates the floating change rate of the current signal strength compared to the median signal strength in historical data; The channel security determination module is used to determine that the channel security quality is poor when the security evaluation parameter is lower than a preset security parameter threshold, and to perform a channel security warning and change the encrypted transmission channel.

Images