Method for generating a key for security of semantic communication and apparatus thereof

Abstract

The present invention relates to a method for generating a key to enhance the security of semantic communication, comprising: a step of acquiring at least one of channel information and time information; a step of generating a first seed value based on at least one of the acquired channel information and time information; a step of acquiring information related to an artificial intelligence (AI) model for semantic communication; a step of generating a second seed value based on the information related to the AI ​​model; and a step of generating a key based on the first seed value and the second seed value.

Description

Technology Field

[0001] The present invention relates to semantic communication technology, and more specifically, to a method and apparatus for generating a key to enhance the security of semantic communication. Background Technology

[0002] Recently, with the advancement of Artificial Intelligence (AI) technology and the expansion of high-performance devices such as laptops and smartphones, On-Device AI is being integrated into a wider range of devices, making AI usage a part of daily life for many users. Since On-Device AI enables the provision of services using AI models without communicating with external servers, it offers advantages such as privacy protection, reduced response times, and lower network costs. Devices equipped with such On-Device AI provide AI services without direct interaction with other devices; when necessary, they request information from a pre-configured database (DB), receive the response, and provide it to the user. However, as On-Device AI devices become commonplace in the near future, communication between AI devices will inevitably occur, requiring a new communication method. Semantic communication is the most frequently discussed method for this inter-AI device communication.

[0003] Semantic communication is a next-generation communication technology being mentioned as a candidate for 6G. It performs communication by focusing on the meaning of the information to be conveyed rather than the bit-level accuracy required in existing traditional communication. Accordingly, the transmitting terminal can vectorize the meaning of the content to be conveyed and transmit it, or selectively transmit only important features, and the receiving terminal can reconstruct the information based on this and provide it to the user.

[0004] To implement such semantic communication, the transmitting and receiving terminals must use the same AI model and share background knowledge. Based on the fact that ideal communication is possible only when this background knowledge exists, semantic communication can offer security advantages over traditional communication methods; however, there is a problem in that some of this background knowledge is merely common sense or can be partially estimated through backtracking. Therefore, an encryption method is required to enhance the security of semantic communication. Prior art literature

[0005] (Patent Document 0001) KR 10-2675382 B The problem to be solved

[0006] The present invention aims to solve the aforementioned problems and other problems. Another objective is to provide a method and apparatus for generating a key to enhance the security of semantic communication.

[0007] Another objective is to provide a method and apparatus for generating a first seed value based on at least one of channel information and time information, generating a second seed value based on parameter information of an AI model, and generating a key based on the first and second seed values. means of solving the problem

[0008] According to one aspect of the present invention for achieving the above or other purposes, a method for generating a key for security of semantic communication is provided, comprising: a step of acquiring at least one of channel information and time information; a step of generating a first seed value based on at least one of the acquired channel information and time information; a step of acquiring information related to an artificial intelligence (AI) model for semantic communication; a step of generating a second seed value based on the information related to the AI ​​model; and a step of generating a key based on the first seed value and the second seed value.

[0009] According to another aspect of the present invention, a key generation device is provided comprising one or more processors that perform a plurality of operations for generating a key to be used in semantic communication, and one or more memories that store a plurality of instructions for executing the plurality of operations, wherein the plurality of operations include: an operation of acquiring at least one of channel information and time information; an operation of generating a first seed value based on at least one of the acquired channel information and time information; an operation of acquiring information related to an artificial intelligence (AI) model for semantic communication; an operation of generating a second seed value based on the information related to the AI ​​model; and an operation of generating a key based on the first seed value and the second seed value.

[0010] According to another aspect of the present invention, a computer program stored on a computer-readable recording medium is provided to enable the following processes to be executed on a computer: a process of acquiring at least one of channel information and time information; a process of generating a first seed value based on at least one of the acquired channel information and time information; a process of acquiring information related to an artificial intelligence (AI) model for semantic communication; a process of generating a second seed value based on the information related to the AI ​​model; and a process of generating a key based on the first seed value and the second seed value. Effects of the invention

[0011] The effects of the key generation method and the device according to embodiments of the present invention are described as follows.

[0012] According to at least one embodiment of the present invention, by generating a key based on a first seed value obtained based on at least one of channel information and time information and a second seed value obtained based on parameter information of an AI model, there is an advantage that the security of semantic communication can be enhanced without the need to perform a separate key exchange process or a key disclosure process.

[0013] According to at least one embodiment of the present invention, by generating a key based on a first seed value obtained based on at least one of channel information and time information and a second seed value obtained based on parameter information of an AI model, there is an advantage in that it is almost impossible for a malicious user to steal all seed information, thereby strengthening the security of semantic communication.

[0014] However, the effects that can be achieved by the key generation method and the apparatus according to the embodiments of the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0015] FIG. 1 is a diagram showing the configuration of a semantic communication system according to one embodiment of the present invention; FIG. 2 is a flowchart illustrating a key generation method in a semantic communication system according to an embodiment of the present invention; FIG. 3 is a drawing referenced to explain the key generation method of FIG. 2; FIG. 4 is a flowchart illustrating a method for generating a key at a terminal according to an embodiment of the present invention; FIG. 5 is a flowchart illustrating the process of analyzing the cause of data decoding failure; FIG. 6 is a drawing referenced to explain a method for verifying information related to the generation of a second seed value; FIG. 7 is a block diagram of a computing device according to an embodiment of the present invention. Specific details for implementing the invention

[0016] Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Identical or similar components regardless of drawing symbols will be assigned the same reference number, and redundant descriptions thereof will be omitted. The suffixes "module" and "part" for components used in the following description are assigned or used interchangeably solely for the ease of drafting the specification and do not inherently possess distinct meanings or roles. That is, the term "part" used in this invention refers to a hardware component such as software, FPGA, or ASIC, and the "part" performs certain roles. However, the meaning of "part" is not limited to software or hardware. The "part" may be configured to reside in an addressable storage medium or may be configured to run one or more processors. Accordingly, as an example, a 'part' includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functionality provided within the components and 'parts' may be combined into a smaller number of components and 'parts' or further separated into additional components and 'parts'.

[0017] In addition, when describing the embodiments disclosed in this specification, if it is determined that a detailed description of related prior art may obscure the essence of the embodiments disclosed in this specification, such detailed description is omitted. Furthermore, the attached drawings are intended only to facilitate understanding of the embodiments disclosed in this specification, and the technical concept disclosed in this specification is not limited by the attached drawings; it should be understood that they include all modifications, equivalents, and substitutions that fall within the spirit and technical scope of the present invention.

[0018] The present invention proposes a method and apparatus for generating a key to enhance the security of semantic communication. Furthermore, the present invention proposes a method and apparatus for generating a first seed value based on at least one of channel information and time information, generating a second seed value based on parameter information of an AI model, and generating a key based on the first and second seed values. Semantic communication described herein is a communication technology that focuses on conveying the meaning of information using a deep learning-based AI model and prior shared background knowledge.

[0020] Hereinafter, various embodiments of the present invention will be described in detail with reference to the drawings.

[0021] FIG. 1 is a diagram showing the configuration of a semantic communication system according to one embodiment of the present invention.

[0022] Referring to FIG. 1, a semantic communication system (100) according to one embodiment of the present invention may include a plurality of terminals (100a, 100b) and a communication network (200).

[0023] Multiple terminals (100a, 100b) can be connected to each other through a communication network (200). The communication network may include wired networks and wireless networks, and specifically, may include various networks such as a Local Area Network (LAN), a Metropolitan Area Network (MAN), and a Wide Area Network (WAN). Additionally, the communication network may include the known World Wide Web (WWW). However, the communication network according to the present invention is not limited to the networks listed above and may include at least one of a known wireless data network, a known telephone network, and a known wired / wireless television network.

[0024] Multiple terminals (100a, 100b) can transmit and receive data using semantic communication. To this end, each terminal (100a, 100b) may include a semantic coding unit (110), an AI model (120), background knowledge (130), a channel coding unit (140), a key generation unit (150), and an encryption / decryption unit (160).

[0025] The semantic coding unit (110) may include a semantic encoder and a semantic decoder. Here, the semantic encoder can extract feature information of data (e.g., images, text, etc.) to be transmitted to another terminal using an AI model (120) installed in the terminal. The semantic decoder can restore data based on feature information received from another terminal using an AI model (120) installed in the terminal.

[0026] The AI ​​model (120) is an artificial intelligence model for providing AI services. The AI ​​model (120) can be generated in advance through a deep learning (DL) algorithm. The AI ​​model (120) can be stored in advance in the storage (not shown) of each terminal (100a, 100b).

[0027] The AI ​​model (120) can be used for semantic communication between multiple terminals (100a, 100b). In this case, the AI ​​model installed in the transmitting terminal (100a) and the AI ​​model installed in the receiving terminal (100b) are identical to each other. Additionally, the hyperparameters applied to the AI ​​model of the transmitting terminal (100a) and the hyperparameters applied to the AI ​​model of the receiving terminal (100b) are also identical to each other.

[0028] Background knowledge (130) can be used as training data to optimize an AI model (120) installed on multiple terminals (100a, 100b). The background knowledge (130) must be shared in advance among the multiple terminals (100a, 100b). The background knowledge (130) may include information related to a specific service domain. The background knowledge (130) may be stored in advance in the storage of each terminal (100a, 100b).

[0029] The channel coding unit (140) may include a channel encoder and a channel decoder. Here, the channel encoder can encode feature information of data to be transmitted to another terminal. The channel decoder can decode the encoded feature information received from another terminal.

[0030] The key generation unit (150) can generate a key to enhance the security of semantic communication. At this time, the key may include at least one of a symmetric key and an asymmetric key.

[0031] The key generation unit (150) generates a first seed value based on at least one of channel information and time information, generates a second seed value based on parameter information obtained through optimization of an AI model, and can generate a key based on the first and second seed values.

[0032] The encryption / decryption unit (160) can encrypt feature information output from the semantic encoder or decrypt feature information output from the semantic decoder using the generated key.

[0033] The terminals (100a, 100b) described in this specification may include, but are not limited to, mobile phones, smartphones, laptop computers, home appliances, digital broadcasting terminals, PDAs (personal digital assistants), PMPs (portable multimedia players), navigation devices, wearable devices, and Internet of Things (IoT) terminals.

[0035] FIG. 2 is a flowchart illustrating a key generation method in a semantic communication system according to an embodiment of the present invention, and FIG. 3 is a diagram referenced to explain the key generation method of FIG. 2. Although the illustrated flowchart describes the key generation method by dividing it into a plurality of steps, at least some of the steps may be performed in a different order, combined with other steps and performed together, omitted, divided into detailed steps, or performed with one or more steps not illustrated. Furthermore, in this embodiment, the first terminal (100a) is a transmitting terminal (Alice) and the second terminal (100b) is a receiving terminal (Bob) as an example.

[0036] Referring to FIGS. 2 and 3, the first terminal (100a) and the second terminal (100b) can share information for semantic communication (S201). Here, the information for semantic communication may include background knowledge, training data for optimizing an AI model, hyperparameter information set by a user, etc. The training data may include a part of the background knowledge.

[0037] The first terminal (100a) and the second terminal (100b) may share a predefined seed generation rule (S202). Here, the seed generation rule may include a first rule for generating a first seed value, a second rule for generating a second seed value, a third rule for generating a third seed value based on the first and second seed values, a fourth rule for generating a key based on the third seed value, and the like. Additionally, the seed generation rule may be shared in the form of a code-book.

[0038] The first terminal (100a) can measure the quality of the first channel between itself and the base station (250) (S203). The first terminal (100a) can store quality information for the measured first channel in storage. At this time, the quality information may include RSRP (Reference Signal Received Power), RSRQ (Reference Signal Received Quality), RSSI (Received Signal Strength Indicator), SINR (Signal to Noise Ratio), etc.

[0039] The first terminal (100a) can measure a timestamp related to the transmission of data (S204). The first terminal (100a) can store the measured time information in storage.

[0040] The first terminal (100a) can transmit first channel information and first time information to the base station (250) (S205). Here, the first channel information may include quality information for the first channel, and the first time information may include time information related to the transmission of data.

[0041] The base station (250) can transmit the first channel information and the first time information received from the first terminal (100a) to the second terminal (100b) (S206). The second terminal (100b) can store the first channel information and the first time information received from the base station (250) in storage.

[0042] The second terminal (100b) can measure the quality of the second channel between itself and the base station (250) (S207). The second terminal (100b) can store the quality information of the measured second channel in storage.

[0043] The second terminal (100b) can measure the time associated with the reception of data (S208). The second terminal (100b) can store the measured time information in storage.

[0044] The second terminal (100b) can transmit second channel information and second time information to the base station (250) (S209). Here, the second channel information may include quality information for the second channel, and the second time information may include time information related to the reception of data.

[0045] The base station (250) can transmit the second channel information and second time information received from the second terminal (100a) to the first terminal (100a) (S206). The second terminal (100b) can store the second channel information and second time information received from the base station (250) in storage.

[0046] Through the information exchange process described above, the first and second terminals (100a, 100b) may each have first channel information, second channel information, first time information, and second time information. That is, the first and second terminals (100a, 100b) may have the same channel information and time information.

[0047] The first and second terminals (100a, 100b) can generate a first seed value based on at least one of channel information and time information (S211). At this time, the first and second terminals (100a, 100b) can generate a first seed value using a first seed generation rule that is shared in advance. The first seed generation rule may include specific information used to generate the first seed value and information regarding a method of calculating the first seed value.

[0048] If a key is generated based on a first seed value, a malicious user (Mallory) can estimate the rule by monitoring the relevant information over a long period. Therefore, it is necessary to generate additional seed values ​​based on information that cannot be obtained by a malicious user through simple monitoring.

[0049] The first and second terminals (100a, 100b) can obtain parameter information by optimizing an AI model based on previously shared training data and hyperparameter information (S212). Here, the parameter information may include weights, biases, batch normalization parameters, output embedding vectors, etc.

[0050] In semantic communication, it is common practice for the transmitting and receiving terminals to use the same AI model and apply identical hyperparameters to it, thereby minimizing distortion of the information intended for transmission between them. Nevertheless, if feature information is distorted due to issues occurring in the transport layer, normal restoration may be impossible, and the parameters applied to the AI ​​model may also have some different values. Therefore, the data transmitted and received between the transmitting and receiving terminals is unsuitable as input data for extracting AI model parameters. For this reason, AI model optimization is performed based on training data consisting of part of background knowledge and hyperparameters set by the user; the resulting AI model parameters generally maintain identical values. Consequently, the transmitting and receiving terminals share only information regarding the training data and hyperparameters, while each secures its own AI model parameter values. Through this, the transmitting and receiving terminals can obtain identical parameter values ​​via the AI ​​model optimization process, regardless of the data intended for transmission.

[0051] Meanwhile, although it is highly likely that identical parameter values ​​will be derived through the optimization process of the AI ​​model, different values ​​within an error range may be obtained; therefore, a predetermined buffer range may be set for the parameter values, and a representative value may be used as seed information when the value falls within that range. For example, if the parameter of the AI ​​model is an integer, 1 may be used as the representative value when the parameter value is between 0 and 2, and 4 may be used as the representative value when the parameter value is between 3 and 5. Additionally, if the parameter of the AI ​​model is a real number, the value obtained by rounding the real number to a specified number of decimal places may be used as the representative value.

[0052] The first and second terminals (100a, 100b) can generate a second seed value based on parameter information generated through the optimization process of an AI model (S213). At this time, the first and second terminals (100a, 100b) can generate a second seed value using a second seed generation rule that is shared in advance. The second seed generation rule may include specific information used to generate the second seed value and information regarding the method of calculating the second seed value.

[0053] The first and second terminals (100a, 100b) can generate a third seed value based on the first and second seed values. At this time, the first and second terminals (100a, 100b) can generate a third seed value using a previously shared third seed generation rule. The third seed generation rule may include information regarding a method for calculating the third seed value.

[0054] The first and second terminals (100a, 100b) can generate a key based on the generated third seed value (S214). At this time, the first and second terminals (100a, 100b) can generate a key using a previously shared fourth seed generation rule. The fourth seed generation rule may include information regarding a method for calculating a key value based on the third seed value.

[0055] Meanwhile, in another embodiment, the first and second terminals (100a, 100b) can generate a key directly based on the first and second seed values ​​without needing to perform the step of generating a third seed value.

[0056] The first and second terminals (100a, 100b) can generate the same key from each other without needing to perform a separate key exchange process. Additionally, the first and second terminals (100a, 100b) can generate the same key from each other without needing to perform a separate key disclosure process.

[0057] Meanwhile, in this embodiment, channel quality and time are measured at the transmitting terminal (100a) and the receiving terminal (100b) and transmitted to the base station (250), but this is not necessarily limited thereto. Accordingly, it will be obvious to those skilled in the art that the base station (250) can directly measure channel quality and time and provide them to the terminals in response to requests from the transmitting terminal (100a) and the receiving terminal (100b).

[0058] As described above, a key generation method according to one embodiment of the present invention can enhance the security of semantic communication without the need to perform a separate key exchange process or a key disclosure process by generating a key based on a first seed value obtained based on at least one of channel information and time information and a second seed value obtained based on parameter information of an AI model. In addition, the key generation method can enhance the security of semantic communication by making it nearly impossible for a malicious user to steal all seed information by generating a key based on a first seed value obtained based on at least one of channel information and time information and a second seed value obtained based on parameter information of an AI model.

[0059] The key generation method according to the present invention can be applied to both symmetric key encryption technology and asymmetric key encryption technology depending on the usage environment.

[0060] In the case of conventional symmetric-key methods, the sender and receiver must use the same key to perform encryption and decryption, thus requiring a procedure for key transmission. Consequently, keys must be transmitted either online or offline, and there is a risk of key theft during this process. Asymmetric-key methods, which emerged to address this, also operate on the principle of disclosing the public key externally; however, due to advancements in quantum computing technology, the likelihood of estimating the private key based on the public key has increased, necessitating countermeasures. In the case of this invention, since the sender and receiver generate symmetric or asymmetric keys based on the same seed value, communication can be enabled without publicly disclosing this information. Furthermore, from Mallory's perspective, obtaining key information would require stealing channel information, time information, the AI ​​model, background knowledge-based training data, and hyperparameter information, as well as being able to estimate the combination rules for these elements. In particular, while other data may be relatively small in size, training data is significantly larger in volume compared to other information, making it virtually impossible to steal all of it.

[0062] FIG. 4 is a flowchart illustrating a method for generating a key in a terminal according to an embodiment of the present invention. The method for generating a key may be performed by a key generation unit (or key generation device, 150) of the terminal (100). Although the method for generating a key is described in the illustrated flowchart by dividing it into a plurality of steps, at least some of the steps may be performed in a different order, combined with other steps and performed together, omitted, divided into detailed steps, or performed with one or more steps not illustrated added.

[0063] Referring to FIG. 4, the key generation device (150) according to the present invention can obtain at least one of channel information and time information (S401). At this time, the channel information may include quality information for a first channel between a transmitting terminal (100a) and a base station (250) and quality information for a second channel between a receiving terminal (100b) and a base station (250). The time information may include time information related to data transmission and time information related to data reception.

[0064] The key generation device (150) can generate a first seed value based on at least one of the acquired channel information and time information (S402). At this time, the key generation device (150) can generate a first seed value using a first seed generation rule that has been shared in advance.

[0065] The key generation device (150) can obtain parameter information by optimizing an AI model based on pre-shared training data and hyperparameter information (S403). At this time, the training data may be composed of part of background knowledge. The parameter information may include at least one of weights, biases, batch normalization parameters, and output embedding vectors.

[0066] The key generation device (150) can generate a second seed value based on parameter information generated through the optimization of an AI model (S404). At this time, the key generation device (150) can generate a second seed value using a previously shared second seed generation rule.

[0067] The key generation device (150) can generate a third seed value based on the first and second seed values ​​(S405). At this time, the key generation device (150) can generate the third seed value using a previously shared third seed generation rule.

[0068] The key generation device (150) can generate a key based on a third seed value (S406). At this time, the key generation device (150) can generate a key using a previously shared fourth seed generation rule.

[0069] Meanwhile, in another embodiment, the key generation device (150) can generate a key based on the first seed value and the second seed value without needing to perform the step of generating the third seed value.

[0070] As described above, a key generation method according to one embodiment of the present invention can enhance the security of semantic communication without the need to perform a separate key exchange process or a key disclosure process by generating a key based on a first seed value obtained based on at least one of channel information and time information and a second seed value obtained based on parameter information of an AI model. In addition, the key generation method can enhance the security of semantic communication by making it nearly impossible for a malicious user to steal all seed information by generating a key based on a first seed value obtained based on at least one of channel information and time information and a second seed value obtained based on parameter information of an AI model.

[0072] FIG. 5 is a flowchart illustrating a process for analyzing the cause of failure in data decoding. The analysis process can be performed by a receiving terminal (100b). Although the analysis process is described in the illustrated flowchart as being divided into multiple steps, at least some of the steps may be performed in a different order, combined with other steps, omitted, divided into detailed steps, or performed with one or more steps not illustrated added.

[0073] Referring to FIG. 5, a receiving terminal (100b) according to the present invention can generate a key based on a first seed value obtained based on at least one of channel information and time information and a second seed value obtained based on parameter information of an AI model.

[0074] The receiving terminal (100b) can receive encrypted data from the transmitting terminal (100a). The receiving terminal (100b) can perform decryption of the encrypted data using a generated key (S501).

[0075] If decryption of the above encrypted data fails (S502), the receiving terminal (100b) can verify information related to the generation of the first seed value (S503).

[0076] The rule for generating the first seed value is set in advance and may be changed periodically to enhance security. Therefore, if the reference time of the transmitting terminal (Alice) and the receiving terminal (Bob) is different, the rule may be applied differently, and different seed values ​​may be generated. To verify this, the receiving terminal (100b) may request and receive channel information and time information set in the rule from the base station (250). The receiving terminal (100b) can verify information related to the generation of the first seed value by checking whether the currently received channel information and time information match the previously received channel information and time information.

[0077] Afterwards, the receiving terminal (100b) can verify information related to the generation of the second seed value (S504).

[0078] Similarly, the rule for generating the second seed value is pre-set and may be changed periodically to enhance security. Therefore, if the reference times of the transmitting terminal and the receiving terminal differ, the rule may be applied differently, resulting in the generation of different seed values. To verify this, as illustrated in FIG. 6, the receiving terminal (100b) can optimize an AI model based on the training data and hyperparameters set in the corresponding rule. The receiving terminal (100b) can restore the output embedding vector of the transmitting terminal (100a) using the optimized AI model. The receiving terminal (100b) can verify information related to the generation of the second seed value by checking whether the restored data and the original data of the transmitting terminal (100a) match each other.

[0079] If, as a result of the verification above, there is an error in the information related to the generation of the first and second seed values ​​(S505), the receiving terminal (100b) can regenerate the key based on the correct information from which the error has been removed (S506). The receiving terminal (100b) can decrypt the encrypted data using the regenerated key.

[0080] Meanwhile, if, as a result of the verification above, there is no error in the information related to the generation of the first and second seed values ​​(S505), the receiving terminal (100b) can detect that an attack by a malicious user has occurred (S507). The receiving terminal (100b) can take appropriate measures regarding the detected attack by a malicious user.

[0082] FIG. 7 is a block diagram of a computing device according to one embodiment of the present invention.

[0083] Referring to FIG. 7, a computing device (700) according to one embodiment of the present invention includes at least one processor (710), a computer-readable storage medium (720), and a communication bus (730). The computing device (700) may implement the terminal (100) or base station (250) described above. Additionally, the computing device (700) may implement the key generation device (150) described above.

[0084] The processor (710) may enable the computing device (700) to operate according to the exemplary embodiment described above. For example, the processor (710) may execute one or more programs (725) stored in a computer-readable storage medium (720). The one or more programs may include one or more computer-executable instructions, and the computer-executable instructions may be configured to enable the computing device (700) to perform operations according to the exemplary embodiment when executed by the processor (710).

[0085] A computer-readable storage medium (720) is configured to store computer-executable instructions or program code, program data and / or other suitable forms of information. A program (725) stored in the computer-readable storage medium (720) includes a set of instructions executable by a processor (710). In one embodiment, the computer-readable storage medium (720) may be memory (volatile memory such as random access memory, non-volatile memory, or a suitable combination thereof), one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, other forms of storage media that are accessed by a computing device (700) and capable of storing desired information, or a suitable combination thereof.

[0086] The communication bus (730) interconnects various other components of the computing device (700), including the processor (710) and the computer-readable storage medium (720).

[0087] The computing device (700) may also include one or more input / output interfaces (740) and one or more network communication interfaces (760) that provide interfaces for one or more input / output devices (750). The input / output interfaces (740) and network communication interfaces (760) are connected to a communication bus (730).

[0088] An input / output device (750) may be connected to other components of the computing device (700) through an input / output interface (740). An exemplary input / output device (750) may include input devices such as a pointing device (such as a mouse or trackpad), a keyboard, a touch input device (such as a touchpad or touchscreen), a voice or sound input device, various types of sensor devices and / or imaging devices, and / or output devices such as a display device, a printer, a speaker and / or a network card. An exemplary input / output device (750) may be included inside the computing device (700) as a component constituting the computing device (700), or it may be connected to the computing device (700) as a separate device distinct from the computing device (700).

[0089] The present invention described above can be implemented as computer-readable code on a medium on which a program is recorded. The computer-readable medium may be one that continuously stores a program executable by a computer, or temporarily stores it for execution or download. Furthermore, the medium may be various recording or storage means in the form of a single or multiple hardware components, and is not limited to a medium directly connected to a computer system but may also exist distributed over a network. Examples of media may include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and media configured to store program instructions, including ROM, RAM, and flash memory. Additionally, other examples of media may include recording or storage media managed by app stores that distribute applications or sites and servers that supply or distribute various other software. Accordingly, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the present invention shall be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the present invention are included within the scope of the present invention. Explanation of the symbols

[0090] 10: Semantic Communication System 100: Terminal 200: Communication Network 110: Semantic Coding Section 120: AI Model 130: Background Knowledge 140: Channel Coding Section 150: Key Generation Section 160: Encryption / Decryption Unit 700: Computing Device

Claims

Claim 1 A method for generating a key for the security of semantic communication, comprising: a step of acquiring at least one of channel information and time information; a step of generating a first seed value based on at least one of the acquired channel information and time information; a step of acquiring information related to an artificial intelligence (AI) model for semantic communication; a step of generating a second seed value based on the information related to the AI ​​model; and a step of generating a key based on the first seed value and the second seed value. Claim 2 A key generation method according to claim 1, wherein the channel information comprises quality information for a first channel between a transmitting terminal and a base station and quality information for a second channel between a receiving terminal and a base station. Claim 3 A key generation method according to claim 1, characterized in that the time information includes time information related to data transmission and time information related to data reception. Claim 4 A key generation method according to claim 1, wherein the channel information and time information are measured by a terminal or a base station. Claim 5 A key generation method according to claim 1, wherein the step of generating the first seed value is characterized by generating the first seed value using a first seed generation rule that is shared in advance. Claim 6 A key generation method according to claim 1, wherein the step of obtaining information related to the AI ​​model comprises: a step of optimizing the AI ​​model based on prior shared training data and hyperparameter information; and a step of obtaining parameter information generated through the optimization of the AI ​​model. Claim 7 A key generation method according to claim 6, characterized in that the learning data includes a portion of prior shared background knowledge. Claim 8 A key generation method according to claim 6, wherein the parameter information comprises at least one of a weight, a bias, a batch normalization parameter, and an output embedding vector. Claim 9 A key generation method according to claim 1, wherein the step of generating the second seed value is characterized by generating the second seed value using a previously shared second seed generation rule. Claim 10 A key generation method according to claim 1, wherein the step of generating the key comprises: a step of generating a third seed value based on the first and second seed values; and a step of generating a key based on the third seed value. Claim 11 A key generation method according to claim 10, wherein the step of generating the third seed value is characterized by generating the third seed value using a previously shared third seed generation rule. Claim 12 A key generation method according to claim 10, wherein the step of generating the key is characterized by generating the key using a previously shared fourth seed generation rule. Claim 13 A key generation method according to claim 1, characterized in that the key includes at least one of a symmetric key and an asymmetric key. Claim 14 A computer program stored on a computer-readable recording medium so that a method according to any one of claims 1 to 13 can be executed on a computer. Claim 15 A key generation device comprising one or more processors that perform a plurality of operations for generating a key to be used in semantic communication, and one or more memories that store a plurality of instructions for executing the plurality of operations, wherein the plurality of operations include: an operation of acquiring at least one of channel information and time information; an operation of generating a first seed value based on at least one of the acquired channel information and time information; an operation of acquiring information related to an artificial intelligence (AI) model for semantic communication; an operation of generating a second seed value based on the information related to the AI ​​model; and an operation of generating a key based on the first seed value and the second seed value. Claim 16 A key generation device according to claim 15, characterized in that the channel information includes quality information for a first channel between a transmitting terminal and a base station and quality information for a second channel between a receiving terminal and a base station. Claim 17 A key generation device according to claim 15, wherein the operation of acquiring information related to the AI ​​model comprises: an operation of optimizing the AI ​​model based on prior shared training data and hyperparameter information; and an operation of acquiring parameter information generated through the optimization of the AI ​​model. Claim 18 A key generation device according to claim 17, characterized in that the above-mentioned learning data includes a portion of prior shared background knowledge. Claim 19 A key generation device according to claim 17, characterized in that the parameter information comprises at least one of a weight, a bias, a batch normalization parameter, and an output embedding vector. Claim 20 A key generation device according to claim 15, wherein the operation of generating the key comprises: an operation of generating a third seed value based on the first and second seed values; and an operation of generating a key based on the third seed value.

Images