Method and device for performing semantic communication on basis of attention map in wireless communication system

Abstract

The present disclosure relates to a method for operating a first node in a wireless communication system. The method comprises the steps of: establishing a connection with a second node; receiving a first node capability information request from the second node; transmitting first node capability information to the second node on the basis of the first node capability information request; receiving semantic communication instruction information from the second node; and encoding input data by using an attention technique through semantic communication-related information in the received semantic communication instruction information and transmitting the encoded input data to the second node.

Description

Method and device for performing semantic communication based on an attention map in a wireless communication system

[0001] The following describes a method and device for performing semantic communication based on an attention map in a wireless communication system. Specifically, the present invention relates to a method and device for performing semantic communication by transmitting an attention map along with a semantic expression in a wireless communication system.

[0002] Wireless access systems are widely deployed to provide various types of communication services, such as voice and data. Typically, wireless access systems are multiple access systems that support communications with multiple users by sharing available system resources (e.g., bandwidth, transmission power). Examples of multiple access systems include code division multiple access (CDMA), frequency division multiple access (FDMA), time division multiple access (TDMA), orthogonal frequency division multiple access (OFDMA), and single-carrier frequency division multiple access (SC-FDMA).

[0003] In particular, as numerous communication devices demand greater communication capacity, enhanced mobile broadband (eMBB) communication technologies are being proposed, improving upon existing radio access technology (RAT). Furthermore, massive machine type communications (mMTC), which connects multiple devices and objects to provide diverse services anytime and anywhere, as well as communication systems that consider reliability and latency-sensitive services / user equipment (UE), are being proposed. Various technological configurations are being proposed for these solutions.

[0004] The present disclosure relates to a method and device for performing semantic communication in a wireless communication system.

[0005] The present disclosure relates to a method and device for operating a single AI (artificial intelligence) task and a multi-AI task operation based on semantic communication in a wireless communication system.

[0006] The present disclosure relates to a method and device for generating an attention map by a source through an attention technique based on semantic communication in a wireless communication system and transmitting the attention map to a destination along with a semantic representation.

[0007] The present disclosure relates to a method and apparatus for transmitting semantic representation, attention map and semantic redundancy data from a source to a destination based on semantic communication in a wireless communication system.

[0008] The present disclosure relates to a method and device for performing error detection by comparing an attention map generated at a destination and a received attention map based on semantic communication in a wireless communication system.

[0009] The present disclosure relates to a method and device for performing error detection by comparing semantic redundancy data generated at a destination and received semantic redundancy data based on semantic communication in a wireless communication system.

[0010] The technical objectives to be achieved in the present disclosure are not limited to those mentioned above, and other technical tasks not mentioned can be considered by a person having ordinary skill in the technical field to which the technical configuration of the present disclosure is applied from the embodiments of the present disclosure described below.

[0011] As an example of the present disclosure, a method for operating a first node in a wireless communication system includes the steps of establishing a connection with a second node, receiving a request for first node capability information from the second node, transmitting the first node capability information to the second node based on the request for first node capability information, receiving semantic communication instruction information from the second node; and encoding input data using an attention technique through semantic communication-related information in the received semantic communication instruction information and transmitting the encoded input data to the second node.

[0012] In addition, as an example of the present disclosure, in a wireless communication system, a first node includes a transceiver and a processor connected to the transceiver, wherein the processor controls the transceiver to establish a connection with a second node, to receive a first node capability information request from the second node, to transmit the first node capability information to the second node based on the first node capability information request, to control the transceiver to receive semantic communication instruction information from the second node, and to control input data to be encoded using an attention technique through semantic communication-related information in the received semantic communication instruction information and to transmit the encoded input data to the second node.

[0013] In addition, as an example of the present disclosure, in a method for operating a second node in a wireless communication system, the method may include the steps of establishing a connection with a first node, transmitting a first node capability information request to the first node, receiving first node capability information from the first node based on the first node capability information request, transmitting semantic communication instruction information to the first node, and receiving input data encoded based on an attention technique through semantic communication-related information in the semantic communication instruction information from the first node.

[0014] Also, as an example of the present disclosure, in a wireless communication system, a second node includes a transceiver and a processor connected to the transceiver, wherein the processor controls the transceiver to establish a connection with a first node, transmit a first node capability information request to the first node, controls the transceiver to receive first node capability information from the first node based on the first node capability information request, controls the transceiver to transmit semantic communication instruction information to the first node, and controls the transceiver to receive input data encoded based on an attention technique through semantic communication-related information in the received semantic communication instruction information from the first node.

[0015] Also, as an example of the present disclosure, in a device including at least one memory and at least one processor functionally connected to the at least one memory, the at least one processor may control the device to establish a connection with another device, receive a device capability information request from the other device, control the device to transmit the device capability information to the other device based on the device capability information request, control the device to receive semantic communication instruction information from the other device, and control the device to encode input data using an attention technique through semantic communication-related information in the received semantic communication instruction information and transmit the encoded input data to the other device.

[0016] Also, as an example of the present disclosure, a non-transitory computer-readable medium storing at least one instruction includes at least one instruction executable by a processor, wherein the at least one instruction controls the device to establish a connection with another device, to receive a device capability information request from the other device, to transmit the device capability information to the other device based on the device capability information request, to control the device to receive semantic communication instruction information from the other device, and to control the device to encode input data using an attention technique through semantic communication-related information in the received semantic communication instruction information and to transmit the encoded input data to the other device.

[0017] The following effects may be achieved by embodiments based on the present disclosure.

[0018] According to the present disclosure, a method for performing semantic communication in a wireless communication system can be provided.

[0019] According to the present disclosure, a multi-AI task operation-based method based on semantic communication can be provided in a wireless communication system.

[0020] According to the present disclosure, a method and device for generating an attention map by a source through an attention technique based on semantic communication in a wireless communication system and transmitting the attention map to a destination together with a semantic representation are provided.

[0021] According to the present disclosure, a method can be provided in a wireless communication system in which a source transmits semantic representation, attention map, and semantic redundancy data to a destination based on semantic communication.

[0022] According to the present disclosure, a method for performing error detection by comparing an attention map generated at a destination and a received attention map based on semantic communication in a wireless communication system can be provided.

[0023] According to the present disclosure, a method for performing error detection by comparing semantic redundancy data generated at a destination and received semantic redundancy data based on semantic communication in a wireless communication system can be provided.

[0024] The effects that can be obtained from the embodiments of the present disclosure are not limited to the effects mentioned above, and other effects that are not mentioned can be clearly derived and understood by those skilled in the art to which the technical configuration of the present disclosure is applied, from the description of the embodiments of the present disclosure below. In other words, unintended effects that result from implementing the configuration described in the present disclosure can also be derived by those skilled in the art from the embodiments of the present disclosure.

[0025] The accompanying drawings are intended to aid understanding of the present disclosure and, together with detailed descriptions, may provide embodiments of the present disclosure. However, the technical features of the present disclosure are not limited to specific drawings, and the features disclosed in each drawing may be combined with each other to form new embodiments. Reference numerals in each drawing may indicate structural elements.

[0026] Figure 1 illustrates an example of a communication system applicable to the present disclosure.

[0027] FIG. 2 illustrates an example of a wireless device applicable to the present disclosure.

[0028] FIG. 3 illustrates a method for processing a transmission signal applicable to the present disclosure.

[0029] Figure 4 illustrates a communication procedure between a terminal and a base station applicable to the present disclosure.

[0030] FIG. 5 illustrates an example of a communication structure that can be provided in a 6G (6th generation) system applicable to the present disclosure.

[0031] Figure 6 illustrates an electromagnetic spectrum applicable to the present disclosure.

[0032] FIG. 7 illustrates a THz wireless communication transceiver applicable to the present disclosure.

[0033] Figure 8 illustrates a THz signal generation method applicable to the present disclosure.

[0034] FIG. 9 illustrates a wireless communication transceiver applicable to the present disclosure.

[0035] Figure 10 illustrates a transmitter structure applicable to the present disclosure.

[0036] Figure 11 illustrates a modulator structure applicable to the present disclosure.

[0037] Figure 12 illustrates the structure of a perceptron included in an artificial neural network applicable to the present disclosure.

[0038] Figure 13 illustrates an artificial neural network structure applicable to the present disclosure.

[0039] Figure 14 is a diagram showing a communication model divided into three stages applicable to the present disclosure.

[0040] FIG. 15 is a diagram illustrating a method for translating an input word, English, into an output word, French, based on an attention technique applicable to the present disclosure.

[0041] FIG. 16 is a diagram illustrating a method of applying an attention technique using a transformer model when the data modality applicable to the present disclosure is an image.

[0042] FIG. 17 is a diagram illustrating a method for performing semantic communication applicable to the present disclosure.

[0043] FIG. 18 is a diagram illustrating a method for operating multiple downstream tasks located at a destination when the modality of input data to which the present disclosure is applicable is an image.

[0044] FIG. 19 is a diagram illustrating a method for performing error detection applicable to the present disclosure.

[0045] FIG. 20 may be a heat map showing a concentrated portion based on a fusion matrix and a discard ratio applicable to the present disclosure.

[0046] FIG. 21 is a diagram illustrating a threshold-based bit operation attention map applicable to the present disclosure.

[0047] Figure 22 is a diagram showing an ARC operation applicable to the present disclosure.

[0048] FIG. 23 is a diagram illustrating a method for performing error detection applicable to the present disclosure.

[0049] Figure 24 may be masked data applicable to the present disclosure, and semantic redundancy data may be obtained using this.

[0050] Figure 25 is a diagram showing an initial setting method of semantic communication applicable to the present disclosure.

[0051] Figure 26 is a diagram showing an initial setting method of semantic communication applicable to the present disclosure.

[0052] FIG. 27 is a diagram illustrating a method for performing semantic communication applicable to the present disclosure.

[0053] FIG. 28 is a diagram illustrating a method for performing semantic communication applicable to the present disclosure.

[0054] Figure 29 is a flowchart illustrating a method for performing semantic communication applicable to the present disclosure.

[0055] Figure 30 is a flowchart illustrating a method for performing semantic communication applicable to the present disclosure.

[0056] The following embodiments combine the components and features of the present disclosure in a predetermined form. Each component or feature may be considered optional unless explicitly stated otherwise. Each component or feature may be implemented without being combined with other components or features. Furthermore, some components and / or features may be combined to form embodiments of the present disclosure. The order of operations described in the embodiments of the present disclosure may be changed. Some components or features of one embodiment may be included in another embodiment or may be replaced with corresponding components or features of another embodiment.

[0057] In the description of the drawings, procedures or steps that may obscure the gist of the present disclosure are not described, and procedures or steps that can be understood by a person skilled in the art are also not described.

[0058] Throughout the specification, when a part is said to "comprising" or "including" a component, this does not mean that other components may be included, but rather that other components may be excluded, unless otherwise specifically stated. In addition, terms such as "...part," "...unit," and "module" described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware, software, or a combination of hardware and software. In addition, the words "a" or "an," "one," "the," and similar related words may be used in the context of describing the present disclosure (especially in the context of the claims below) to include both the singular and the plural, unless otherwise indicated herein or clearly contradicted by context.

[0059] Embodiments of the present disclosure described herein focus on the data transmission and reception relationship between a base station and a mobile station. Here, the base station is understood as a terminal node of a network that directly communicates with the mobile station. Certain operations described herein as being performed by the base station may, in some cases, be performed by an upper node of the base station.

[0060] That is, in a network consisting of multiple network nodes including a base station, various operations performed for communication with a mobile station may be performed by the base station or other network nodes other than the base station. In this case, the term 'base station' may be replaced by terms such as fixed station, Node B, eNB (eNode B), gNB (gNode B), ng-eNB, advanced base station (ABS), or access point.

[0061] Additionally, in the embodiments of the present disclosure, the term terminal may be replaced with terms such as user equipment (UE), mobile station (MS), subscriber station (SS), mobile subscriber station (MSS), mobile terminal, or advanced mobile station (AMS).

[0062] Additionally, a transmitter refers to a fixed and / or mobile node that provides data or voice services, and a receiver refers to a fixed and / or mobile node that receives data or voice services. Therefore, for uplink, a mobile station can be the transmitter, and a base station can be the receiver. Similarly, for downlink, a mobile station can be the receiver, and a base station can be the transmitter.

[0063] Embodiments of the present disclosure may be supported by standard documents disclosed in at least one of wireless access systems, such as IEEE 802.xx system, 3rd Generation Partnership Project (3GPP) system, 3GPP Long Term Evolution (LTE) system, 3GPP 5th generation (5G) NR (New Radio) system and 3GPP2 system, and in particular, embodiments of the present disclosure may be supported by 3GPP TS (technical specification) 38.211, 3GPP TS 38.212, 3GPP TS 38.213, 3GPP TS 38.321 and 3GPP TS 38.331 documents.

[0064] Furthermore, the embodiments of the present disclosure can be applied to other wireless access systems and are not limited to the systems described above. For example, they can be applied to systems implemented after the 3GPP 5G NR system and are not limited to a specific system.

[0065] That is, obvious steps or parts not described in the embodiments of the present disclosure can be explained by referring to the above documents. In addition, all terms disclosed in this document can be explained by the above standard documents.

[0066] Hereinafter, preferred embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. The detailed description set forth below, together with the accompanying drawings, is intended to illustrate exemplary embodiments of the present disclosure and is not intended to represent the only embodiments in which the technical configurations of the present disclosure may be implemented.

[0067] Additionally, specific terms used in the embodiments of the present disclosure are provided to aid in understanding of the present disclosure, and the use of such specific terms may be changed to other forms without departing from the technical spirit of the present disclosure.

[0068] The following technology can be applied to various wireless access systems such as CDMA (code division multiple access), FDMA (frequency division multiple access), TDMA (time division multiple access), OFDMA (orthogonal frequency division multiple access), and SC-FDMA (single carrier frequency division multiple access).

[0069] For clarity, the following description is based on 3GPP communication systems (e.g., LTE, NR, etc.), but the technical spirit of the present disclosure is not limited thereto. LTE may refer to technology after 3GPP TS 36.xxx Release 8. Specifically, LTE technology after 3GPP TS 36.xxx Release 10 may be referred to as LTE-A, and LTE technology after 3GPP TS 36.xxx Release 13 may be referred to as LTE-A pro. 3GPP NR may refer to technology after TS 38.xxx Release 15. 3GPP 6G may refer to technology after TS Release 17 and / or Release 18. "xxx" refers to a standard document detail number. LTE / NR / 6G may be collectively referred to as a 3GPP system.

[0070] For background information, terms, abbreviations, etc. used in this disclosure, reference may be made to standard documents published prior to this disclosure. For example, reference may be made to standard documents 36.xxx and 38.xxx.

[0071] Communication system applicable to the present disclosure

[0072] Although not limited thereto, the various descriptions, functions, procedures, proposals, methods and / or operational flowcharts of the present disclosure disclosed in this document may be applied to various fields requiring wireless communication / connectivity (e.g., 5G) between devices.

[0073] Hereinafter, more specific examples will be provided with reference to the drawings. In the drawings / descriptions below, the same drawing reference numerals may represent identical or corresponding hardware blocks, software blocks, or functional blocks, unless otherwise described.

[0074] Figure 1 illustrates an example of a communication system applied to the present disclosure.

[0075] Referring to FIG. 1, a communication system (100) applied to the present disclosure includes a wireless device, a base station, and a network. Here, the wireless device refers to a device that performs communication using a wireless access technology (e.g., LTE, LTE-A, LTE-A pro, NR, 5G, 5G-A, 6G) and may be referred to as a communication / wireless / 5G device. Although not limited thereto, the wireless device may include a robot (100a), a vehicle (100b-1, 100b-2), an XR (extended reality) device (100c), a hand-held device (100d), a home appliance (100e), an IoT (Internet of Things) device (100f), and an AI (artificial intelligence) device / server (100g). For example, the vehicle may include a vehicle equipped with a wireless communication function, an autonomous vehicle, a vehicle capable of performing vehicle-to-vehicle communication, etc. Here, the vehicles (100b-1, 100b-2) may include unmanned aerial vehicles (UAVs) (e.g., drones). The XR devices (100c) include augmented reality (AR) / virtual reality (VR) / mixed reality (MR) devices, and may be implemented in the form of head-mounted devices (HMDs), head-up displays (HUDs) installed in vehicles, televisions, smartphones, computers, wearable devices, home appliances, digital signage, vehicles, robots, etc. The portable devices (100d) may include smartphones, smart pads, wearable devices (e.g., smartwatches, smart glasses), computers (e.g., laptops, etc.), etc. The home appliances (100e) may include TVs, refrigerators, washing machines, etc. The IoT devices (100f) may include sensors, smart meters, etc.For example, the base station (120) and the network (130) may also be implemented as wireless devices, and a specific wireless device (120a) may act as a base station / network node to other wireless devices.

[0076] Wireless devices (100a to 100f) can be connected to a network (130) via a base station (120). AI technology can be applied to the wireless devices (100a to 100f), and the wireless devices (100a to 100f) can be connected to an AI server (100g) via a network (130). The network (130) can be configured using a 3G network, a 4G (e.g., LTE) network, a 5G (e.g., NR), or a 6G network. The wireless devices (100a to 100f) can communicate with each other via the base station (120) / network (130), but can also communicate directly (e.g., sidelink communication) without going through the base station (120) / network (130). For example, vehicles (100b-1, 100b-2) can communicate directly (e.g., V2V (vehicle to vehicle) / V2X (vehicle to everything) communication). Additionally, an IoT device (100f) (e.g., a sensor) can communicate directly with another IoT device (e.g., a sensor) or another wireless device (100a to 100f).

[0077] Wireless communication / connection (150a, 150b, 150c) can be established between wireless devices (100a to 100f) / base stations (120), and base stations (120) / base stations (120). Here, the wireless communication / connection can be established through various wireless access technologies such as uplink / downlink communication (150a), sidelink communication (150b) (or D2D communication), and base station-to-base station communication (150c) (e.g., relay, IAB (integrated access backhaul)). Through the wireless communication / connection (150a, 150b, 150c), the wireless device and base station / wireless device, and base stations and base stations can transmit / receive wireless signals to / from each other. For example, the wireless communication / connection (150a, 150b, 150c) can transmit / receive signals through various physical channels. To this end, based on various proposals of the present disclosure, at least some of various configuration information setting processes for transmitting / receiving wireless signals, various signal processing processes (e.g., channel encoding / decoding, modulation / demodulation, resource mapping / demapping, etc.), resource allocation processes, etc. may be performed.

[0078] Devices applicable to the present disclosure

[0079] FIG. 2 illustrates an example of a wireless device applicable to the present disclosure.

[0080] Referring to FIG. 2, the wireless device (200) can transmit and receive wireless signals via various wireless access technologies (e.g., LTE, LTE-A, LTE-A pro, NR, 5G, 5G-A, 6G). The wireless device (200) includes at least one processor (202) and at least one memory (204), and may additionally include at least one transceiver (206) and / or at least one antenna (208).

[0081] The processor (202) controls the memory (204) and / or the transceiver (206), and may be configured to implement the descriptions, functions, procedures, proposals, methods, and / or operational flowcharts disclosed in this document. For example, the processor (202) may process information in the memory (204) to generate first information / signal, and then transmit a wireless signal including the first information / signal via the transceiver (206). In addition, the processor (202) may receive a wireless signal including second information / signal via the transceiver (206), and then store information obtained from signal processing of the second information / signal in the memory (204). The memory (204) may be connected to the processor (202) and may store various information related to the operation of the processor (202). For example, the memory (204) may store software code including instructions for performing some or all of the processes controlled by the processor (202), or for performing the descriptions, functions, procedures, proposals, methods, and / or operational flowcharts disclosed herein. Here, the processor (202) and the memory (204) may be part of a communication modem / circuit / chip designed to implement wireless communication technology. The transceiver (206) may be connected to the processor (202) and may transmit and / or receive wireless signals via at least one antenna (208). The transceiver (206) may include a transmitter and / or a receiver. The transceiver (206) may be used interchangeably with an RF (radio frequency) unit. In the present disclosure, a wireless device may also mean a communication modem / circuit / chip.

[0082] Hereinafter, the hardware elements of the wireless device (200) will be described in more detail. Although not limited thereto, at least one protocol layer may be implemented by at least one processor (202). For example, at least one processor (202) may implement at least one layer (e.g., a functional layer such as physical (PHY), media access control (MAC), radio link control (RLC), packet data convergence protocol (PDCP), radio resource control (RRC), and service data adaptation protocol (SDAP)). At least one processor (202) may generate at least one Protocol Data Unit (PDU) and / or at least one Service Data Unit (SDU) according to the descriptions, functions, procedures, proposals, methods, and / or operation flowcharts disclosed in this document. At least one processor (202) may generate a message, control information, data, or information according to the descriptions, functions, procedures, proposals, methods, and / or operation flowcharts disclosed in this document. At least one processor (202) can generate a signal (e.g., a baseband signal) including a PDU, an SDU, a message, control information, data or information according to the functions, procedures, proposals and / or methods disclosed in this document, and provide the signal to at least one transceiver (206). At least one processor (202) can receive a signal (e.g., a baseband signal) from at least one transceiver (206) and obtain the PDU, SDU, message, control information, data or information according to the descriptions, functions, procedures, proposals, methods and / or operational flowcharts disclosed in this document.

[0083] At least one processor (202) may be referred to as a controller, a microcontroller, a microprocessor, or a microcomputer. The at least one processor (202) may be implemented by hardware, firmware, software, or a combination thereof. For example, at least one application specific integrated circuit (ASIC), at least one digital signal processor (DSP), at least one digital signal processing device (DSPD), at least one programmable logic device (PLD), or at least one field programmable gate array (FPGA) may be included in the at least one processor (202). The descriptions, functions, procedures, proposals, methods, and / or operation flowcharts disclosed in this document may be implemented using firmware or software, and the firmware or software may be implemented to include modules, procedures, functions, etc. The descriptions, functions, procedures, proposals, methods, and / or operation flowcharts disclosed in this document may be included in the at least one processor (202), or may be stored in at least one memory (204) and driven by the at least one processor (202). The descriptions, functions, procedures, suggestions, methods and / or flowcharts disclosed in this document may be implemented using firmware or software in the form of code, instructions and / or sets of instructions.

[0084] At least one memory (204) can be connected to at least one processor (202) and can store various forms of data, signals, messages, information, programs, codes, instructions and / or commands. The at least one memory (204) can be configured as a read only memory (ROM), a random access memory (RAM), an erasable programmable read only memory (EPROM), a flash memory, a hard drive, a register, a cache memory, a computer readable storage medium and / or a combination thereof. The at least one memory (204) can be located internally and / or externally to the at least one processor (202). In addition, the at least one memory (204) can be connected to the at least one processor (202) via various technologies such as a wired or wireless connection.

[0085] At least one transceiver (206) can transmit user data, control information, wireless signals / channels, etc., mentioned in the methods and / or flowcharts of this document to at least one other device. At least one transceiver (206) can receive user data, control information, wireless signals / channels, etc. mentioned in the descriptions, functions, procedures, proposals, methods and / or flowcharts disclosed in this document from at least one other device. For example, at least one transceiver (206) can be connected to at least one processor (202) and can transmit and receive wireless signals. For example, at least one processor (202) can control at least one transceiver (206) to transmit user data, control information, or wireless signals to at least one other device. Furthermore, at least one processor (202) can control at least one transceiver (206) to receive user data, control information, or wireless signals from at least one other device. In addition, at least one transceiver (206) may be connected to at least one antenna (208), and at least one transceiver (206) may be configured to transmit and receive user data, control information, wireless signals / channels, etc. mentioned in the descriptions, functions, procedures, proposals, methods and / or operation flowcharts disclosed in this document through at least one antenna (208). In this document, at least one antenna may be a plurality of physical antennas or a plurality of logical antennas (e.g., antenna ports). At least one transceiver (206) may convert the received wireless signals / channels, etc. from RF band signals to baseband signals in order to process the received user data, control information, wireless signals / channels, etc. using at least one processor (202). At least one transceiver (206) may convert the processed user data, control information, wireless signals / channels, etc. from baseband signals to RF band signals using at least one processor (202).For this purpose, at least one transceiver (206) may include an (analog) oscillator and / or filter.

[0086] The components of the wireless device described with reference to FIG. 2 may be referred to by different terms in terms of functionality. For example, the processor (202) may be referred to as a control unit, the transceiver (206) as a communication unit, and the memory (204) as a storage unit. In some cases, the communication unit may be used to mean at least a portion of the processor (202) and the transceiver (206).

[0087] The structure of the wireless device described with reference to FIG. 2 can be understood as the structure of at least a portion of various devices. For example, the structure of the wireless device illustrated in FIG. 2 can be at least a portion of various devices described with reference to FIG. 1 (e.g., a robot (100a), a vehicle (100b-1, 100b-2), an XR device (100c), a portable device (100d), a home appliance (100e), an IoT device (100f), an AI device / server (100g)). Furthermore, according to various embodiments, in addition to the components illustrated in FIG. 2, the device may further include other components.

[0088] For example, the device may be a portable device such as a smartphone, a smart pad, a wearable device (e.g., a smart watch, smart glasses), or a portable computer (e.g., a laptop, etc.). In this case, the device may further include at least one of a power supply unit that supplies power and includes a wired / wireless charging circuit, a battery, etc., an interface unit that includes at least one port for connection with another device (e.g., an audio input / output port, a video input / output port), and an input / output unit for inputting and outputting image information / signals, audio information / signals, data, and / or information input from a user.

[0089] For example, the device may be a mobile device such as a mobile robot, a vehicle, a train, an aerial vehicle (AV), a ship, etc. In this case, the device may further include at least one of a driving unit including at least one of an engine, a motor, a power train, wheels, brakes, and a steering unit of the device, a power supply unit including a wired / wireless charging circuit, a battery, etc. that supplies power, a sensor unit that senses status information, environmental information, and user information of the device or its surroundings, an autonomous driving unit that performs functions such as path maintenance, speed control, and destination setting, and a position measurement unit that obtains location information of the mobile device through a global positioning system (GPS) and various sensors.

[0090] For example, the device may be an XR device such as an HMD, a head-up display (HUD) installed in a vehicle, a television, a smartphone, a computer, a wearable device, a home appliance, a digital signage, a vehicle, a robot, etc. In this case, the device may further include at least one of a power supply unit that supplies power and includes a wired / wireless charging circuit, a battery, etc., an input / output unit that obtains control information, data, etc. from the outside and outputs the generated XR object, and a sensor unit that senses status information, environmental information, and user information of the device or the surroundings of the device.

[0091] For example, the device may be a robot that can be classified into industrial, medical, household, military, etc. types depending on the purpose or field of use. In this case, the device may further include at least one of a sensor unit that senses status information, environmental information, and user information of the device or its surroundings, and a driving unit that performs various physical actions, such as moving the robot joints.

[0092] For example, the device may be an AI device such as a TV, a projector, a smartphone, a PC, a laptop, a digital broadcasting terminal, a tablet PC, a wearable device, a set-top box (STB), a radio, a washing machine, a refrigerator, digital signage, a robot, a vehicle, etc. In this case, the device may further include at least one of an input unit that acquires various types of data from the outside, an output unit that generates output related to sight, hearing, or touch, a sensor unit that senses status information, environmental information, and user information of the device or its surroundings, and a training unit that trains a model composed of an artificial neural network using learning data.

[0093] The structure of the wireless device illustrated in FIG. 2 may be understood as a part of a RAN node (e.g., a base station, DU, RU, RRH, etc.). That is, the device illustrated in FIG. 2 may be a RAN node. In this case, the device may further include a wired transceiver for front haul and / or back haul communications. However, if the front haul and / or back haul communications are based on wireless communications, at least one transceiver (206) illustrated in FIG. 2 may be used for front haul and / or back haul communications, and a wired transceiver may not be included.

[0094] FIG. 3 illustrates a method for processing a transmission signal applicable to the present disclosure. For example, the transmission signal may be processed by a signal processing circuit. At this time, the signal processing circuit (300) may include a scrambler (310), a modulator (320), a layer mapper (330), a precoder (340), a resource mapper (350), and a signal generator (360). At this time, as an example, the operations / functions of FIG. 3 may be performed in the processor (202) and / or the transceiver (206) of FIG. 2. Furthermore, as an example, the hardware elements of FIG. 3 may be implemented in the processor (202) and / or the transceiver (206) of FIG. 2. As an example, blocks 310 to 360 may be implemented in the processor (202) of FIG. 2. Additionally, blocks 310 to 350 may be implemented in the processor (202) of FIG. 2, and block 360 may be implemented in the transceiver (206) of FIG. 2, and are not limited to the above-described embodiment.

[0095] The codeword can be converted into a wireless signal through the signal processing circuit (300) of FIG. 3. Here, the codeword is an encoded bit sequence of an information block. The information block may include a transport block (e.g., a UL-SCH transport block, a DL-SCH transport block). The wireless signal may be transmitted through various physical channels (e.g., a PUSCH, a PDSCH). Specifically, the codeword can be converted into a bit sequence scrambled by a scrambler (310). The scramble sequence used for scrambling is generated based on an initialization value, and the initialization value may include ID information of the wireless device, etc. The scrambled bit sequence can be modulated into a modulation symbol sequence by a modulator (320). The modulation scheme may include pi / 2-binary phase shift keying (pi / 2-BPSK), m-phase shift keying (m-PSK), m-quadrature amplitude modulation (m-QAM), etc.

[0096] A complex modulation symbol sequence can be mapped to at least one transmission layer by a layer mapper (330). The modulation symbols of each transmission layer can be mapped to corresponding antenna port(s) by a precoder (340). The output z of the precoder (340) can be obtained by multiplying the output y of the layer mapper (330) by a precoding matrix W of NХM, where N is the number of antenna ports and M is the number of transmission layers. Here, the precoder (340) can perform precoding after performing transform precoding (e.g., discrete Fourier transform (DFT) transform) on the complex modulation symbols. Additionally, the precoder (340) can perform precoding without performing transform precoding.

[0097] The resource mapper (350) can map modulation symbols of each antenna port to time-frequency resources. The time-frequency resources can include multiple symbols (e.g., CP-OFDMA symbols, DFT-s-OFDMA symbols) in the time domain and multiple subcarriers in the frequency domain. The signal generator (360) generates a wireless signal from the mapped modulation symbols, and the generated wireless signal can be transmitted to another device through each antenna. To this end, the signal generator (360) can include an inverse fast Fourier transform (IFFT) module, a cyclic prefix (CP) inserter, a digital-to-analog converter (DAC), a frequency uplink converter, and the like.

[0098] The signal processing process for a received signal in a wireless device may be configured in reverse order of the signal processing process (310 to 360) of FIG. 3. For example, a wireless device (e.g., 200 of FIG. 2) may receive a wireless signal from the outside through an antenna port / transceiver. The received wireless signal may be converted into a baseband signal through a signal restorer. For this purpose, the signal restorer may include a frequency downlink converter, an analog-to-digital converter (ADC), a CP remover, and a fast Fourier transform (FFT) module. Thereafter, the baseband signal may be restored to a codeword through a resource demapper process, a postcoding process, a demodulation process, and a descrambling process. The codeword may be restored to the original information block through decoding. Therefore, a signal processing circuit (not shown) for a received signal may include a signal restorer, a resource demapper, a postcoder, a demodulator, a descrambler, and a decoder.

[0099] Figure 4 illustrates a communication procedure between a terminal and a base station applicable to the present disclosure. Figure 4 illustrates operations of a terminal (410) and a base station (420) transmitting and / or receiving data and operations performed prior thereto.

[0100] Referring to FIG. 4, in step 401, the terminal (410) and the base station (420) perform synchronization. For example, the terminal (410) performs an initial cell search operation. Specifically, the terminal (410) can detect at least one synchronization signal transmitted from the base station (420) according to a predefined rule. Here, the synchronization signal can include multiple synchronization signals classified according to structure or purpose (e.g., primary synchronization signal, secondary synchronization signal). Through this, the terminal (410) can check the boundary of the frame, subframe, slot, and / or symbol of the base station (420) and obtain information about the base station (420) (e.g., cell identifier).

[0101] In step 403, the terminal (410) obtains system information transmitted from the base station (420). The system information is information related to the properties, characteristics, and / or capabilities of the base station (420) required to access the base station (420) and use the service, and may be classified according to the content (e.g., whether it is essential for access), transmission structure (e.g., channel used, whether provided on-demand), etc., and may be classified into, for example, a master information block (MIB) and a system information block (SIB). If necessary, the terminal (410) may transmit a signal requesting system information before receiving the system information. However, the request and provision of the system information may be performed after the random access procedure described below.

[0102] In step 405, the terminal (410) and the base station (420) perform a random access procedure. The terminal (410) may transmit and / or receive at least one message (e.g., a random access preamble, a random access response (RAR) message, etc.) for the random access procedure based on information related to the random access channel of the base station (420) obtained through system information (e.g., channel position, channel structure, supported preamble structure, etc.). For example, the terminal (410) may transmit a preamble (e.g., MSG1) through the random access channel, receive an RAR message (e.g., MSG2), transmit a message (e.g., MSG3) including information related to the terminal (410) (e.g., identification information) to the base station (420) using scheduling information included in the RAR message, and receive a message (e.g., MSG4) for contention resolution and / or connection establishment. As another example, MSG1 and MSG3 may be sent and received as one message, or MSG2 and MSG4 may be sent and received as one message.

[0103] In step 407, the terminal (410) and the base station (420) perform signaling of control information. Here, the control information may be defined in various layers, such as a layer that controls a connection (e.g., a radio resource control (RRC) layer), a layer that handles mapping between logical channels and transport channels (e.g., a media access control (MAC) layer), and a layer that handles physical channels (e.g., a physical (PHY) layer). For example, the terminal (410) and the base station (420) may perform at least one of signaling for establishing a connection, signaling for determining settings related to communication, and signaling for indicating allocated resources.

[0104] In step 409, the terminal (410) and the base station (420) transmit and / or receive data. In other words, the terminal (410) and the base station (420) can process, transmit, and / or receive data based on the signaling of the control information. For example, when transmitting data, the terminal (410) or the base station (420) can perform at least one of channel encoding, rate matching, scrambling, constellation mapping, layer mapping, waveform modulation, antenna mapping, and resource mapping on the information bits. Conversely, when receiving data, the terminal (410) or the base station (420) can perform at least one of signal extraction from resources, waveform demodulation for each antenna, signal arrangement considering layer mapping, constellation demapping, descrambling, and channel decoding.

[0105] 6G communication systems and core implementation technologies of 6G systems

[0106] The 5G system defines various operating bands within FR1 (frequency range 1), which covers 410 MHz to 7125 MHz, and FR2 (frequency range 2), which covers 24,250 MHz to 71,000 MHz. Various frequencies are being discussed as operating bands for the subsequent 6G system, and the use of higher frequencies than 5G systems is also being considered for wider bandwidth and higher transmission speeds. One such band is the THz (terahertz) frequency band, which covers approximately 100 GHz to 10 THz. The THz frequency band is a band that has both the transparency of radio waves and the straightness of light waves, and communications using the THz frequency band are expected to play a transitional role from existing radio-centered communications to lightwave-based communications.

[0107] 6G systems utilizing the THz frequency band are aimed at i) very high data rates per device, ii) a very large number of connected devices, iii) global connectivity, iv) very low latency, v) reducing energy consumption of battery-free IoT devices, vi) ultra-reliable connectivity, and vii) connected intelligence with machine learning capabilities. The vision of the 6G system can be divided into four aspects: “intelligent connectivity,” “deep connectivity,” “holographic connectivity,” and “ubiquitous connectivity,” and the 6G system can be designed to satisfy the requirements as shown in [Table 1] below.

[0108] [Table 1]

[0109]

[0110] At this time, the 6G system may have key factors such as enhanced mobile broadband (eMBB), ultra-reliable low latency communications (URLLC), massive machine type communications (mMTC), AI integrated communication, tactile internet, high throughput, high network capacity, high energy efficiency, low backhaul and access network congestion, and enhanced data security.

[0111] Figure 5 illustrates an example of a communication structure that can be provided in a 6G system applicable to the present disclosure. Referring to Figure 5, a 6G system is expected to have 50 times higher simultaneous wireless communication connectivity than a 5G wireless communication system. URLLC, a key feature of 5G, is expected to become an even more important technology in 6G communications by providing end-to-end latency of less than 1 ms. Furthermore, 6G systems will have significantly superior volumetric spectral efficiency, unlike the frequently used area spectral efficiency. 6G systems can provide extremely long battery life and advanced battery technologies for energy harvesting, so mobile devices in 6G systems may not need to be separately charged.

[0112] As core implementation technologies of the 6G system, technologies such as artificial intelligence (AI), THz (terahertz) communication, optical wireless technology, FSO backhaul network, massive MIMO technology, blockchain, 3D networking, quantum communication, unmanned aerial vehicles, cell-free communication, wireless information and energy transfer (WIET), integration of sensing and communication, integration of access backhaul networks, holographic beamforming, big data analysis, and large intelligent surface (LIS) can be adopted.

[0113] For example, THz communication is a communication that utilizes a spectrum in a frequency band between 0.1 THz and 10 THz with a corresponding wavelength in the range of 0.03 mm to 3 mm as shown in Fig. 6, and can be implemented using circuit elements having a structure as shown in Fig. 7. In addition, optical wireless technology is a technology that generates and modulates THz signals using optical elements, and can be implemented based on devices having structures as shown in Figs. 8, 9, 10, and 11.

[0114] In addition, artificial intelligence can be implemented based on various models such as neural networks and machine learning (machine models). For example, an artificial intelligence model of a neural network structure can be based on the structure of a perceptron as in Fig. 12. Referring to Fig. 12, an artificial neural network can be composed of multiple perceptrons. According to the structure of the perceptron, when an input vector x={x1, x2, …, xd} is input, each component is multiplied by a weight {W1, W2, …, Wd}, and all the results are added, and then an activation function σ(·) is applied. A large artificial neural network structure can be formed by extending the simplified perceptron structure illustrated in Fig. 12, and the input vector can be applied to perceptrons of different dimensions. When perceptrons are stacked, a neural network having an input layer, a hidden layer, and an output layer as in Fig. 13 can be configured.

[0115] Figure 14 is a diagram illustrating a three-stage communication model applicable to the present disclosure. Referring to Figure 14, the communication model can be explained by dividing it into three stages. Stage 1, from a technical perspective, may be concerned with whether symbols for communication are accurately transmitted, and Shannon's information theory can be viewed as a theory focusing on this technical aspect. On the other hand, Stage 2, from a semantic perspective, may be concerned with how accurately the transmitted symbols convey the correct meaning, and Stage 3, from an effectiveness perspective, may be concerned with how effectively the received meaning influences the correct operation.

[0116] One of the various goals of 6G communications may be to provide services that interconnect humans and machines. Semantic communication, based on the concept of "meaning transfer," can be considered as one of the next-generation wireless communication paradigms for this purpose. For example, conventional communication involves a receiver (e.g., destination) decoding an encoded signal received from a transmitter (e.g., source) into a conventional signal without error. In contrast, semantic communication focuses on the meaning intended to be conveyed through the signal, as people exchange information through the "meaning" of words when communicating.

[0117] For example, semantic communication can be a communication method based on whether the concept related to the message sent by the source is correctly interpreted by the destination. For example, the source can generate a semantic representation based on raw data and transmit it to the destination. The destination can perform communication by evaluating whether the task intended by the source is performed properly using the received semantic representation, rather than with the goal of reducing reconstruction errors. In other words, in semantic communication, communication can be performed based on whether the task is performed properly (i.e., whether the interpretation is correct) at the destination, rather than with the goal of reducing reconstruction errors.

[0118] As a specific example, a source can generate a semantic representation and transmit it to a destination. Here, the source needs to generate the semantic representation by considering the task operations performed at the destination. In other words, the source can operate based on a task-oriented semantic communication system that enables the generation of a semantic representation based on whether the task operations are performed well at the destination. Considering the above, there is a need to introduce useful invariances for the tasks performed at the destination while preserving task-relevant information. For example, to perform semantic communication considering the above, a new layer can be added as a semantic layer that governs the overall operation of semantic representations and messages. For example, the semantic layer can be located at both the source and the destination, reflecting the task-oriented semantic communication system. To perform communication between semantic radars located at each of the source and the destination, a protocol and a series of operation procedures, which are rules between the layers, need to be determined.

[0119] As an example, the following describes a method for performing attention map-based error detection in semantic communication. The attention map can be generated based on an attention technique. Attention techniques can be applied in the field of neural machine translation (NMT). Specifically, when applying the attention technique to the field of NMT, the decoder can refer back to the input sequence of the encoder at each time step to predict an output word. Here, the decoder can perform prediction by referring to an input sequence that is highly related to the predicted output word among the input sequences. In other words, the decoder can focus (attention) on the input sequence that is highly related to the predicted word rather than referring to the input sequence equally when performing output word prediction, thereby improving prediction accuracy. Considering the above, the technique can be an attention technique because it utilizes the focused (attentioned) portion, but it may not be limited to that name. However, for convenience of explanation, it is referred to as the attention technique below.

[0120] As an example, FIG. 15 is a diagram illustrating a method for translating an input word, English, into an output word, French, based on an attention technique applicable to the present disclosure. Referring to FIG. 15, an attention technique may be applied in sequence modeling between the input word, English, and the output word, French. Here, the sequence modeling between the input word, English, and the output word, French, may include weight values ​​expressed as a correlation matrix. For example, in FIG. 15, the closer the color of a pixel is to white, the higher the correlation between the words may be. Referring to FIG. 15, even if the order in which English sentences are written differs from the order in which French sentences are written, a weight may be higher for a part that is to be translated into a word that has a correlation regardless of the position, thereby increasing the prediction accuracy.

[0121] Here, as an example, the attention technique, based on the transformer model, can be applied to network architectures other than neural machine translation. Specifically, the transformer model architecture allows attention to be applied to all significant parts regardless of input length. Accordingly, attention-based processing can be performed on text, images, graphs, and other diverse data modalities, but may not be limited to a specific format.

[0122] FIG. 16 is a diagram illustrating a method of applying an attention technique using a transformer model when the data modality applicable to the present disclosure is an image. Referring to FIG. 16, a vision transformer may be applied as a transformer model for image processing. As a more specific example, the vision transformer may be a data-efficient image transformer (DeiT), but this is only one example and may not be limited thereto. Referring to FIG. 16, the vision transformer model may use a class token for matching the true label of the input image with the input value of DeiT, tokens obtained by tokenizing the input image in units of patches, and distillation tokens that play a role in learning the knowledge of the teacher model by reducing the error with the inferred value of the teacher model. Here, the classic token and the distilled token are randomly initialized before being applied to the transformer structure, and may be changed to an optimized value while passing through multiple layers.

[0123] As described above, the attention technique may be a technique for processing data by focusing on a part that is highly related to the predicted output from the input data. Here, as an example, FIG. 17 is a diagram illustrating a method for performing semantic communication applicable to the present disclosure. Referring to FIG. 17, a transmitter (transmitter, 1710) may perform semantic encoding and channel encoding on input data (sentence, s) and transmit it to a receiver (receiver, 1720). The receiver (1720) may perform channel decoding and semantic encoding on data received through the channel to reconstruct data ( ) can be obtained. Here, semantic encoding and semantic decoding can be performed through the transformer model-based encoder and transformer model-based decoder described above from the perspective of semantic communication. For example, in the case of performing semantic communication based on FIG. 17, the loss function can be as shown in Equation 1 below. Referring to Equation 1, the loss function can be composed of a part for cross entropy and a part for mutual information. Here, the cross entropy part is composed of input data (s) and reconstructed data ( ) can be the cross entropy between the input data (s), which can be as shown in Equation 2. The cross entropy part is how well the input data (s) is restored to the reconstructed data ( ) may be the part that determines whether or not it has been achieved. For example, in mathematical expression 2, is the lth word in the sentence (sentence, s) as input data. is the actual probability, is reconstructed data The lth word in It can be a predicted probability for . That is, the cross entropy part can mean the difference in the probability between the input data and the reconstructed data.

[0124] [Mathematical Formula 1]

[0125]

[0126] [Equation 2]

[0127]

[0128] Here, the mathematical expression described above can be a complete comparison between the input data and the reconstructed data. However, in semantic communication, the intent conveyed by the source can be crucial for performing downstream tasks. Therefore, it is necessary to consider the reconstructive operation for the portion of the input data that the source intends to convey, rather than the entire data. Considering the above, the following describes a method for utilizing an attention map obtained through an attention technique.

[0129] For example, when applying attention techniques from the perspective of semantic communication, the source can use attention techniques to generate and transmit a semantic representation (SR) to express the intention it wants to convey. The destination can then use attention techniques based on the received semantic representation to determine which part of the original data was more focused on, and through this, determine whether it can interpret the meaning that the source intended to convey. In other words, in semantic communication, attention techniques can perform error detection not based on the reconstruction of the entire data, but only on the part of the entire data that contains the meaning it wants to convey, and a method for this may be necessary.

[0130] Here, the existing system was configured using the existing Transformer model structure as is, and learning was performed to reconstruct the input data based on this, and the trained model was used for inference. In addition, the existing system did not detect errors in the reconstruction of the source input data at the destination by utilizing the attention map (attention matrix), which is the result of the attention technique utilized in each layer constituting the Transformer model. In other words, it was possible to perform only the detection of errors in the reconstruction of the input data, which is the result of the application of the attention technique, without focusing on the attention map, which is the result of the attention technique utilized in each layer constituting the Transformer model. However, semantic communication can be a method of communicating meaning based on the reconstructed data at the semantic level as described above. Therefore, semantic error detection, which detects errors only in the portion of the entire data that contains the meaning to be conveyed, may be necessary, and the following describes a method for performing this based on the attention technique.

[0131] As an example, the following describes a method and procedure for performing error detection using an attention map generated through an attention technique in a semantic communication-based system. The following may be a case where two-way communication is performed at a semantic level based on Level B of FIG. 14, but is not limited thereto.

[0132] Task-based operations can be performed in a semantic communication system. Specifically, a source can generate a semantic representation based on the intended message and transmit it to a destination. The destination can then reconstruct the received semantic representation at the semantic level and use the reconstructed data as input to multiple downstream tasks located at the destination to obtain the task's output. Here, semantic-level reconstruction, unlike conventional reconstruction that perfectly restores input data, can be a reconstruction that performs the operations of downstream tasks normally (or according to the intended intent) so that the destination obtains an output corresponding to the intended operation of the source.

[0133] As an example, FIG. 18 is a diagram illustrating a method for operating multiple downstream tasks located at a destination when the modality of input data applicable to the present disclosure is an image. Referring to FIG. 18, a source and a destination may have a labeled dataset necessary for learning a multi-task. For example, a source may be an entity that transmits data, such as a terminal, a base station, or other device. In addition, a destination may be an entity that receives data and performs a task appropriate for the purpose, such as a terminal, a base station, or other device. In other words, the source and the destination may be each entity that performs communication, and may not be limited to a specific type of device. For convenience of explanation, the following description will be based on the source and the destination. For example, the source and the destination may have a labeled dataset necessary for learning a task, and the destination may perform a multi-task. Here, each multi-task can operate based on a different purpose, and the semantic representation can be passed from the source to the destination, where the reconstructed data at the semantic level can be passed as input to each multi-task. For example, each multi-task can operate based on its input for a different purpose.

[0134] As a concrete example, a source can encode data given as input (image x, 1810) to generate a semantic representation (latent representation y, 1820). Here, the semantic representation (1820) can be transmitted to a destination through a channel. The destination decodes the received semantic representation (1830) to reconstruct data at the semantic level. (reconstruction data , 1840) can be obtained. After that, the destination is You can use each multi-task as input and check the output after performing an action suitable for the purpose of each task.

[0135] As an example, the following describes a method for performing inference operations based on an encoder and decoder that have completed training by applying an attention technique, and performing multi-tasks at a destination. The destination can perform error detection on reconstructed data at the semantic level. Here, the source can construct a bitwise attention map based on a set threshold and transmit the bitwise attention map to the destination. The destination can perform error detection using the bitwise attention map received from the source.

[0136] As a specific example, FIG. 19 is a diagram illustrating a method for performing error detection applicable to the present disclosure. For example, FIG. 19 describes the case where the modality of the input data is an image, but this is merely an example for convenience of explanation and may not be limited thereto. Furthermore, FIG. 19 describes one-way communication from a source (1910) to a destination (1920), but this is merely for convenience of explanation and can be equally applied to two-way communication.

[0137] Referring to FIG. 19, a source (1910) may include an encoder (1911) that has completed training based on an attention technique. When input data is provided to the encoder (1911), the encoder (1911) may generate a semantic representation and an attention matrix. Here, the semantic representation may include the meaning to be conveyed based on semantic communication. In addition, the attention matrix may be a matrix generated based on an attention technique. In addition, an attention map may be obtained based on the attention matrix. For example, when the encoder (1911) includes a single layer to which an attention technique is applied, an attention map may be obtained through the attention matrix used in the layer. On the other hand, if the encoder (1911) is composed of multiple layers to which the attention technique is applied, an attention map can be obtained by considering all attention matrices obtained from each layer to which the attention technique is applied.

[0138] As a specific example, when the attention technique is applied to multiple layers, the attention matrix of each layer can be constructed using the following mathematical formula 3 based on the transformer model. Referring to mathematical formula 3, the attention matrix for each layer constituting the transformer model can be obtained. Here, can be an attention matrix indicating how much attention flow will occur from token j of the previous layer to token i of the next layer. That is, represents how much attention the jth token of the previous layer will give to the ith token of the current layer, and multiplying the matrix between the two layers can obtain the total attention flow between the two layers. In addition, the transformer model can use residual connections to prevent positional encoding information from being lost due to backpropagation during the learning process. That is, positional information (or order information) can be provided during the learning process, and residual blocks based on residual connections can be utilized to prevent the information from being lost during the backpropagation process, thereby ensuring that learning converges well even when the number of layers increases. In the transformer model, the unit matrix of Equation 3 may be an additional configuration considering the above-mentioned points.

[0139] [Equation 3]

[0140]

[0141] In addition, as an example, if the downstream task located at the destination (1920) can transmit the target class information it wants to obtain to the source (1910), the source (1910) can first receive the target class information and then calculate an attention aware matrix for obtaining an attention map. That is, the target class information that the downstream task wants to obtain can be reflected in the process of obtaining the attention map. Here, the attention aware matrix is ​​expressed by Equation 3. The gradient value can be obtained by reflecting it as in the following mathematical expression 4. For example, in mathematical expression 4, can be the gradient value related to the target class between token j of the previous layer and token i of the current layer.

[0142] [Equation 4]

[0143]

[0144] In addition, as an example, each layer is applied with multi-head attention according to the transformer model structure. Here, the attention head may be a part to focus on, and the part to focus on may be different for each attention head. Here, the part of interest (or focus) in the data may be set differently depending on the way each attention head is combined (fused). That is, the part to focus on in the data may be set differently based on the multi-head attention structure, and a fusion metric that considers the attention head combination may be applied. As an example, the fusion metric is the above-described mathematical expression 3 (or mathematical expression 4). (or * ) can be used to compute the attention recognition matrix representing the

[0145] In addition, as an example, in the attention technique, rather than being interested in each attention value constituting the attention matrix, it is necessary to focus on the upper attention value, and a discard ratio can be set for this purpose. Specifically, only attention values ​​with upper values ​​based on the attention matrix can be utilized, and attention values ​​with lower values ​​may not be needed. Considering the above, a setting to discard attention values ​​with lower values ​​can be performed, and a discard ratio can be set for this. In addition, the set discard ratio is expressed as in Equation 3 (or Equation 4). (or * ) can be applied.

[0146] Here, the attention recognition matrix can be calculated recursively after applying the fusion matrix and discard ratio set in a specific layer L, which can be as shown in Equation 5 below. For example, the final attention recognition matrix is ​​calculated as a result of calculating the attention matrix of the entire layer through Equation 5. can be obtained.

[0147] [Equation 5]

[0148]

[0149] Based on mathematical expression 5, the final attention recognition matrix is When obtaining , each layer maintains the total attention flow as 1. Perform normalization to make rows 1 for .

[0150] However, the above-described mathematical formulas 3 to 5 are only one example for obtaining an attention matrix and may not be limited thereto.

[0151] For example, FIG. 20 may be a heat map showing a focused area based on a fusion matrix and a discard ratio applicable to the present disclosure. In FIG. 20, the DeiT model may be used with the aforementioned vision transformer series, the number of multi-heads may be 3, the number of attention layers may be 12, and an attention map may be derived based on mathematical equations 3 to 5. However, this is merely an example and is not limited thereto. Referring to FIG. 20, when all attention is acquired by taking the attention average of the heads constituting each layer, the heat map by the attention matrix may be located broadly overall because each head focuses on a different location. Therefore, in order to focus on a part of interest in the data, it is necessary to set a fusion matrix and a discard ratio, as described above. The fusion matrix and the discard ratio may be set and adjusted based on the operation of the downstream task. As a concrete example, for downstream task operations that require a lot of information from input data, the discard ratio may be adjusted lower and error detection may be performed on a larger portion, but this is only one example and may not be limited thereto.

[0152] When the attention map is obtained based on the above, a threshold-based bitwise attention map (bitwise attention map, ) can be generated. For example, the source (1910) is an attention recognition matrix obtained through an encoder (1911). For each element of It can be applied as in the following mathematical formula 6, and based on this, a threshold-based bit operation attention map can be obtained. In mathematical formula 6 It can mean the configuration of the attention recognition matrix i row j column obtained based on the encoder (1911). The threshold-based bit operation attention map finally obtained through this is This can be derived.

[0153] [Equation 6]

[0154]

[0155] Also, as an example, It may be a reference value for generating an attention map that takes into account the part that determines whether data reconstruction is normal when performed at the destination. As a specific example, FIG. 21 is a diagram illustrating a threshold-based bit operation attention map applicable to the present disclosure. For example, if a threshold-based bit operation attention map is derived based on an experimental environment such as Table 2 below in FIG. 20, the threshold-based bit operation attention map may be as shown in FIG. 21.

[0156] [Table 2]

[0157]

[0158] Based on the above, the source (1911) can finally generate a semantic representation and a threshold-based bit operation attention map and transmit them to the destination (1920). Thereafter, the destination (1920) can perform a reconstruction operation using the semantic representation. Specifically, the decoder (1921) of the destination (1920) reconstructs the semantic representation to generate reconstructed data at the semantic level. can be obtained. After that, the encoder (1922) located at the destination (1920) reconstructs the semantic level data. Attention map with input can be obtained. For example, the encoder (1922) of the destination (1920) may obtain an attention map based on the same method as the encoder (1911) of the source (1910). That is, the encoder (1922) of the destination (1920) applies the fusion matrix and the discard ratio based on the same attention matrix calculation method as the encoder (1911) of the source (1910) to obtain a threshold-based bit operation attention map based on the above-described mathematical expression 6. can be obtained. After that, it is transmitted from the source (1910). Generated through the encoder (1921) of the destination (1920) Error detection can be performed by comparing the values ​​received from the source (1910). Specifically, Generated through the encoder (1921) of the destination (1920) An attention aware redundancy check (ARC) can be performed by performing a comparison. ARC is Contrast It can be a process of measuring the number of configurations that do not match 1 as an error rate by checking how many configurations match 1. After that, error detection can be performed using. For example, This can be a criterion for determining whether the reconstructed data is of a semantic level that can convey semantic information to downstream tasks. That is, If it is not satisfied, it may not be passed as input to the downstream task.

[0159] Fig. 22 is a diagram illustrating an ARC operation applicable to the present disclosure. Referring to Fig. 22, based on the above, when the reconstruction data of the semantic level contains noise compared to the input data, it may be an attention map. Here, the case where the configuration of the attention map is 1 and the case where the configuration of the attention map is 0 can be distinguished. For example, in semantic communication, the part where the attention map is 1 may be important information that conveys meaning, so the part where the configuration is 1 in each attention map can be compared, and based on this can be compared with. After that, the measured error rate is If it is less than the value, it is judged to satisfy the condition, and the reconstructed data at the semantic level The task output can be obtained by passing it to the downstream task.

[0160] As another example, FIG. 23 is a diagram illustrating a method for performing error detection applicable to the present disclosure. Referring to FIG. 23, the source (2310) can derive additional information about the portion of the input data to which the bit operation attention map is applied while constructing a threshold-based bit operation attention map as in FIG. 19. The source (2310) can derive an attention area from the input data and generate attention-aware redundancy of data based on the attention area. The source (2310) can transmit the semantic expression, threshold-based bit operation attention map, and attention-aware redundancy of data described above in FIG. 19 to the destination (2320).

[0161] Specifically, the source (2310) is processed by the encoder (2311) into a semantic representation and a threshold-based bit operation attention map. , which may be as shown in Fig. 19. Here, the source (2310) may further extract a part related to the meaning to be conveyed from the input data. That is, the source (2310) may extract the part related to the meaning to be conveyed from the input data. By applying the method, masked data can be obtained by extracting only the areas that are judged to be important information for conveying meaning. As an example, FIG. 24 may be masked data applicable to the present disclosure, and semantic redundancy data can be obtained using the same. As an example, the semantic redundancy data can be obtained by generating CRC redundancy for the masked data based on the CRC technique, but is not limited thereto.

[0162] After that, the source (2310) can transmit the semantic representation, the threshold-based bit operation attention map, and the semantic redundancy data to the destination (2320). Here, the decoder (2321) of the destination (2320) performs reconstruction of the semantic representation in the same manner as in Fig. 19 to reconstruct the semantic level data. can be obtained. For example, reconstruction data at the semantic level The part that focuses on the meaning to be conveyed can be extracted. Specifically, the reconstructed data at the semantic level By applying the bit operation attention map received in , the area that focuses on the meaning that the source (2310) wants to convey can be extracted. For example, the reconstruction data at the semantic level The remaining areas that are not extracted using the bitwise operation attention map may not affect the meaning to be conveyed, but may not be limited thereto.

[0163] After that, a semantic redundancy check can be performed to detect errors. For example, the destination (2320) can perform a semantic redundancy check using the masked data obtained using the received bit operation attention map and the received semantic redundancy data. For example, when semantic redundancy data generated using a CRC technique is transmitted from the source (2310), the destination (2320) can perform a CRC check using the masked data and the received semantic redundancy data and measure the error rate. For example, the threshold value is as described above. It can be determined whether the reconstructed data is of a semantic level that can convey semantic information to downstream tasks based on this. Here, the measured error rate is If less than the value, the reconstruction data at the semantic level is passed as input to the downstream task so that the task can obtain the output. That is, the reconstructed data at the semantic level if the threshold is satisfied. is passed as input to the downstream task so that the task can obtain its output.

[0164] FIG. 25 is a diagram illustrating an initial setup method of semantic communication applicable to the present disclosure. Referring to FIG. 25, a first node (2510) may be connected by obtaining a synchronization signal from a second node (2520) and performing synchronization. For example, the first node (2510) may be the aforementioned source, and the second node (2520) may be the aforementioned destination. As another example, the first node (2510) may be the aforementioned destination, and the second node (2520) may be the aforementioned source. In addition, as an example, in FIG. 25, the first node (2510) may be a terminal, and the second node (2520) may be a base station. In addition, as an example, the first node (2510) may be a base station, and the second node (2520) may be a terminal, and the present invention is not limited to a specific form. Additionally, as an example, FIG. 25 can be applied to both one-way and two-way communications and is not limited to a specific form.

[0165] However, for the convenience of explanation, the description is based on the case where the first node (2510) is a terminal and the second node is a base station (2520). The second node (2520) can request capability information from the first node (2510). Here, the capability information may include various information based on the first node type. For example, the capability information may include information on whether the first node (2510) has the above-described semantic communication performance capability. That is, the capability information may include semantic communication capability information of the first node (2510). The first node (2510) can transmit the capability information to the second node (2520) based on a request from the second node (2520). Here, for example, the capability information may further include information on the types of raw data that the first node (2510) can generate, collect, or process, and the computational capabilities of the device, and may not be limited to a specific form. Thereafter, the second node (2520) can determine whether to perform semantic communication. For example, the second node (2520) can recognize that the first node (2510) has the capability to perform semantic communication based on the capability information of the first node (2510). In addition, the second node (2520) can further consider other acquired information and other parameters to determine whether to perform semantic communication, and can instruct the first node (2510) whether to perform semantic communication. That is, if the second node (2520) determines to perform communication with the first node (2510) based on semantic communication, the second node (2520) can transmit information indicating semantic communication to the first node (2510). Here, the information indicating semantic communication can include semantic communication-related information. Afterwards, the first node (2510) can store semantic communication related information transmitted together with semantic communication instruction information.For example, the semantic communication-related information may include an encoder / decoder model, an attention matrix calculation method, a fusion matrix, an attention value discard ratio, a threshold related to a bit operation attention map, and a threshold related to error detection. In addition, for example, the semantic communication-related information may further include information related to a method for generating semantic redundancy, and is not limited to the above-described information. For example, the attention matrix calculation method included in the semantic communication-related information may be information related to mathematical expressions 3 to 5, but may not be limited thereto. In addition, the threshold related to a bit operation attention map is. And the threshold for error detection is It may be, but is not limited to,

[0166] For example, semantic communication instruction information and semantic communication related information may be transmitted from the second node (2520) to the first node (2510) via at least one of a DCI (downlink control information), a MAC (medium access control) CE (control element), and an RRC (radio resource control) message.

[0167] As another example, semantic communication indication information may be indicated through upper layer signaling, and semantic communication-related information may also be transmitted through DCI. That is, the second node (2520) may obtain capability information from the first node (2510) and, if semantic communication in the above-described manner is possible, transmit semantic communication-related information to the first node (2510). The first node (2510) may store the received information and perform the above-described semantic communication. In addition, as an example, each step and operation of FIG. 25 may be omitted or merged based on settings or configurations, and may not be limited to a specific form.

[0168] FIG. 26 is a diagram illustrating a method for performing semantic communication based on an attention map applicable to the present disclosure. For example, in FIG. 26, the source (2610) may be a terminal, and the destination (2620) may be a base station. However, this is merely an example for convenience of explanation, and the same may apply even when the source (2610) is a base station and the destination (2620) is a terminal. Furthermore, FIG. 26 can be equally applied to both unidirectional and bidirectional communications, and is not limited to a specific form.

[0169] Referring to FIG. 26, the source (2610) and the destination (2620) can perform initialization for semantic communication. For example, the source (2610) and the destination (2620) can perform communication using the attention technique described above as semantic communication. More specifically, the source (2610) and the destination (2620) can perform communication by extracting the meaning to be conveyed by considering the downstream task operation performed at the destination (2620) based on semantic communication, as described above. Here, error detection can be performed using the attention technique described above, as described above.

[0170] For example, the source (2610) can obtain input data and obtain a semantic representation and bit operation attention map through an encoder operation. Here, as an example, the destination (2620) can transmit target class information that the downstream task wants to obtain to the source (2610), but is not limited thereto. Here, when the source (2610) receives target class information, it can calculate an attention recognition matrix to obtain an attention map by reflecting the target class information, as described above.

[0171] Additionally, as an example, the source (2610) can further obtain semantic redundancy together with the semantic representation and bit operation attention map. Specifically, the source (2610) can further be configured with a method for generating semantic redundancy. The source (2610) can derive an attention region from input data and generate semantic redundancy data based on the attention region. Thereafter, the source (2610) can transfer the semantic representation and the bit operation attention map to the destination (2620). Here, when semantic redundancy is generated, the semantic redundancy can be further transferred to the destination (2620) together with the semantic representation and the bit operation attention map. Thereafter, the destination (2620) can perform reconstruction at the semantic level.

[0172] FIG. 27 is a diagram illustrating a method for performing semantic communication based on an attention map applicable to the present disclosure. Referring to FIG. 27, a source (2710) and a destination (2720) can perform semantic communication based on S2601 of FIG. 26 described above, and the destination (2720) can perform semantic level reconstruction by receiving a semantic representation and a bit operation attention map. Here, the destination (2720) can generate semantic level reconstruction data from the semantic representation through a decoder. In addition, the destination (2720) can derive an attention region based on the bit operation attention map received from the semantic level reconstruction data. Thereafter, the destination (2720) can perform a semantic redundancy check using the received semantic redundancy data based on the attention region. Here, an error rate can be measured based on the semantic redundancy check. For example, the destination (2720) can perform a semantic redundancy check using the masked data obtained using the received bit operation attention map and the received semantic redundancy data. For example, when semantic redundancy data generated using a CRC technique is transmitted from the source (2710), the destination (2720) can perform a CRC check using the masked data and the received semantic redundancy data and measure the error rate. For example, the threshold value is as described above. It can be determined whether the reconstructed data is of a semantic level that can convey semantic information to downstream tasks based on this. Here, the measured error rate is If less than the value, the reconstruction data at the semantic level is passed as input to the downstream task to obtain the task output. That is, the reconstructed data at the semantic level is obtained when the threshold is satisfied. can be passed as input to a downstream task to obtain task output, which may be as shown in Fig. 23.

[0173] On the other hand, if semantic redundancy data is not received or a semantic redundancy check is not performed, the encoder of the destination (2720) can generate a bit operation attention map and measure the error rate by performing a comparison between bit operation attention maps.

[0174] FIG. 28 is a diagram illustrating a method for performing semantic communication based on an attention map applicable to the present disclosure. Referring to FIG. 28, a source (2810) and a destination (2820) can perform semantic communication based on S2601 of FIG. 26 described above, and the destination (2820) can perform semantic level reconstruction by receiving a semantic representation and a bit operation attention map. Here, the destination (2820) can generate semantic level reconstruction data from the semantic representation through a decoder, and provide the semantic level reconstruction data as an input to an encoder of the destination (2820) to generate a bit operation attention map in the same manner as the source (2820). Here, the destination (2820) can perform a semantic redundancy check using the semantic redundancy data received based on the attention area, and measure an error rate based on the semantic redundancy check, which may be the same as S2701 of FIG. 27.

[0175] As another example, if semantic redundancy data is not received or semantic redundancy check is not performed, the destination (2820) can generate a bit operation attention map in the same manner as the source (2820) by providing it as input to the destination's encoder, and can perform an error check by comparing the generated bit operation attention map with the bit operation attention map received from the source (2810), which may be the same as S2701 of FIG. 27.

[0176] For example, if the error rate is lower than the threshold, the destination (2820) may perform downstream tasks based on the semantic-level reconstruction data. Conversely, if the error rate is higher than the threshold, the destination (2820) may forward the error to the source (2810). For example, the source (2810) may control the semantic communication method based on the error, and is not limited to a specific embodiment.

[0177] Fig. 29 is a flowchart illustrating a first node operation method applicable to the present disclosure. Referring to Fig. 29, a first node can establish a connection with a second node (S2910). Here, the first node may be the aforementioned source, and the second node may be the aforementioned destination. As another example, the first node may be the aforementioned destination, and the second node may be the aforementioned source. In addition, as an example, the first node may be a terminal, and the second node may be a base station. In addition, as an example, the first node may be a base station, and the second node may be a terminal, and the present invention is not limited to a specific form. In addition, as an example, Fig. 29 may be applicable to both one-way communication and two-way communication, and is not limited to a specific form.

[0178] Here, the first node can receive a request for first node capability information from the second node (S2920), and based on this, transmit the first node capability information to the second node. (S2930) For example, the first node capability information can include information indicating whether the first node supports downstream task-based semantic communication. In addition, the first node capability information can further include at least one of first node creation information, first node collection information, processable raw data type information, and device computation capability information, as described above. Next, the first node can receive semantic communication instruction information from the second node. (S2940) Here, the semantic communication-related information in the semantic communication instruction information can include at least one of an encoder model, a decoder model, an attention matrix calculation method, a fusion matrix, an attention value discard ratio, a bitwise attention map threshold, an error detection threshold, and a semantic redundancy generation method, as described above. Thereafter, the first node can encode the input data using the attention technique through the received semantic communication-related information and transmit it to the second node. (S2950) Here, if the first node and the second node support downstream task-based semantic communication, the first node can generate a semantic representation and a bitwise attention map from the input data and transmit them to the second node. Here, the semantic representation can be generated from the input data based on the downstream task performed in the second node, as described above.

[0179] In addition, as an example, a bitwise attention map can be generated based on an attention matrix. Here, if the layer to which the attention technique is applied is a single layer, the bitwise attention map is generated using an attention matrix, and if the layer to which the attention technique is applied is a multi-layer, the bitwise attention map can be generated based on the attention matrices obtained from each layer. In addition, each of the attention matrices obtained from each layer can be generated based on the attention flow occurring from the previous layer to the next layer, as described above.

[0180] In addition, as an example, when the first node obtains target class information of a downstream task from the second node, the target class information may be reflected in the attention matrices obtained from each layer to generate a bit operation attention map. Here, a fusion matrix and an attention value discard ratio may be set and applied to the attention matrices obtained from each layer, and an attention matrix of the entire layer may be obtained based on the attention matrices of each layer to which the fusion matrix and the attention value discard ratio are applied. In addition, a bit operation attention map transmitted to the second node may be generated by comparing the attention matrix of the entire layer with a bit operation attention map threshold, as described above.

[0181] Additionally, as an example, the first node can apply a bitwise attention map to input data to derive an attention region, and generate semantic redundancy data, which is attention-aware redundancy, based on the attention region. Thereafter, the first node can transmit the semantic redundancy data, along with the semantic representation and the bitwise attention map, to the second node, as described above.

[0182] FIG. 30 is a flowchart illustrating a second node operation method applicable to the present disclosure. Referring to FIG. 30, the second node can establish a connection with the first node (S3010). Here, the first node may be the aforementioned source, and the second node may be the aforementioned destination. As another example, the first node may be the aforementioned destination, and the second node may be the aforementioned source. In addition, as an example, the first node may be a terminal, and the second node may be a base station. In addition, as an example, the first node may be a base station, and the second node may be a terminal, and the method is not limited to a specific form. In addition, as an example, FIG. 30 may be applicable to both one-way communication and two-way communication, and is not limited to a specific form.

[0183] Here, the second node can transmit a request for first node capability information to the first node (S3020), and based on this, can receive first node capability information from the first node (S3030). For example, the first node capability information can include information indicating whether the first node supports downstream task-based semantic communication. In addition, the first node capability information can further include at least one of first node creation information, first node collection information, processable raw data type information, and device computation capability information, as described above. Next, the second node can transmit semantic communication instruction information to the first node. (S3040) Here, the semantic communication-related information in the semantic communication instruction information can include at least one of an encoder model, a decoder model, an attention matrix calculation method, a fusion matrix, an attention value discard ratio, a bitwise attention map threshold, an error detection threshold, and a semantic redundancy generation method, as described above. Thereafter, the second node can receive input data encoded by an attention technique from the first node through the received semantic communication-related information. (S3050)

[0184] Here, when the first node and the second node support downstream task-based semantic communication, the second node can receive an encoded semantic representation and a first bitwise attention map from the first node based on input data. For example, the semantic representation may be generated by the first node from the input data based on a downstream task performed by the second node, and the second node may generate semantic-level reconstruction data based on the semantic representation. Thereafter, the second node may generate a second bitwise attention map from the semantic-level reconstruction data in the same manner as the first node. The second node may compare the first bitwise attention map and the second bitwise attention map to generate an error rate, and may perform error detection by comparing the error rate with an error detection threshold. Here, the second node may obtain an output by providing the semantic-level reconstruction data as an input to the downstream task if the error rate is less than the error detection threshold.

[0185] As another example, the second node may receive, from the first node, first semantic redundancy data generated based on a first attention region derived from input data, together with a semantic representation and a first bit operation attention map. Here, the second node may apply the first bit operation attention map received to the reconstruction data of the semantic level to derive a second attention region, and generate second semantic redundancy data based on the second attention region. The second node may generate an error rate by comparing the first semantic redundancy data with the second semantic redundancy data, and perform error detection by comparing the error rate with an error detection threshold, and if the error rate is less than the error detection threshold, provide the reconstruction data of the semantic level as an input to a downstream task to obtain an output, as described above.

[0186] It is clear that the examples of the proposed methods described above can also be considered as a type of proposed methods, as they can be included as one of the implementation methods of the present disclosure. Furthermore, the proposed methods described above can be implemented independently, but they can also be implemented in the form of a combination (or merge) of some of the proposed methods. Information regarding the applicability of the proposed methods (or information regarding the rules of the proposed methods) can be defined by a rule such that the base station notifies the terminal of the application of the proposed methods through a predefined signal (e.g., a physical layer signal or a higher layer signal).

[0187] The present disclosure may be embodied in other specific forms without departing from the technical ideas and essential features described herein. Therefore, the above detailed description should not be construed as limiting in all respects but rather as illustrative. The scope of the present disclosure should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the present disclosure are intended to be included within the scope of the present disclosure. Furthermore, claims that do not explicitly cite each other in the claims may be combined to form embodiments or incorporated into new claims through post-filing amendments.

[0188] Embodiments of the present disclosure can be applied to various wireless access systems. As an example of various wireless access systems, 3GPP (3 rd Generation Partnership Project) or 3GPP2 system.

[0189] The embodiments of the present disclosure can be applied not only to the various wireless access systems described above, but also to all technical fields that utilize these various wireless access systems. Furthermore, the proposed method can also be applied to mmWave and THz communication systems utilizing ultra-high frequency bands.

[0190] Additionally, embodiments of the present disclosure can be applied to various applications such as autonomous vehicles and drones.

Claims

1. In a method for operating a first node in a wireless communication system, Step of establishing a connection with a second node; A step of receiving a request for first node capability information from the second node; A step of transmitting first node capability information to the second node based on the first node capability information request; A step of receiving semantic communication instruction information from the second node; and A first node operation method for encoding input data using an attention technique through semantic communication-related information in the received semantic communication instruction information and transmitting the encoded input data to the second node.

2. In paragraph 1, The above first node capability information includes information indicating whether the first node supports downstream task-based semantic communication, A first node operation method, wherein the first node capability information further includes at least one of first node creation information, first node collection information, processable raw data type information, and device operation capability information.

3. In paragraph 2, A first node operating method, wherein the semantic communication-related information in the semantic communication instruction information includes at least one of an encoder model, a decoder model, an attention matrix calculation method, a fusion matrix, an attention value discard ratio, a bitwise attention map threshold, an error detection threshold, and a semantic redundancy generation method.

4. In paragraph 3, If the first node and the second node support the downstream task-based semantic communication, A first node operation method, wherein the first node generates a semantic representation and a bitwise attention map from the input data using the attention technique and transmits the generated semantic representation and a bitwise attention map to the second node.

5. In paragraph 4, A method of operating a first node, wherein the semantic representation is generated from the input data based on a downstream task performed in the second node.

6. In paragraph 4, The above bit operation attention map is generated based on the attention matrix, If the layer to which the attention technique is applied is a single layer, the bit operation attention map is generated with the attention matrix, A first node operation method, wherein, when the layer to which the above attention technique is applied is a multi-layer, the bit operation attention map is generated based on attention matrices obtained from each layer.

7. In paragraph 6, A first node operation method in which each of the attention matrices obtained from each of the above layers is generated based on the attention flow occurring from the previous layer to the next layer.

8. In paragraph 6, A first node operation method in which, when the first node obtains target class information of a downstream task from the second node, the target class information is reflected in the attention matrices obtained from each layer to generate the bit operation attention map.

9. In paragraph 6, The fusion matrix and the attention value discard ratio are set and applied to the attention matrices obtained from each layer above, An attention matrix of the entire layer is obtained based on the attention matrices of each layer to which the above fusion matrix and the attention value discard ratio are applied, A first node operation method, wherein the bit operation attention map transmitted to the second node is generated by comparing the attention matrix of the entire layer with the bit operation attention map threshold.

10. In paragraph 6, A first node operation method, wherein the bit operation attention map is applied to the input data to derive an attention region from the input data, and semantic redundancy data, which is attention recognition redundancy, is generated based on the attention region and transmitted to the second node together with the semantic expression and the bit operation attention map.

11. In a second node operation method in a wireless communication system, Step of establishing a connection with the first node; A step of transmitting a request for first node capability information to the first node; A step of receiving first node capability information from the first node based on the first node capability information request; a step of transmitting semantic communication instruction information to the first node; and A second node operation method for receiving a semantic expression and bit operation attention map generated from input data of the first node based on an attention technique through semantic communication-related information in the semantic communication instruction information from the first node.

12. In paragraph 11, The above first node capability information includes information indicating whether the first node supports downstream task-based semantic communication, A second node operating method, wherein the first node capability information further includes at least one of first node creation information, first node collection information, processable raw data type information, and device operation capability information.

13. In paragraph 12, A second node operating method, wherein the semantic communication-related information in the semantic communication instruction information includes at least one of an encoder model, a decoder model, an attention matrix calculation method, a fusion matrix, an attention value discard ratio, a bitwise attention map threshold, an error detection threshold, and a semantic redundancy generation method.

14. In paragraph 13, If the first node and the second node support the downstream task-based semantic communication, A second node operation method, wherein the second node receives an encoded semantic representation and a first bitwise attention map based on the input data from the first node.

15. In paragraph 14, The above semantic representation is generated by the first node from the input data based on the downstream task performed by the second node, A second node operation method, wherein the second node generates reconstructed data at a semantic level based on the semantic expression.

16. In paragraph 15, The second node generates a second bit operation attention map from the reconstruction data of the semantic level in the same manner as the first node, Comparing the first bit operation attention map and the second bit operation attention map to generate an error rate, A second node operation method, wherein error detection is performed by comparing the error rate with the error detection threshold, and if the error rate is less than the error detection threshold, the semantic level reconstruction data is provided as an input to the downstream task to obtain an output.

17. In paragraph 15, A second node operation method, wherein the second node receives first semantic redundancy data generated based on the first attention region derived from the input data from the first node together with the semantic expression and the first bit operation attention map.

18. In paragraph 17, Applying the first bit operation attention map received to the reconstructed data of the semantic level to derive a second attention region, and generating second semantic redundancy data based on the second attention region, Generate an error rate by comparing the first semantic redundancy data and the second semantic redundancy data, A second node operation method for performing error detection by comparing the error rate with the error detection threshold, and obtaining output by providing the reconstruction data of the semantic level as an input to the downstream task if the error rate is less than the error detection threshold.

19. In a first node in a wireless communication system, Transmitter and receiver; and A processor connected to the above transmitter and receiver is included, The above processor, Establish a connection with the second node, Control the transceiver to receive a first node capability information request from the second node, Control the transceiver to transmit the first node capability information to the second node based on the first node capability information request, Controlling the transceiver to receive semantic communication instruction information from the second node, and A first node that controls the transmission of input data to the second node by encoding the input data using an attention technique through semantic communication-related information in the received semantic communication instruction information.

20. In a second node in a wireless communication system, Transmitter and receiver; and A processor connected to the above transmitter and receiver is included, The above processor, Establish a connection with the first node, Control the transceiver to transmit a first node capability information request to the first node, Control the transceiver to receive first node capability information from the first node based on the first node capability information request, Controlling the transceiver to transmit semantic communication instruction information to the first node, and A second node that controls the transceiver to receive, from the first node, a semantic expression and bit operation attention map generated from the input data of the first node based on an attention technique through semantic communication-related information in the received semantic communication instruction information.

21. A device comprising at least one memory and at least one processor functionally connected to the at least one memory, At least one processor of the device, Establish a connection with another device, Controlling the device to receive a device capability information request from the other device; Controlling the device to transmit device capability information to the other device based on the device capability information request; Controlling the device to receive semantic communication instruction information from the other device, and A device that controls the transmission of input data to another device by encoding the input data using an attention technique through semantic communication-related information in the received semantic communication instruction information.

22. In a non-transitory computer-readable medium storing at least one instruction, comprising at least one instruction executable by the processor, At least one of the above commands causes the device to: Establish a connection with another device, Controlling the device to receive a device capability information request from the other device; Controlling the device to transmit device capability information to the other device based on the device capability information request; Controlling the device to receive semantic communication instruction information from the other device, and A computer-readable medium that controls the transmission of input data to another device by encoding the input data using an attention technique through semantic communication-related information in the received semantic communication instruction information.

Images