KR-20260063854-A - SEMANTIC COMMUNICATION SYSTEM AND METHOD THEREOF

Abstract

A semantic communication method of a semantic communication system according to an embodiment of the present disclosure comprises: performing encoder-decoder learning to estimate element data representing detailed elements of a first input data from a first input data using machine learning; generating additional information based on the learning result; distributing the additional information to a transmitter and a receiver of the semantic communication system, respectively; the transmitter receiving a second input data; the transmitter extracting an index from the second input data; the transmitter transmitting the index to the receiver; and the receiver restoring the index to the second input data based on the additional information.

Inventors

• 김근영
• 김민식
• 조원철

Assignees

• 한국전자통신연구원

Dates

Publication Date

20260507

Application Date

20241031

Claims (4)

1. In a semantic communication method of a semantic communication system, A step of performing encoder-decoder learning to estimate element data representing detailed elements of the first input data from the first input data using machine learning; A step of generating additional information based on the above learning results; A step of distributing the above additional information to the transmitter and receiver of the semantic communication system, respectively; The above transmitter receives second input data; The above transmitter extracts an index from the second input data; The step of the transmitter transmitting the index to the receiver; and The receiver comprises the step of restoring the index to the second input data based on the additional information. method.
2. In paragraph 1, The step of generating the above additional information is, A method comprising the step of generating additional information using a vector-quantized variable autoencoder, method.
3. In paragraph 2, The above additional information is characterized as being a latent vector having a finite number of matrices, method.
4. In paragraph 3, In the above encoder-decoder learning step, Characterized that the encoder refers to the transmitter and the decoder refers to the receiver, method.

Description

Semantic Communication System and Semantic Communication Method The present disclosure relates to a semantic communication method and a communication device for generating additional information and transmitting semantic information in semantic communication technology. The goal of existing communication methods is to transmit information accurately and quickly without errors, and to send image data, the entire image must be sent. In contrast, the purpose of semantic communication is to achieve the desired mission by extracting the necessary meaning and transmitting it without altering its semantics. Therefore, there is no need to transmit the original image as is; the transmitter sends only the amount of data required for mission success, and the receiver simply needs to restore it appropriately. This significantly reduces the amount of transmitted data while enabling the efficient completion of the mission. Meanwhile, in information theory, Wyner-Ziv Coding proposes a theoretical method that compresses data while allowing for errors, utilizes supplementary information available at the receiver to achieve greater compression compared to cases without such information, and restores the data within a desired error range. However, as a conceptual approach based on data properties that occur when data length becomes infinite, it does not provide methods for generating supplementary information or compressing data of finite length. Meanwhile, VQ-VAE provides a method for compressing data into a finite number of latent vectors and generating new data using these latent vectors. FIG. 1 illustrates a method of transmitting, receiving, and restoring an index regardless of the meaning of the data according to the prior art. FIG. 2 is a diagram comparing the communication technology of the prior art and the semantic communication technology of the present disclosure. FIG. 3 illustrates the process of transmitting and receiving images according to the present disclosure. FIG. 4 illustrates one of the applications of the present disclosure. FIG. 5 illustrates the process of a plurality of terminals performing a task according to the present disclosure. Figure 6 illustrates a communication model divided into three stages in semantic communication. Figure 7 illustrates the compression process utilized in rate distortion theory. Figure 8 illustrates the compression process in Wyner-Ziv coding theory. Figure 9 illustrates a vector quantization variable autoencoder. Figure 10 illustrates a modified form of an autoencoder. Figure 11 illustrates an example of using a continuous vector of an autoencoder. FIG. 12 is a flowchart illustrating a semantic communication method according to the present disclosure. FIG. 13 illustrates an encoder and a decoder according to an embodiment of the present disclosure. FIG. 14 illustrates a semantic communication system according to an embodiment of the present disclosure. [Explanation of Terms in the Present Disclosure] All embodiments described below are provided by way of example to aid in understanding the present disclosure and may be implemented in various forms with modifications different from the embodiments described herein. Furthermore, in describing the present disclosure, if it is determined that a detailed description of related known functions or known components might unnecessarily obscure the essence of the present disclosure, such detailed description will be omitted. The attached drawings are not drawn to actual scale to aid in understanding the disclosure, and the dimensions of some components may be exaggerated. When assigning reference numbers to each component, the same component is indicated by the same symbol whenever possible, even if it appears in different drawings. Additionally, terms such as first, second, A, B, (a), (b), etc., may be used to describe the components of the embodiments of the present disclosure. These terms are intended merely to distinguish the components from other components, and the nature, order, or sequence of the components is not limited by these terms. Where it is stated that a component is 'connected,' 'combined,' or 'connected' to another component, it should be understood that the component may be directly connected, combined, or connected to the other component, but that another component may also be 'connected,' 'combined,' or 'connected' between the component and the other component. Accordingly, the embodiments described in this disclosure and the configurations illustrated in the drawings are merely the most preferred embodiments of this disclosure and do not represent all of the technical ideas of this disclosure; therefore, various modified embodiments of this disclosure may exist. Furthermore, terms or words used in this disclosure and claims should not be limited to their ordinary or dictionary meanings, and should be interpreted in a meaning and concept consistent with the technical spirit of this disclosure, based on the pr