Semantic communication method, device and system

Abstract

The invention provides a semantic communication method and a related product. The method comprises the following steps: receiving semantic lexical elements and lexical element configuration from central equipment; determining whether the perception data is matched with the semantic lexical elements or not, wherein the lexical element configuration is used for obtaining the perception lexical elements from the perception data; when the perception data is matched with the semantic lexical elements, a perception result is sent to the center equipment, and the perception result indicates the perception data. The sending of the perception result is triggered when the perception data is matched with the semantic lexical elements, namely, the perception result is not sent all the time, so that transmission resources are saved; besides, the sent sensing result meets the requirements of semantic lexical elements, so that irrelevant information is filtered, and the sent data is required by the central equipment, so that the response accuracy is ensured.

Description

[0001] Cross Reference to Related Applications

[0002] This application claims priority to U.S. Provisional Patent Application No. 63 / 509,453, filed on June 21, 2023, entitled “Semantic Communication Method, Apparatus and System,” the entire contents of which are incorporated herein by reference. TECHNICAL FIELD

[0003] The present disclosure relates generally to the field of communication technology, and more particularly to a communication method, a communication apparatus, a communication system, and related products. BACKGROUND

[0004] Sensing functionalities will be integrated into the 6th generation (6G) systems. A large number of sensing user equipments (UEs) or sensing devices will be densely deployed in cities, factories, farms, etc. In addition to cellphones, sensing devices will become an important class of UEs or devices that announce the arrival of the Internet of Things (IoT) era. Like Internet search engines, 6G will launch a corresponding IoT search engine in the real physical world. In fact, billions of IoT-based applications, such as self-driving cars, automated factories, smart cities, autonomous farms, etc., will largely rely on an efficient, real-time search engine in the physical world.

[0005] In recent years, artificial intelligence (AI) has conquered various fields of knowledge and cognitive fields. Some AI is exploring the frontier knowledge in the fields of chemistry, games, mathematics, genetic engineering, etc., while other AI is providing a human-level question and answer platform in the digital world. The field that AI has not yet conquered is the real-time physical world. The AI of the physical world can be built on the ubiquitous IoT connectivity brought by 6G, in which AI technology will penetrate into all aspects of society and life.

[0006] More challenging than Internet search engines is that the search engine of the real world must search the physical world in real time in a large-scale physical area and handle multiple types of data and information. In addition, green technology, low energy consumption, and low emissions are also proposed as key features of 6G. Sensing devices can be powered by batteries and / or completely powered by solar and wind energy. In some implementations, the sensing device can be a UE, a cellphone, or a headset, in which it is assumed that any two sensing devices are independent of each other; therefore, the wireless system associated with the sensing device can schedule the sensing device independently; the sensing data measured by the sensing device can be an application-level payload of the wireless system and protocol. The above-mentioned scheme of scheduling the sensing device is low in terms of wireless bandwidth and energy consumption.

[0007] The purpose of providing this background art is to disclose information which the applicant believes to be relevant to the present application, and is not admitted to be prior art to the present application by virtue of prior art. SUMMARY

[0008] In a first aspect, the present disclosure provides a communication method, comprising:

[0009] receiving a semantic token and a tokenization configuration from a central device;

[0010] determining whether perception data matches the semantic token, the tokenization configuration being used to obtain a perception token from the perception data;

[0011] when the perception data matches the semantic token, sending a perception result to the central device, the perception result indicating the perception data.

[0012] The semantic token from the central device wakes up the perception device to perform measurement and send the perception result when the perception data matches the semantic token, that is, the perception device initiates the calculation of the matching determination in response to the semantic token, and in other cases, the calculation of the matching determination can not be initiated, thereby reducing the energy consumption of the perception device. In addition, since the form of the semantic token can provide a more accurate true intention, the signaling overhead is saved, thereby improving the accuracy of the matching result between the perception data and the semantic token and saving the signaling overhead. The tokenization configuration can be used for tokenization of the perception data, thereby ensuring that the matching object adopts a common token form, so that the perception data and the semantic token can be easily compared with each other. And the sending of the perception result is triggered when the perception data matches the semantic token, that is, the perception result is not always sent, thereby saving transmission resources; in addition, the perception result sent meets the requirements of the semantic token, thereby filtering irrelevant information, and the data sent is required by the central device, thereby ensuring response accuracy.

[0013] In a possible implementation manner of the first aspect, the receiving the semantic token and the tokenization configuration from the central device comprises:

[0014] receiving the semantic token and the tokenization configuration broadcast or multicasted by the central device.

[0015] The received semantic token and tokenization configuration are broadcast or multicasted by the central device, the semantic token and the tokenization configuration can be sent to multiple perception devices, thereby improving transmission efficiency; through the same tokenization configuration, the perception data of multiple perception devices can be tokenized, so that the perception devices can determine whether to send their perception results under the same standard.

[0016] In a possible implementation form of the first aspect, the semantic wordpiece comprises a semantic wordpiece of one task or a semantic wordpiece of one modality.

[0017] The perception device can serve one task or one modality, that is, the perception device is dedicated to one task or one modality, and thus the response efficiency of the perception device can be improved.

[0018] In a possible implementation form of the first aspect, the method further comprises:

[0019] receiving an indication of the task identity or an indication of the modality identity.

[0020] In a possible implementation form of the first aspect, the semantic wordpiece comprises a plurality of semantic wordpieces of a plurality of tasks and a plurality of task identities respectively associated with the plurality of semantic wordpieces.

[0021] The perception device can serve a plurality of tasks at the same time, and thus the resource utilization of the perception device is improved.

[0022] In a possible implementation form of the first aspect, the semantic wordpiece comprises a plurality of semantic wordpieces of a plurality of modalities and a plurality of modality identities respectively associated with the plurality of semantic wordpieces.

[0023] The perception device can serve a plurality of modalities at the same time, that is, the perception device can perceive and feedback more than one modality (i.e. one type of data), which enriches the diversity of data used by the perception device for response.

[0024] In a possible implementation form of the first aspect, the semantic wordpiece comprises a first plurality of semantic wordpieces of a plurality of tasks, a second plurality of semantic wordpieces of a plurality of modalities, a plurality of task identities respectively associated with the first plurality of semantic wordpieces, and a plurality of modality identities respectively associated with the second plurality of semantic wordpieces.

[0025] The perception device can serve a plurality of tasks and a plurality of modalities at the same time, and thus not only the resource utilization of the perception device is improved, but also the diversity of data used by the perception device for response is enriched.

[0026] In a possible implementation form of the first aspect, the perception result comprises one of:

[0027] raw perception data;

[0028] perception semantics obtained from the raw perception data;

[0029] semi-raw perception data and perception semantics obtained from the raw perception data;

[0030] the raw perception data and a matching score between the perception data and the semantic wordpiece.

[0031] a distance between the raw perception data and the semantic token;

[0032] a matching score between the raw perception data and the semantic token;

[0033] a distance between the raw perception data and the semantic token;

[0034] semi-raw perception data, a matching score between the raw perception data and the semantic token, and a perception semantic obtained from the raw perception data;

[0035] semi-raw perception data, a distance between the raw perception data and the semantic token, and a perception semantic obtained from the raw perception data.

[0036] The perception result can take various forms related to the perception data, which provides more flexibility and can meet different needs.

[0037] In a possible implementation form of the first aspect, the perception result further comprises a task identification or a modality identification.

[0038] The task identification is used to distinguish a certain task, and the modality identification is used to distinguish a certain modality. Since the task identification or the modality identification is included in the perception result, the central device can easily identify which task or which modality is carried in a certain perception result.

[0039] In a possible implementation form of the first aspect, the perception result further comprises an identification of a semantic token having a matching score greater than or equal to a first threshold or a distance less than a second threshold.

[0040] The identification of the semantic token is used to distinguish the semantic token, and the semantic token having the matching score greater than or equal to the first threshold or the distance less than the second threshold indicates that the perception device has the perception data corresponding to the semantic token. Since the identification is included in the perception result, the central device can easily identify which semantic token is carried in a certain perception result.

[0041] In a possible implementation form of the first aspect, the perception result further comprises a plurality of identifications of a plurality of semantic tokens having a matching score greater than or equal to a first threshold or a distance less than a second threshold.

[0042] In a possible implementation form of the first aspect, the receiving, from the central device, the semantic token and the tokenization configuration comprises:

[0043] receiving, from the central device, a sequence of a predefined length and the tokenization configuration.

[0044] The semantic token can be encoded as a sequence of a predefined length, and thus the transmission of the semantic token can be integrated into a physical layer of a wireless communication system for the purpose of short latency.

[0045] In a possible implementation form of the first aspect, the sequence of the predefined length comprises a pilot code, a mask code, or an interleaving sequence.

[0046] In a possible implementation form of the first aspect, the receiving the semantic token and the tokenization configuration from the central device comprises:

[0047] receiving, from the central device, a payload with a fixed modulation and coding scheme (MCS) and the tokenization configuration.

[0048] The semantic token can be encoded as a payload with a fixed MCS, and thus the semantic token can be encoded directly with a specific code, and the transmission of the semantic token can be integrated into a physical layer of a wireless communication system for the purpose of short latency.

[0049] In a possible implementation form of the first aspect, the method further comprises:

[0050] obtaining the perception data;

[0051] translating the perception data into a perception semantic according to a semanticization configuration;

[0052] tokenizing the perception semantic into the perception token according to the tokenization configuration;

[0053] determining whether the perception data matches the semantic token comprises:

[0054] determining whether the perception token matches the semantic token.

[0055] In case the perception data is in a natural language form, the perception data can be processed into a perception token, and then a comparison between the perception token and the semantic token can be implemented. Since the form of the token can provide a more accurate true intention, signaling overhead is saved, and thus the accuracy of the comparison result is improved and signaling overhead is saved. In addition, tokenization can be used to prevent the perception device from recovering a complete query message from the semantic token. Tokenization can provide a certain privacy protection for the query message.

[0056] In a possible implementation form of the first aspect, the determining whether the perception token matches the semantic token comprises:

[0057] determining whether a matching score between the perception token and the semantic token is greater than or equal to a first threshold; or

[0058] determine whether the distance between the perceptual token and the semantic token is less than or equal to a second threshold value.

[0059] According to the matching score between the perceptual data and the semantic token, or according to the distance between the perceptual data and the semantic token, it can be determined whether the perceptual data matches the semantic token, thereby providing an efficient implementation. It should be noted that the correlation score between the perceptual data and the semantic token, the similarity between the perceptual data and the semantic token, and other indicators can also be used to indicate whether the perceptual data matches the semantic token.

[0060] In a possible implementation of the first aspect, the tokenization configuration comprises one of the following:

[0061] a tokenization model;

[0062] a tokenization function;

[0063] a projection matrix;

[0064] a graph-based or topology-based pruning; or

[0065] a compression method.

[0066] There are various tokenization configurations to choose from, providing more flexibility, thereby meeting different needs.

[0067] In a possible implementation of the first aspect, the method further comprises:

[0068] receiving, from the central device, at least one of a scoring function for determining a matching score between the perceptual data and the semantic token, a first threshold value, a function for determining a distance between the perceptual data and the semantic token, or a second threshold value.

[0069] The matching score between the perceptual data and the semantic token can be calculated by a scoring function, and the distance between the perceptual data and the semantic token can be calculated by a corresponding function, thereby providing an efficient implementation. At least one of the scoring function for determining the matching score, the first threshold value, the function for determining the distance, and the second threshold value can be sent from the central device to the perceptual device, which provides more flexibility, thereby meeting different needs.

[0070] In a possible implementation of the first aspect, the receiving, from the central device, the semantic token and the tokenization configuration comprises:

[0071] receiving, from the central device, the semantic token, the tokenization configuration, and at least one of a scoring function for determining a matching score between the perception data and the semantic token, a first threshold, a function for determining a distance between the perception data and the semantic token, or a second threshold.

[0072] The at least one of the scoring function for determining a matching score, the first threshold, the function for determining a distance, the second threshold can be sent from the central device to the perception device together with the semantic token and the tokenization configuration, which provides more flexibility so that different requirements can be met.

[0073] In a possible implementation manner of the first aspect, the semantic token is carried in high layer signaling or physical layer signaling or a combination of high layer signaling and physical signaling, the high layer signaling including one of radio resource control (RRC) signaling, medium access control (MAC) layer signaling, and the physical layer signaling including downlink control information (DCI).

[0074] The semantic token can be sent from the central device to the perception device through specific signaling, which guarantees high security and high efficiency of data transmission.

[0075] In a possible implementation manner of the first aspect, the semantic token includes a query token.

[0076] In a second aspect, the present disclosure provides a communication method, the method comprising:

[0077] sending, to a perception device, a semantic token and a tokenization configuration;

[0078] receiving, from the perception device, a perception result, wherein perception data of the perception device matches the semantic token, the tokenization configuration is used to obtain a perception token from the perception data, and the perception result indicates the perception data.

[0079] The central device sends a semantic token and a tokenization configuration to a perception device, the form of the semantic token can provide more accurate real intention and save signaling overhead, and the perception device can tokenize perception data using the tokenization configuration. The central device receives a perception result in the case that perception data matches the semantic token, the received perception result meets the requirement of the semantic token, that is, irrelevant information is filtered, and the received data is what the central device needs, thereby guaranteeing the accuracy of data transmission.

[0080] In a possible implementation manner of the second aspect, the sending of the semantic token and the tokenization configuration to the perception device comprises:

[0081] The semantic token and the tokenization configuration are broadcasted or multicasted to a plurality of perception devices.

[0082] The central device broadcasts or multicasts the semantic token and the tokenization configuration to a plurality of perception devices, thereby improving transmission efficiency; through the same tokenization configuration, the perception data of a plurality of perception devices can be tokenized, so that the perception devices can determine whether to send their perception results under the same standard.

[0083] In a possible implementation manner of the second aspect, the semantic token comprises a semantic token of one task or a semantic token of one modality.

[0084] In a possible implementation manner of the second aspect, the method further comprises:

[0085] Sending an indication of a task identity or an indication of a modality identity.

[0086] In a possible implementation manner of the second aspect, the semantic token comprises a plurality of semantic tokens of a plurality of tasks and a plurality of task identities respectively associated with the plurality of semantic tokens.

[0087] In a possible implementation manner of the second aspect, the semantic token comprises a plurality of semantic tokens of a plurality of modalities and a plurality of modality identities respectively associated with the plurality of semantic tokens.

[0088] In a possible implementation manner of the second aspect, the semantic token comprises a first plurality of semantic tokens of a plurality of tasks, a second plurality of semantic tokens of a plurality of modalities, a plurality of task identities respectively associated with the first plurality of semantic tokens, and a plurality of modality identities respectively associated with the second plurality of semantic tokens.

[0089] In a possible implementation manner of the second aspect, the perception result comprises one of:

[0090] Original perception data;

[0091] Perception semantics obtained from the original perception data;

[0092] Semi-original perception data and perception semantics obtained from the original perception data;

[0093] The original perception data and a matching score between the perception data and the semantic token;

[0094] The original perception data and a distance between the perception data and the semantic token;

[0095] a semantic wordpiece obtained from raw perception data and a matching score between the perception data and the semantic wordpiece;

[0096] a semantic wordpiece obtained from raw perception data and a distance between the perception data and the semantic wordpiece;

[0097] semi-raw perception data, a semantic wordpiece obtained from raw perception data and a matching score between the perception data and the semantic wordpiece;

[0098] semi-raw perception data, a semantic wordpiece obtained from raw perception data and a distance between the perception data and the semantic wordpiece.

[0099] The perception result can take various forms related to the perception data, which provides more flexibility and can meet different needs.

[0100] In a possible implementation of the second aspect, the perception result further comprises a task identification or a modality identification.

[0101] The task identification is used to distinguish a certain task, and the modality identification is used to distinguish a certain modality. Since the task identification or the modality identification is included in the perception result, the central device can easily identify which task or which modality is carried in a certain perception result.

[0102] In a possible implementation of the second aspect, the perception result further comprises an identification of a semantic wordpiece whose matching score is greater than or equal to a first threshold or whose distance is less than a second threshold.

[0103] In a possible implementation of the second aspect, the perception result further comprises a plurality of identifications of a plurality of semantic wordpieces whose matching scores are greater than or equal to a first threshold or whose distances are less than a second threshold.

[0104] The identification of the semantic wordpiece is used to distinguish the semantic wordpiece, and the semantic wordpiece whose matching score is greater than or equal to the first threshold or whose distance is less than the second threshold indicates that the perception device has the perception data corresponding to the semantic wordpiece. Since the identification is included in the perception result, the central device can easily identify which semantic wordpiece is carried in a certain perception result.

[0105] In a possible implementation of the second aspect, the method further comprises:

[0106] encoding the semantic wordpiece into a sequence of a predefined length;

[0107] The sending, to the perception device, of the semantic wordpiece and the wordpiece configuration comprises:

[0108] sending, to the perception device, the sequence of the predefined length and the wordpiece configuration.

[0109] The semantic token can be encoded as a sequence of a predefined length, and thus the transmission of the semantic token can be integrated into a physical layer of a wireless communication system for the purpose of short latency.

[0110] In a possible implementation of the second aspect, the sequence of the predefined length comprises a pilot code, a mask code, or an interleaving sequence.

[0111] In a possible implementation of the second aspect, the method further comprises:

[0112] encoding the semantic token into a payload using a fixed modulation and coding scheme (MCS);

[0113] The sending of the semantic token and the tokenization configuration to the perception device comprises:

[0114] sending the payload using the fixed MCS and the tokenization configuration to the perception device.

[0115] The semantic token can be encoded into a payload using a fixed MCS, and thus the semantic token can be encoded directly using a specific code, and the transmission of the semantic token can be integrated into a physical layer of a wireless communication system for the purpose of short latency.

[0116] In a possible implementation of the second aspect, the semantic token comprises a query token.

[0117] In a possible implementation of the second aspect, the method further comprises:

[0118] receiving semantics from a generative pre-trained transformer (GPT) device;

[0119] tokenizing the semantic token into the semantic token;

[0120] outputting the perception result to the GPT device.

[0121] The center device receives semantics from the GPT device and processes the semantics into a semantic token. Next, the center device provides the semantic token and a tokenization configuration for the perception device to make a decision, and then receives a perception result from the perception device and outputs the perception result to the GPT device. The center device serves as a bridge between the perception device and the GPT device, and thus facilitates smooth communication between the perception device and the GPT device.

[0122] In a possible implementation of the second aspect, the method further comprises:

[0123] determining the tokenization configuration.

[0124] The tokenization process can have different tokenization configurations, and a suitable tokenization configuration can correspond to a specific tokenization, thereby ensuring high efficiency of data processing.

[0125] In a possible implementation of the second aspect, the method further includes:

[0126] determining a token length of the semantic token.

[0127] Determining the token length of the semantic token can facilitate subsequent processing, for example, the token length of the perceptual token can be the same as the length of the semantic token.

[0128] In a possible implementation of the second aspect, the tokenization configuration includes one of the following:

[0129] a tokenization model;

[0130] a tokenization function;

[0131] a projection matrix;

[0132] a graph-based or topology-based pruning; or

[0133] a compression method.

[0134] There are various tokenization configurations to choose from, providing more flexibility, thereby meeting different needs.

[0135] In a possible implementation of the second aspect, the method further includes: sending, to the perception device, at least one of a scoring function for determining a matching score between the perception data and the semantic token, a first threshold, a function for determining a distance between the perception data and the semantic token, or a second threshold.

[0136] The matching score between the perception data and the semantic token can be calculated by the scoring function, and the distance between the perception data and the semantic token can be calculated by the corresponding function, thereby providing an efficient implementation. At least one of the scoring function for determining the matching score, the first threshold, the function for determining the distance, and the second threshold can be sent from the central device to the perception device, which provides more flexibility, thereby meeting different needs.

[0137] In a possible implementation of the second aspect, the sending, to the perception device, of the semantic token and the tokenization configuration includes:

[0138] The semantic wordpiece, the wordpiece configuration, and at least one of a scoring function for determining a matching score between the perception data and the semantic wordpiece, a first threshold, a function for determining a distance between the perception data and the semantic wordpiece, or a second threshold are sent to the perception device.

[0139] At least one of the scoring function for determining a matching score, the first threshold, the function for determining a distance, the second threshold can be sent from the central device to the perception device together with the semantic wordpiece and the wordpiece configuration, which provides more flexibility so that different requirements can be met.

[0140] In a possible implementation manner of the second aspect, the semantic wordpiece is carried in high layer signaling or physical layer signaling or a combination of high layer signaling and physical signaling, the high layer signaling including one of radio resource control (RRC) signaling, medium access control (MAC) layer signaling, and the physical layer signaling including downlink control information (DCI).

[0141] The semantic wordpiece can be sent from the central device to the perception device through specific signaling, which guarantees high security and high efficiency of data transmission.

[0142] In a third aspect, the present disclosure provides a communication apparatus, the apparatus comprising various modules for performing the method according to the first aspect or any possible implementation manner of the first aspect.

[0143] In a fourth aspect, the present disclosure provides a communication apparatus, the apparatus comprising various modules for performing the method according to the second aspect or any possible implementation manner of the second aspect.

[0144] In a fifth aspect, the present disclosure provides a perception device, the perception device comprising processing circuitry for performing the method according to the first aspect or any possible implementation manner of the first aspect.

[0145] In a sixth aspect, the present disclosure provides a central device, the central device comprising processing circuitry for performing the method according to the second aspect or any possible implementation manner of the second aspect.

[0146] In a seventh aspect, the present disclosure provides a communication system, the communication system comprising the perception device according to the fifth aspect and the central device according to the sixth aspect.

[0147] In an eighth aspect, the present disclosure provides a chip comprising an input / output (I / O) interface and a processor configured to invoke and run a computer program stored in a memory, so that a device installed with the chip is capable of performing the method according to the first or second aspect or any possible implementation manner of the first or second aspect.

[0148] In a ninth aspect, the present disclosure provides a computer readable medium storing computer-executable instructions that, when executed by a processor, cause the processor to perform the method according to the first or second aspect or any possible implementation manner of the first or second aspect.

[0149] In a tenth aspect, the present disclosure provides a computer program product comprising computer-executable instructions that, when executed by a processor, cause the processor to perform the method according to the first or second aspect or any possible implementation manner of the first or second aspect.

[0150] The present disclosure provides a communication method and related products. A central device sends a semantic token and tokenization configuration to a sensing device, and the sensing device initiates the calculation of a matching judgment in response to the semantic token. The form of the semantic token can provide more accurate real intentions and save signaling overhead, and the sensing device can use the tokenization configuration to tokenize the sensing data. The sending of the sensing result is triggered in the case that the sensing data of the sensing device matches the semantic token, that is, the sending of the sensing result of the sensing device has certain conditions, rather than always sending the sensing result, which saves transmission resources; in addition, the sent sensing result meets the requirements of the semantic token, that is, irrelevant information is filtered, and the sent data is required by the central device, thereby ensuring the accuracy of data transmission. BRIEF DESCRIPTION OF DRAWINGS

[0151] The accompanying drawings are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the present disclosure and serve to explain the present disclosure, and should not be understood as limiting the present disclosure.

[0152] Figure 1 is a schematic diagram of a communication system provided by one or more example embodiments of the present disclosure.

[0153] Figure 2 is another schematic diagram of a communication system provided by one or more example embodiments of the present disclosure.

[0154] Figure 3 is a schematic diagram of a basic component structure of a communication system provided by one or more example embodiments of the present disclosure.

[0155] Figure 4A block diagram of a device in a communication system provided by one or more examples of the present disclosure is shown.

[0156] Figure 5 A schematic flow diagram of a communication method provided by one or more examples of the present disclosure is shown.

[0157] Figure 6 A schematic flow diagram of another communication method provided by one or more examples of the present disclosure is shown.

[0158] Figure 7 A schematic flow diagram of yet another communication method provided by one or more examples of the present disclosure is shown.

[0159] Figure 8 A further schematic diagram of a communication system provided by one or more examples of the present disclosure is shown.

[0160] Figure 9 A schematic diagram of a partitioning of a perception device provided by one or more examples of the present disclosure is shown.

[0161] Figure 10 A schematic diagram of an interaction between devices in a communication system provided by one or more examples of the present disclosure is shown.

[0162] Figure 11 A further schematic diagram of an interaction between devices in a communication system provided by one or more examples of the present disclosure is shown.

[0163] Figure 12 A schematic diagram of a generation of a sequence of query messages in a communication system provided by one or more examples of the present disclosure is shown.

[0164] Figure 13 A schematic diagram of an interaction between a central device and two perception devices in a communication system provided by one or more examples of the present disclosure is shown.

[0165] Figure 14 A further schematic diagram of an interaction between a central device and two perception devices in a communication system provided by one or more examples of the present disclosure is shown.

[0166] Figure 15 A schematic diagram of a generation of query semantics by a GPT device in a communication system provided by one or more examples of the present disclosure is shown.

[0167] Figure 16 A schematic diagram of a recovery of query messages from query semantics in a communication system provided by one or more examples of the present disclosure is shown.

[0168] Figure 17 A schematic diagram of a generation of query tokens by a GPT device in a communication system provided by one or more examples of the present disclosure is shown.

[0169] Figure 18is a diagram of a perception device responding to query tokens from a hub device provided by one or more examples of the present disclosure.

[0170] Figure 19 is a diagram of a scoring operation implemented by a perception device provided by one or more examples of the present disclosure.

[0171] Figure 20 is a diagram of a perception result provided by one or more examples of the present disclosure.

[0172] Figure 21 is a diagram of two GPT devices generating query tokens provided by one or more examples of the present disclosure.

[0173] Figure 22 is a diagram of a perception device processing two query tokens using one common semantic model and two tokenization models provided by one or more examples of the present disclosure.

[0174] Figure 23 is a diagram of a perception device processing two query tokens using one common semantic model and one common tokenization model provided by one or more examples of the present disclosure.

[0175] Figure 24 is a diagram of a perception device processing two query tokens using two semantic models and two tokenization models provided by one or more examples of the present disclosure.

[0176] Figure 25 is a diagram of a perception device processing two query tokens using two semantic models and one common tokenization model provided by one or more examples of the present disclosure.

[0177] Figure 26 is a diagram of independently processing two perception semantics provided by one or more examples of the present disclosure.

[0178] Figure 27 is a block diagram of a communication apparatus provided by one or more examples of the present disclosure.

[0179] Figure 28 is a block diagram of another communication apparatus provided by one or more examples of the present disclosure. DETAILED DESCRIPTION

[0180] In the following description, reference is made to the accompanying drawings which form a part hereof, and which are shown by way of illustration of specific aspects of examples of the present disclosure. It is understood that examples of the present disclosure can be used in other aspects not depicted herein. Thus, the following detailed description is not to be taken in a limiting sense, and the scope of the present disclosure is defined by the appended claims.

[0181] To help understand the present disclosure, examples of wireless communication systems and devices are described below.

[0182] Examples of communication systems and devices

[0183] The present disclosure takes the interaction and processing procedures between at least one UE (i.e., a sensing device, also referred to as a sensing node, in Figure 1 , marked as ED), at least one BS (i.e., a central device), and at least one GPT device as an illustrative example. The exchanged information and protocol flows can also be used between other network nodes described below, for example, between ED 110 and TRP 170, between ED 110 and the core network, between ED 110 and ED 110, between TRP 170 and TRP 170, between TRP 170 and GPT device 180. The UE in the processes described in the present disclosure can be replaced by the sensing node mentioned below. The BS in the processes described in the present disclosure can be replaced by a sensing coordinator. The sensing coordinator is a node in the network that can assist in sensing operations. These nodes can be independent nodes dedicated only to sensing operations, or other nodes that perform sensing operations in parallel with communication transmission (e.g., TRP 170, ED 110, or core network nodes shown below).

[0184] Referring to Figure 1 , a simplified schematic diagram of a communication system is provided as an illustrative example but without limitation. The communication system 100 (which can be a wireless system in Figure 1 ) includes a wireless access network 120. The wireless access network 120 can be a next generation (e.g., sixth generation (6G) or higher) wireless access network or a legacy (e.g., 5G, 4G, 3G, or 2G) wireless access network. One or more communication electric devices (EDs) 110a, 110b, 110c, 110d, 110e, 110f, 110g, 110h, 110i, 110j (collectively referred to as 110) can be interconnected to each other and / or connected to one or more network nodes (170a, 170b, collectively referred to as 170) in the wireless access network 120. The core network 130 can be part of the communication system and can rely on or be independent of the wireless access technology used in the communication system 100. In addition, the communication system 100 includes a public switched telephone network (PSTN) 140, the Internet 150, and other networks 160.

[0185] Further, the communication system 100 includes at least one GPT device 180. The GPT device 180 can be located within one or more network nodes 170. The GPT device 180 can be a standalone device connected to the network 170, e.g., an ED 110 connected to a network node 170 over a Uu interface. The GPT device 180 can be a device connected to a network node 170 over the core network 130. When the GPT device 180 is an ED, uplink messages / data sent between a central device (e.g., a network node 170) and the GPT device 180 can be carried in higher layer signaling, e.g., RRC signaling, or MAC layer signaling. Alternatively, they can be carried in physical layer signaling, e.g., UCI. Alternatively, they can be carried in a combination of higher layer signaling and physical signaling. It should be noted that messages in this disclosure can be replaced by information, which can be carried in one single message, or in more than one single message. Downlink messages / data sent between a central device and the GPT device 180 can be carried in higher layer signaling, e.g., RRC signaling, or MAC layer signaling. Alternatively, they can be carried in physical layer signaling, e.g., DCI. Alternatively, they can be carried in a combination of higher layer signaling and physical signaling. It should be noted that messages in this disclosure can be replaced by information, which can be carried in one single message, or in more than one single message.

[0186] Figure 2 An example communication system 100 is shown. Generally, the communication system 100 is capable of enabling multiple wireless or wireline elements to communicate data and other content. The communication system 100 can be used to provide voice, data, video, signaling, and / or text content, among other content, through broadcasting, multicasting, and unicasting, among other techniques. The communication system 100 can operate by sharing resources, such as a carrier frequency bandwidth, among its constituent elements. The communication system 100 can include a terrestrial communication system and / or a non-terrestrial communication system. The communication system 100 can provide a wide range of communication services and applications, such as earth monitoring, remote sensing, passive sensing and positioning, navigation and tracking, autonomous delivery and mobility, among others. The communication system 100 can provide high availability and robustness through joint operation of the terrestrial communication system and the non-terrestrial communication system. For example, integrating a non-terrestrial communication system (or components thereof) into a terrestrial communication system can enable a heterogeneous network comprising multiple tiers. The heterogeneous network can achieve better overall performance compared to traditional communication networks through efficient multi-link joint operation, more flexible function sharing, and faster physical layer link switching between the terrestrial network and the non-terrestrial network.

[0187] The terrestrial communication system and the non-terrestrial communication system can be considered as subsystems of a communication system. In Figure 2In the illustrated example, the communication system 100 includes electronic devices (EDs) 110a, 110b, 110c, 1 lOd (collectively referred to as EDs 110), radio access networks (RANs) 120a and 120b, a non-terrestrial communication network 120c, a core network 130, a public switched telephone network (PSTN) 140, the Internet 150, and other networks 160. The RANs 120a and 120b include respective base stations (BSs) 170a and 170b, which can generally be referred to as terrestrial transmit and receive points (T-TRPs) 170a and 170b. The non-terrestrial communication network 120c includes an access node 172, which can generally be referred to as a non-terrestrial transmit and receive point (NT-TRP) 172.

[0188] Alternatively or additionally, any of the EDs 110 can be configured to connect, access, or communicate with any of the T-TRPs 170a and 170b, the NT-TRP 172, the Internet 150, the core network 130, the PSTN 140, the other networks 160, or any combination of the foregoing. In some examples, the ED 110a can communicate uplink and / or downlink transmissions with the T-TRP 170a over a terrestrial air interface 190a. In some examples, the EDs 110a, 110b, 110c, and 1 lOd can also communicate directly with each other over one or more sidelink air interfaces 190b. In some examples, the ED 1 lOd can communicate uplink and / or downlink transmissions with the NT-TRP 172 over a non-terrestrial air interface 190c.

[0189] The air interfaces 190a and 190b can use similar communication techniques, for example, any suitable wireless access technique. For example, the communication system 100 can implement one or more channel access methods in the air interfaces 190a and 190b, such as code division multiple access (CDMA), space division multiple access (SDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), Direct Fourier Transform spread OFDMA (DFT-OFDMA), or single-carrier FDMA (SC-FDMA). The air interfaces 190a and 190b can utilize other multi-dimensional signal spaces that can include combinations of orthogonal and / or non-orthogonal dimensions.

[0190] The non-terrestrial air interface 190c can enable communication between the ED 110d and one or more NT-TRPs 172 through a wireless link or simply a link. For some examples, the link is a dedicated connection for unicast transmission, a connection for broadcast transmission, or a connection between a group of EDs 110 and one or more NT-TRPs 172 for groupcast transmission.

[0191] The RANs 120a and 120b are in communication with the core network 130 to provide the EDs 110a, 110b, and 110c with access to various services, such as voice, data, and other services. The RANs 120a and 120b and / or the core network 130 can be in direct or indirect communication with one or more other RANs (not shown) that can or can not be of the same

[0192] Basic component structure

[0193] Figure 3Another example of an ED 110 and base stations 170a, 170b, and / or 170c is shown. The ED 110 is used to connect people, objects, machines, etc. The ED 110 can be widely used in various scenarios, such as cellular communication, device-to-device (D2D), vehicle to everything (V2X), peer-to-peer (P2P), machine-to-machine (M2M), machine-type communication (MTC), Internet of things (IOT), virtual reality (VR), augmented reality (AR), mixed reality (MR), metaverse, digital twin, industrial control, autonomous driving, telemedicine, smart grid, smart home, smart office, smart wearable device, smart transportation, smart city, unmanned aerial vehicle, robot, remote sensing, passive sensing, positioning, navigation and tracking, automatic distribution, mobility, etc.

[0194] Each ED 110 represents any suitable end user device for wireless operation, which can include these devices as (or can be referred to as): user equipment / device (UE), wireless transmit / receive unit (WTRU), mobile station, fixed or mobile subscriber unit, cellular telephone, station (STA), machine type communication (MTC) device, personal digital assistant (PDA), smartphone, laptop, computer, tablet, wireless sensor, consumer electronics, smartbook, vehicle, car, truck, bus, train, or IoT device, watch, head-mounted device, glasses, etc. wearable device, industrial device, or means of the above devices (e.g., a communication module, modem, or chip), etc. Future generations of ED 110 can be referred to using other terms. Each base station 170a and 170b is a T-TRP, which will be referred to as T-TRP 170 hereinafter. Also in Figure 3As shown, NT-TRP will be referred to as NT-TRP 172 below. Each ED 110 connected to T-TRP 170 and / or NT-TRP 172 can be dynamically or semi-statically turned on (i.e., established, activated, or enabled), turned off (i.e., released, deactivated, or disabled), and / or configured in response to one or more of connectivity availability and connectivity necessity.

[0195] ED 110 includes a transmitter 201 and a receiver 203 coupled to one or more antennas 204. Only one antenna 204 is shown in the figure. One, some, or all of the antennas 204 may also be panels. The transmitter 201 and receiver 203 may, for example, be integrated as a transceiver. The transceiver is used to modulate data or other content for transmission through at least one antenna 204 or a network interface controller (NIC). The transceiver is also used to demodulate data or other content received through at least one antenna 204. Each transceiver includes any suitable structure for generating signals for wireless or wired transmission and / or for processing signals received wirelessly or wiredly. Each antenna 204 includes any suitable structure for transmitting and / or receiving wireless or wired signals.

[0196] ED 110 includes at least one memory 208. Memory 208 stores instructions and data used, generated, or acquired by ED 110. For example, memory 208 may store software instructions or modules for implementing some or all of the functions and / or embodiments described herein and executed by one or more processing units (e.g., processor 210). Each memory 208 includes any suitable volatile and / or non-volatile storage and retrieval device. Any suitable type of memory can be used, such as random access memory (RAM), read-only memory (ROM), hard disk, optical disk, subscriber identity module (SIM) card, memory stick, secure digital (SD) memory card, processor cache, etc.

[0197] ED 110 may also include one or more input / output devices (not shown) or interfaces (e.g., connected to...). Figure 1 (Wired interface to the Internet 150). Input / output devices support interaction with the user or other devices in the network. Each input / output device includes any suitable (e.g., by operation) structure for providing or receiving information from the user, such as a speaker, microphone, keypad, keyboard, display, or touchscreen, including network interface communication.

[0198] ED 110 includes a processor 210 for performing operations including operations related to preparing uplink transmissions to be sent to NT-TRPs 172 and / or T-TRPs 170, operations related to processing downlink transmissions received from NT-TRPs 172 and / or T-TRPs 170, and operations related to processing sidelink transmissions to and from another ED 110. The processing operations related to preparing uplink transmissions to be sent can include operations such as encoding, modulating, transmit beamforming, and generating symbols for transmission. The processing operations related to processing downlink transmissions can include operations such as receive beamforming, demodulating received symbols, and decoding received symbols. According to embodiments, receiver 203 can receive a downlink transmission (possibly using receive beamforming), and processor 210 can extract signaling (e.g., by detecting and / or decoding the signaling) from the downlink transmission. For example, the signaling can be a reference signal transmitted by NT-TRPs 172 and / or T-TRPs 170. In some embodiments, processor 210 implements transmit beamforming and / or receive beamforming according to an indication of a beam direction, e.g., beam angle information (BAI), received from T-TRPs 170. In some embodiments, processor 210 can perform operations related to network access (e.g., initial access) and / or downlink synchronization, e.g., operations related to detecting synchronization sequences, decoding and acquiring system information, etc. In some embodiments, processor 210 can perform channel estimation using a reference signal received from NT-TRPs 172 and / or T-TRPs 170, for example.

[0199] Although not shown, processor 210 can form part of transmitter 201 and / or part of receiver 203. Although not shown, memory 208 can form part of processor 210.

[0200] Processor 210, processing components in transmitter 201, and processing components in receiver 203 can be respectively implemented by the same or different one or more processors for executing instructions stored in memory (e.g., memory 208). Alternatively, some or all of processor 210, processing components in transmitter 201, and processing components in receiver 203 can be implemented using a special-purpose circuitry, such as an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a graphical processing unit (GPU), a Central Processing Unit (CPU), or a special-purpose computer.

[0201] In some implementations, the ED 110 can be a communication module, a modem, a chip or a chipset, or the like device (may also be referred to as a component), including the at least one processor 210, an interface, or at least one pin. In such a scenario, the transmitter 201 and the receiver 203 can be replaced by the interface or the at least one pin, where the interface or the at least one pin is used to connect the device (e.g., a chip) and another device (e.g., a chip, a memory, or a bus). Thus, sending information to the NT-TRP 172 and / or the T-TRP 170 and / or another ED 110 can be referred to as sending information to the interface or the at least one pin, or as sending information to the NT-TRP 172 and / or the T-TRP 170 and / or another ED 110 through the interface or the at least one pin, and receiving information from the NT-TRP 172 and / or the T-TRP 170 and / or another ED 110 can be referred to as receiving information from the interface or the at least one pin, or as receiving information from the NT-TRP 172 and / or the T-TRP 170 and / or another ED 110 through the interface or the at least one pin. The information can include control signaling and / or data.

[0202] In some implementations, the T-TRP 170 can be referred to by other names, such as a base station, a base transceiver station (BTS), a wireless base station, a network node, a network device, a network-side device, a transmission-reception node, a NodeB, an evolved NodeB (eNodeB or eNB), a Home eNodeB, a Generation NodeB (gNB), a transmission point (TP), a site controller, an access point (AP), a wireless router, a relay, a remote radio head, a ground node, a ground network device, a ground base station, a base band unit (BBU), a remote radio unit (RRU), an active antenna unit (AAU), a remote radio head (RRH), a central unit (CU), a distributed unit (DU), a positioning node, or the like. The T-TRP 170 can be a macro base station, a pico base station, a relay node, a host node, or the like, or a combination thereof. The T-TRP 170 can refer to a device as described above, or to a device (e.g., a communication module, a modem, or a chip) in the device as described above.

[0203] In some embodiments, various portions of T-TRP 170 can be distributed. For example, some of the modules in T-TRP 170 can be remote from the device that houses the antennas 256 of T-TRP 170, and can be coupled to the device that houses the antennas 256 by a communication link (not shown), sometimes referred to as front-haul, such as common public radio interface (CPRI). Thus, in some embodiments, the term "T-TRP 170" can also refer to modules that perform the processing operations of ED 110 position determination, resource allocation (scheduling), message generation and encoding / decoding, etc. on the network side, which are not necessarily part of the device that houses the antennas 256 of T-TRP 170. These modules can also be coupled to other T-TRPs. In some embodiments, T-TRP 170 can actually be multiple T-TRPs that work together, e.g., by using coordinated multipoint transmission, to serve ED 110.

[0204] The T-TRP 170 includes at least one transmitter 252 and at least one receiver 254 coupled to one or more antennas 256. Only one antenna 256 is shown in the figure. One, some or all of the antennas 256 can also be panels. The transmitter 252 and receiver 254 can be integrated as a transceiver. The T-TRP 170 also includes a processor 260 for performing various operations, including operations related to preparing downlink transmissions to the ED 110, processing uplink transmissions received from the ED 110, preparing backhaul transmissions to the NT-TRP 172, and processing transmissions received from the NT-TRP 172 over the backhaul. Processing operations related to preparing a downlink transmission or a backhaul transmission can include operations such as encoding, modulation, precoding (e.g., multiple input multiple output (MIMO) precoding), transmit beamforming, and generating symbols for transmission. Processing operations related to processing received transmissions in the uplink or over the backhaul can include operations such as receive beamforming, demodulating received symbols, and decoding received symbols. The processor 260 can also perform operations related to network access (e.g., initial access) and / or downlink synchronization, such as generating content of a synchronization signal block (SSB), generating system information, etc. In some embodiments, the processor 260 also generates an indication of a beam direction, e.g., a BAI, which can be scheduled for transmission by the scheduler 253. The processor 260 performs other network-side processing operations described herein, e.g., determining a location of the ED 110, determining a location at which to deploy the NT-TRP 172, etc. In some embodiments, the processor 260 can generate signaling, e.g., to configure one or more parameters of the ED 110 and / or one or more parameters of the NT-TRP 172. Any signaling generated by the processor 260 is transmitted by the transmitter 252. It should be noted that “signaling” used herein can also be referred to as control signaling. Dynamic signaling can be transmitted in a control channel such as a physical downlink control channel (PDCCH), and static or semi-static higher layer signaling can be included in packets transmitted in a data channel such as a physical downlink shared channel (PDSCH).

[0205] The scheduler 253 can be coupled to the processor 260. The scheduler 253 can be included within the T-TRP 170 or can operate separately from the T-TRP 170. The scheduler 253 can schedule uplink, downlink, and / or backhaul transmissions, including issuing scheduling grants and / or configuring grant-free (“configured grant”) resources. The T-TRP 170 also includes memory 258 for storing information and data. The memory 258 stores instructions and data used, generated, or collected by the T-TRP 170. For example, the memory 258 can store software

[0206] Although not shown, the processor 260 can form part of the transmitter 252 and / or part of the receiver 254. Further, although not shown, the processor 260 can implement the scheduler 253. Although not shown, the memory 258 can form part of the processor 260.

[0207] The processor 260, the scheduler 253, processing components in the transmitter 252, and processing components in the receiver 254 can be respectively implemented by the same or different one or more processors that are used to execute instructions stored in a memory (e.g., the memory 258). Alternatively, some or all of the processor 260, the scheduler 253, processing components in the transmitter 252, and processing components in the receiver 254 can be implemented using FPGA, GPU, CPU, or ASIC, etc., special circuitry.

[0208] When the T-TRP 170 is an apparatus (also referred to as a component) in a device, a modem, a chip, or a chipset, etc., it includes at least one processor, an interface, or at least one pin. In this scenario, the transmitter 252 and the receiver 254 can be replaced by the interface or at least one pin, which is used to connect the apparatus (e.g., a chip) and another apparatus (e.g., a chip, a memory, or a bus). Thus, transmitting information to the NT-TRP 172 and / or the T-TRP 170 and / or the ED 110 can be referred to as transmitting information to the interface or at least one pin, and receiving information from the NT-TRP 172 and / or the T-TRP 170 and / or the ED 110 can be referred to as receiving information from the interface or at least one pin. The information can include control signaling and / or data.

[0209] Notably, while the NT-TRP 172 is exemplified as a drone, the NT-TRP 172 can be implemented using any suitable non-ground form, such as a high-altitude platform, a satellite, a high-altitude platform that is an international mobile telecommunications base station and an unmanned flight vehicle, which are discussed below. Moreover, in some implementations, the NT-TRP 172 can go by other names, such as a non-ground node, a non-ground network device, or a non-ground base station. The NT-TRP 172 includes a transmitter 272 and a receiver 274 coupled to one or more antennas 280. Only one antenna 280 is shown in the figure. One, some or all of the antennas can also be panels. The transmitter 272 and receiver 274 can be integrated as a transceiver. The NT-TRP 172 also includes a processor 276 for performing various operations, including operations related to preparing downlink transmissions to the ED 110, processing uplink transmissions received from the ED 110, preparing backhaul transmissions to the T-TRP 170, and processing transmissions received from the T-TRP 170 over the backhaul. Processing operations related to preparing a downlink transmission or a backhaul transmission can include operations such as encoding, modulating, precoding (e.g., MIMO precoding), transmit beamforming, and generating symbols for transmission. Processing operations related to processing received transmissions in the uplink or over the backhaul can include operations such as receive beamforming, demodulating received symbols, and decoding received symbols. In some embodiments, the processor 276 implements transmit beamforming and / or receive beamforming in accordance with beam direction information (e.g., BAI) received from the T-TRP 170. In some embodiments, the processor 276 can generate signaling, such as for configuring one or more parameters of the ED 110. In some embodiments, the NT-TRP 172 implements physical layer processing, but not higher layer functions such as functions of a medium access control (MAC) layer or a radio link control (RLC) layer. Since this is just one example, the NT-TRP 172 can generally implement higher layer functions in addition to physical layer processing.

[0210] The NT-TRP 172 also includes a memory 278 for storing information and data. Although not shown, the processor 276 can form part of the transmitter 272 and / or part of the receiver 274. Although not shown, the memory 278 can form part of the processor 276.

[0211] The processor 276, processing components in the transmitter 272, and processing components in the receiver 274 can be implemented by the same or different one or more processors that are used to execute instructions stored in a memory (e.g., the memory 278). Alternatively, some or all of the processor 276, processing components in the transmitter 272, and processing components in the receiver 274 can be implemented using a programmed FPGA, GPU, CPU, or ASIC, among other specialized circuits. In some embodiments, the NT-TRP 172 can actually be multiple NT-TRPs that work together, e.g., through coordinated multipoint transmission, to serve the ED 110.

[0212] When the NT-TRP 172 is a device (e.g., a communication module, modem, chip, or chip set) in an apparatus, at least one processor, interface, or at least one pin is included. In this scenario, the transmitter 272 and the receiver 257 can be replaced by an interface or at least one pin that is used to connect the device (e.g., a chip) and another device (e.g., a chip, memory, or bus). Thus, transmitting information to the T-TRP 170 and / or another NT-TRP 172 and / or the ED 110 can be referred to as transmitting information to the interface or at least one pin, and receiving information from the T-TRP 170 and / or another NT-TRP 172 and / or the ED 110 can be referred to as receiving information from the interface or at least one pin. The information can include control signaling and / or data.

[0213] It should be noted that “TRP” as used herein can refer to a T-TRP or an NT-TRP. A T-TRP can also be referred to as a terrestrial network TRP (TN TRP), and an NT-TRP can also be referred to as a non-terrestrial network TRP (NTN TRP).

[0214] The T-TRP 170, the NT-TRP 172, and / or the ED 110 can include other components, which for brevity, have been omitted.

[0215] Any or all of the EDs 110 and the BSs 170 can be sensing nodes in the system 100. A sensing node is a network entity that senses by transmitting and receiving sensing signals. Some sensing nodes are communication devices that communicate and sense simultaneously. However, it is also possible for some sensing nodes to not communicate and to be dedicated to sensing. The sensing agent 174 is an example of a sensing node that is dedicated to sensing. Unlike the EDs 110 and the BSs 170, the sensing agent 174 does not transmit or receive communication signals. However, the sensing agent 174 can communicate configuration information, sensing information, signaling information, or other information within the communication system 100. The sensing agent 174 can communicate with the core network 130 to communicate information with the rest of the communication system 100. As an example, the sensing agent 174 can determine the location of the ED 110a and transmit that information to the base station 170a through the core network 130. Although only one sensing agent 174 is shown in Figure 2 Although only one sensing agent 174 is shown in

[0216] A sensing node can combine sensing-based techniques with reference signal-based techniques to enhance UE pose determination. This type of sensing node can also be referred to as a sensing management function (SMF). In some networks, the SMF can also be referred to as a location management function (LMF). The SMF can be implemented as a physically separate entity located at the core network 130 that connects with multiple BSs 170. In other aspects of the application, the SMF can be implemented as a logical entity co-located within a BS 170 through logic executed by the processor 260.

[0217] Although Figure 3 Although not shown in

[0218] Basic module structure

[0219] One or more steps of the methods provided herein can be performed by the corresponding units or modules shown in Figure 4 Figure 4 ​Units or modules in the ED 110, T-TRP 170, NT-TRP 172, or GPT device 180, etc. are shown. For example, a signal can be transmitted by a transmitting unit or module. A signal can be received by a receiving unit or module. A signal can be processed by a processing unit or module. Other steps can be performed by an artificial intelligence (AI) module or a machine learning (ML) module. The corresponding units or modules can be implemented using hardware, one or more components or devices executing software, or a combination thereof. For example, one or more of these units or modules can be an integrated circuit, such as a programmed FPGA, GPU, CPU, or ASIC. It will be understood that where modules are implemented by software for execution by a processor, such modules can be retrieved, processed individually or together in a single or multiple instances as needed by the processor, and the modules themselves can include instructions for further deployment and instantiation, in conjunction with Figure 3 The mentioned transmitter can be a specific implementation of a transmitting module. In conjunction with Figure 3 The mentioned receiver can be a specific implementation of a receiving module. In conjunction with Figure 3 The mentioned processor can be a specific implementation of a processing module.

[0220] Other details about the ED 110, T-TRP 170, NT-TRP 172, and GPT device 180 are known to those skilled in the art. Therefore, these details are omitted here.

[0221] Exemplary concepts of some terms

[0222] Message: Payload in a natural language such as English, French, or Chinese.

[0223] Query message: Query statement in a natural language.

[0224] Perception message: Description of observations or perception data in a natural language.

[0225] Semantics: Scalar vector, matrix, tensor embedding a message.

[0226] Query semantics: Semantics embedding a query message.

[0227] Perception semantics: Semantics embedding a perception message.

[0228] Token: Scalar vector encoded from semantics.

[0229] Query token: Token encoded from query semantics.

[0230] Perception token: Token encoded from perception semantics.

[0231] GPT device: a device running on one or more generative AI models, generating one or more query messages according to one or more perception messages.

[0232] Central device: a device as a BS, connecting multiple terminal devices through wireless access in DL and UL, and connecting to a core network through a backbone network.

[0233] Perception device: a device as a terminal, connected to one or more BSs, equipped with a perception gadget to measure nearby data of interest.

[0234] The above describes possible scenarios or general descriptions of examples of the present disclosure, and the motivation and technical concepts of the present disclosure will be described hereinafter.

[0235] Perception functions will be integrated into the 6th generation (6G) system. A large number of perception user equipment (UE) or perception devices will be densely deployed in cities, factories, farms, etc. In addition to mobile phones, perception devices will also become an important type of UE or device announcing the advent of the IoT era.

[0236] Like Internet search engines, 6G will launch a corresponding Internet of Things (IoT) search engine in the real physical world. In fact, billions of IoT-based applications such as self-driving cars, automated factories, smart cities, and autonomous farms will largely rely on efficient and real-time search engines in the physical world.

[0237] In recent years, artificial intelligence (AI) has conquered various fields of knowledge and cognitive fields. Some AI is exploring the frontier knowledge of chemistry, games, mathematics, genetic engineering, etc., while other AI is providing a human-level question and answer platform in the digital world. The field that AI has not yet conquered is the real-time physical world. AI of the physical world can be built on ubiquitous IoT connections brought by 6G, in which AI technology will penetrate into all aspects of society and life.

[0238] More challenging than Internet search engines, real-world search engines must search the physical world in real time over large physical areas and handle multiple types of data and information (some of which can be novel, some of which can not yet have been invented). In addition, green technology, low energy consumption, and low emissions are also proposed as key features of 6G. Sensing devices can be powered by batteries and / or fully powered by solar and wind energy. It is costly and unrealistic to require all sensing devices at a large scale to feedback what they sense at the same time. On one hand, frequent sensing and transmission consume a large amount of energy of the sensing devices, shortening their battery life; on the other hand, such a high density of IoT deployment can block the uplink channel, especially when the uplink (UL) bandwidth is more expensive than the downlink (DL) bandwidth.

[0239] In some implementations, the sensing devices can be UEs, cellphones, or earpieces, where it is assumed that any two sensing devices are independent of each other; thus, the wireless system associated with the sensing devices can schedule the sensing devices individually; the sensing data measured by the sensing devices can be application-level payloads of the wireless system and protocol.

[0240] The above scheme of scheduling the sensing devices is low in terms of wireless bandwidth and energy consumption. For example, the sensing devices blindly send their sensing data to the center device all the time, regardless of whether the sensing data is needed.

[0241] From a higher level perspective, it is better to wake up the sensing devices for measurement and transmission only when their sensing data will serve one or more targets; for example, when a self-driving car or a generative pre-trained transformer (GPT) device can request information about obstacles in movement near itself, it is futile to send irrelevant information to the self-driving car all the time, or to send all the moving obstacles near the car to the car when the car is parked at the roadside.

[0242] To avoid any possibility of missing information, the resources in the wireless system in the above implementations can be over-scheduled.

[0243] The basic idea of the present disclosure can be illustrated as follows. When a semantic token is received, not all sensing devices feedback what they are sensing, only the sensing devices whose sensing data have sufficient relevance to the semantic token (i.e., the sensing data match the semantic token) respond and send their sensing data. For example, a center device (or called BS) can broadcast a semantic token of a query or other task, only the sensing devices (or called UEs) with the corresponding results will feedback the semantic results (i.e., the sensing results below).

[0244] The technical concept of the present disclosure is briefly described above, and specific examples of the present disclosure will be described in detail below.

[0245] The present disclosure provides a communication method, such as Figure 5 As shown, the method can be implemented by a perception device, and can include the following steps:

[0246] Step 502: The perception device receives semantic tokens and tokenization configuration from the center device.

[0247] The semantic tokens are used to retrieve relevant data from the perception device, for example, the semantic tokens can be semantic tokens for a query or other tasks, that is, the semantic tokens can be query tokens or tokens for other tasks. The tokenization configuration can be used for tokenization of perception data, which can be sent from the center device to the perception device, or predefined for the perception device. The specific transmission method of semantic tokens and tokenization configuration will be described later.

[0248] Step 504: The perception device determines whether the perception data matches the semantic tokens, and the tokenization configuration is used to obtain perception tokens from the perception data.

[0249] The perception device is responsible for measuring and / or collecting local physical world data. It can be a perception UE, a perception device, an IoT device, a UE, a mobile phone, a handset, or other devices. The perception device can be equipped with a perception gadget or component for measuring local physical world data, which can be referred to as perception data. In addition, the perception device encodes and sends the perception data to the center device.

[0250] According to the matching score between the perception data and the semantic tokens, or according to the distance between the perception data and the semantic tokens, etc., it can be determined whether the perception data matches the semantic tokens. The specific implementation details will be described later.

[0251] Step 506: When the perception data matches the semantic tokens, the perception device sends a perception result to the center device, and the perception result indicates the perception data.

[0252] The center device can be a base station (BS), such as a gNB or an eNB, etc., or the center device can be an access point (AP). The perception result is related to the perception data obtained by the perception device. If the perception data matches the semantic tokens, the perception device will respond with the perception result. If the perception data does not match the semantic tokens, the perception device will not respond. The specific content of the perception result and the details of the transmission method will be described in detail later.

[0253] The semantic token from the central device wakes up the perception device to take measurement and send the perception result when the perception data matches the semantic token, i.e. the perception device initiates the calculation of the matching judgment in response to the semantic token, and in other cases, the calculation of the matching judgment can not be initiated, thereby reducing the energy consumption of the perception device. In addition, since the form of the semantic token can provide more accurate true intention, save signaling overhead, thereby improving the accuracy of the matching result between the perception data and the semantic token and saving signaling overhead. The tokenization configuration can be used for tokenization of the perception data, thereby ensuring that the matching object adopts a common token form, so that the perception data and the semantic token can be easily compared with each other. And the sending of the perception result is triggered when the perception data matches the semantic token, i.e. the perception result is not always sent, thereby saving transmission resources; in addition, the perception result sent meets the requirements of the semantic token, thereby filtering irrelevant information, and the data sent is required by the central device, thereby ensuring response accuracy.

[0254] In a possible implementation, the perception device can receive the semantic token and the tokenization configuration broadcast or multicast by the central device. It should be noted that in some cases, some perception devices can actively send their perception results without receiving any semantic token from the central device. The perception devices that actively send the perception results can respond to some emergency queries, such as fire alarms or car accidents. In a sense, some queries have been predefined and configured into the system by default. The received semantic token and tokenization configuration are broadcast or multicast by the central device, and the semantic token and tokenization configuration can be sent to multiple perception devices, thereby improving transmission efficiency; through the same tokenization configuration, the perception data of multiple perception devices can be tokenized, so that the perception devices can determine whether to send their perception results under the same standard.

[0255] The semantic token is target-oriented or task-oriented, and can trigger the perception device to complete one or more tasks and collect and send data of one or more modalities. It should be noted that the task and the mode can be independent or associated with each other. For example, the central device (or referred to as BS) can broadcast or multicast the semantic token for a single task, a single mode, multiple modes, multiple tasks. In this case, the perception device can feed back data of one or more modalities in response to a task, and the perception data of one mode can be the requested data corresponding to one or more tasks. Some implementation modes are described below.

[0256] In a possible implementation, the semantic token includes semantic tokens of one task or semantic tokens of one modality. Specifically, the perception device can receive an indication of a task identity or an indication of a modality identity. The indication of the task identity and the indication of the modality identity can be certain field(s), which is not limited herein. The task identity is used to distinguish a certain task, and the modality identity is used to distinguish a certain modality. The perception device can serve one task or one modality, that is, the perception device is dedicated to one task or one modality, and thus the response efficiency of the perception device can be improved.

[0257] In a possible implementation, the semantic token includes a plurality of semantic tokens of a plurality of tasks and a plurality of task identities respectively associated with the plurality of semantic tokens. The perception device can serve a plurality of tasks at the same time, and thus the resource utilization of the perception device is improved.

[0258] In a possible implementation, the semantic token includes a plurality of semantic tokens of a plurality of modalities and a plurality of modality identities respectively associated with the plurality of semantic tokens. The perception device can serve a plurality of modalities at the same time, that is, the perception device can perceive and feedback more than one type of data, which enriches the diversity of data used by the perception device for response.

[0259] In a possible implementation, the semantic token includes a first plurality of semantic tokens of a plurality of tasks, a second plurality of semantic tokens of a plurality of modalities, a plurality of task identities respectively associated with the first plurality of semantic tokens, and a plurality of modality identities respectively associated with the second plurality of semantic tokens. The perception device can serve a plurality of tasks and a plurality of modalities at the same time, and thus not only the resource utilization of the perception device is improved, but also the diversity of data used by the perception device for response is enriched.

[0260] The possible implementations of the information included in the semantic token are described above, and the transmission of the semantic token will be described below.

[0261] In a possible implementation, the perception device can receive, from the central device, a sequence of a predefined length and a tokenization configuration. The semantic token can be encoded as the sequence of the predefined length, and thus the transmission of the semantic token can be integrated into the physical layer of the wireless communication system for the purpose of short latency. Specifically, the sequence of the predefined length can be applied or transmitted in various forms, for example, a pilot code, a mask code, an interleaving sequence, and the like. In another possible implementation, the perception device can receive, from the central device, a payload with a fixed modulation and coding scheme (MCS) and a tokenization configuration. The semantic token can be encoded as the payload with the fixed MCS, and thus the semantic token can be directly encoded with a specific code, and the transmission of the semantic token can be integrated into the physical layer of the wireless communication system for the purpose of short latency.

[0262] As to the specific content of the perception result, in a possible implementation, the perception result comprises one of the following: raw perception data; perception semantics obtained from the raw perception data; semi-raw perception data and perception semantics obtained from the raw perception data; the raw perception data and a matching score between the perception data and a semantic token; the raw perception data and a distance between the perception data and a semantic token; the perception semantics obtained from the raw perception data and a matching score between the perception data and a semantic token; the perception semantics obtained from the raw perception data and a distance between the perception data and a semantic token; semi-raw perception data, the perception semantics obtained from the raw perception data, and a matching score between the perception data and a semantic token; semi-raw perception data, the perception semantics obtained from the raw perception data, and a distance between the perception data and a semantic token.

[0263] The perception result can adopt various forms related to the perception data, which provides more flexibility and can meet different requirements.

[0264] In a possible implementation, the perception result further comprises a task identifier or a modality identifier. The task identifier is used to distinguish a certain task, and the modality identifier is used to distinguish a certain modality. Since the task identifier or the modality identifier is included in the perception result, the center device can easily identify which task or which modality is carried in a certain perception result.

[0265] In a possible implementation, the perception result further comprises an identifier of a semantic token with a matching score greater than or equal to a first threshold or a distance less than a second threshold. In another possible implementation, the perception result further comprises multiple identifiers of multiple semantic tokens with a matching score greater than or equal to a first threshold or a distance less than a second threshold. The identifier of the semantic token is used to distinguish the semantic token, and the semantic token with a matching score greater than or equal to a first threshold or a distance less than a second threshold indicates that the perception device has the perception data corresponding to the semantic token. Since the identifier is included in the perception result, the center device can easily identify which semantic token is carried in a certain perception result.

[0266] As to the sending manner of the perception result, in a possible implementation, the perception device can initiate a random access, a state report (SR), or a buffer state report (BSR) procedure to send the perception result to the center device.

[0267] In one example, the BS (or the aforementioned central device) can broadcast or multicast semantic terms (or query keys), where term k can be a compressed / privacy-protected version of the original query q. For example, for n queries, the message format of the query key can be: {k1, k2, ..., k} n}, n>=1. For example, for m tasks, the message format can be: {{{t1,{k} 1,1 , .., k 1,n1}},.., {t m , {k m,1 , .., k 1m,nm}}}, each task t i Having n i There are one query, m>=1, n i >=1. Each term (or query key) k can be of fixed length, selected from a given set of lengths {LEN_1, LEN_2, …LEN_P}, which facilitates UE detection. The message can be carried in the SSB for broadcast, or in a multicast message for a group of UEs (or the aforementioned sensing devices), or even in a multicast message dedicated to a single UE.

[0268] The UE receives / detects semantic tokens, then obtains their semantic observation value o (or embedding vector), obtains their local semantic token c based on o, and combines c with {k1, k2, ..., k}. n} or {k i,1 , k i,2 , .., k i, ni} compare. If c is compared with any k j If a match is found, the UE can determine whether to respond; otherwise, the UE can determine whether to not to respond. The BS can be configured to calculate the semantic unit c based on the semantic observation o, and can be configured to calculate the distance between two units: d(c, k) j The threshold t used for the response can be configured, i.e., if d(c, k) j If t < 0, the UE will respond. If the UE determines that it needs to respond, it will generate a semantic response, which includes a semantic observation o or a local semantic unit c, represented by a vector of length N, or by N... j ×M j Matrix representation. The semantic response also includes the task identifier, i.e., i, matching query q. jidentifiers, i.e., j. If there are multiple matching queries, multiple identifiers can be included. In addition, the semantic response or local semantic tokens can also be compressed. If there are multiple observations or multiple semantic tokens, the semantic response can include multiple observations or multiple semantic tokens. The UE initiates the following procedures to send the semantic response to the BS: random access, state report (SR), or buffer state report (BSR), etc.

[0269] In a possible implementation, the perception device can obtain perception data, translate the perception data into perception semantics according to the semantic configuration, tokenize the perception semantics into perception tokens according to the tokenization configuration, and then determine whether the perception tokens match the semantic tokens. It should be noted that the translation operation in the present disclosure refers to semantic processing, and the translation operation can be replaced by embedding operation, conversion operation, transform operation, etc. For example, translating the perception data into perception semantics can be replaced by embedding the perception data into the perception semantics, converting the perception data into the perception semantics, transforming the perception data into the perception semantics, etc. The specific method of the translation operation, the embedding operation, the conversion operation, and the transform operation is not limited herein. For example, the transform operation can be implemented by existing methods. In the case that the perception data is in the form of natural language, the perception data can be processed into perception tokens, and then the comparison between the perception tokens and the semantic tokens is implemented. Since the form of the token can provide more accurate real intention, signaling overhead is saved, thereby improving the accuracy of the comparison result and saving the signaling overhead. In addition, tokenization can be used to prevent the perception device from recovering the complete query message from the semantic tokens. Tokenization can provide a certain privacy protection for the query message.

[0270] In a possible implementation, the tokenization configuration includes one of the following: a tokenization model, a tokenization function, a projection matrix, a graph-based or topology-based pruning, and a compression method. There are multiple tokenization configurations to choose from, providing more flexibility, so that different needs can be met. Some implementations of tokenization are listed above, and the specific method of the tokenization implementation is not limited herein. For example, the tokenization model can be implemented by existing methods.

[0271] In a possible implementation, the perception device with the perception function acquires the perception data when receiving the semantic token, and determines whether the perception data matches the semantic token. The specific method of determining whether the perception data matches the semantic token can be: if a matching score between the perception data and the semantic token is greater than or equal to a first threshold, it is determined that the perception data matches the semantic token; otherwise, it is determined that the perception data does not match the semantic token. Alternatively, if a distance between the perception data and the semantic token is less than or equal to a second threshold, it is determined that the perception data matches the semantic token; otherwise, it is determined that the perception data does not match the semantic token. According to the matching score between the perception data and the semantic token, or according to the distance between the perception data and the semantic token, whether the perception data matches the semantic token can be determined, thereby providing an efficient implementation. It should be noted that other indicators such as a correlation score between the perception data and the semantic token, a similarity between the perception data and the semantic token, and the like can also be used to indicate whether the perception data matches the semantic token.

[0272] In a possible implementation, the perception device can receive at least one of a scoring function for determining a matching score between the perception data and the semantic token, a first threshold, a function for determining a distance between the perception data and the semantic token, or a second threshold from the center device. The matching score between the perception data and the semantic token can be calculated by the scoring function, and the distance between the perception data and the semantic token can be calculated by the corresponding function, thereby providing an efficient implementation. At least one of the scoring function for determining the matching score, the first threshold, the function for determining the distance, and the second threshold can be sent from the center device to the perception device, which provides more flexibility and can meet different requirements.

[0273] In another possible implementation, the perception device can receive the semantic token, the tokenization configuration, and at least one of the scoring function for determining the matching score between the perception data and the semantic token, the first threshold, the function for determining the distance between the perception data and the semantic token, or the second threshold from the center device. At least one of the scoring function for determining the matching score, the first threshold, the function for determining the distance, and the second threshold can be sent from the center device to the perception device together with the semantic token and the tokenization configuration, which provides more flexibility and can meet different requirements.

[0274] It should be noted that at least one of the scoring function used to determine the matching score, the first threshold, the function used to determine the distance, and the second threshold can be predefined for the perception device, can be sent to the perception device by the center device together with the semantic word and tokenization configuration, or can be sent to the perception device before the center device sends the semantic word and tokenization configuration. The scoring function can take the form of an inner product or Euclidean distance; the function used to determine the first distance can take the form of an inner product, a cross-correlation matrix, or a cross-entropy function, and the above implementation manners are only illustrative but not limiting, and it should be noted that there can be other methods to implement the above functions. This provides more flexibility, so that different needs can be met. It should be noted that the threshold in the present disclosure can be predefined or selected according to actual needs.

[0275] In a possible implementation manner, the semantic word is carried in high layer signaling or physical layer signaling or a combination of high layer signaling and physical signaling, the high layer signaling includes one of radio resource control (RRC) signaling and medium access control (MAC) layer signaling, and the physical layer signaling includes downlink control information (DCI). The semantic word can be sent from the center device to the perception device through specific signaling, which guarantees high security and high efficiency of data transmission.

[0276] The above describes the communication method of the present disclosure from the perspective of the perception device. The communication method of the present disclosure will be described from the perspective of the center device as follows. Figure 5 The method can include: Figure 6 Step 602: The center device sends a semantic word and a tokenization configuration to the perception device.

[0277] Step 604: The center device receives a perception result from the perception device, wherein the perception data of the perception device matches the semantic word, the tokenization configuration is used to obtain a perception word from the perception data, and the perception result indicates the perception data.

[0278]

[0279] ​As to the description of step 602 and step 604, reference can be made to the description of step 502 and step 506, which will not be repeated here. The center device sends the semantic token and the tokenization configuration to the perception device, the form of the semantic token can provide more accurate real intention and save signaling overhead, and the perception device can use the tokenization configuration to tokenize the perception data. The center device receives the perception result in the case that the perception data matches the semantic token, the received perception result meets the requirement of the semantic token, that is, irrelevant information is filtered, and the received data is what the center device needs, thereby ensuring the accuracy of data transmission.

[0280] In a possible implementation, the center device can broadcast or multicast the semantic token and the tokenization configuration to multiple perception devices. The center device broadcasts or multicasts the semantic token and the tokenization configuration to multiple perception devices, thereby improving transmission efficiency; through the same tokenization configuration, the perception data of multiple perception devices can be tokenized, so that the perception devices can judge whether to send their perception results under the same standard.

[0281] Next, the content included in the semantic token is briefly introduced, and the following details of the possible implementation can be the same as or similar to the description on the perception device side, which will not be repeated here.

[0282] In a possible implementation, the semantic token includes a semantic token of one task or a semantic token of one modality. Specifically, the center device can send an indication of a task identifier or an indication of a modality identifier.

[0283] In another possible implementation, the semantic token includes multiple semantic tokens of multiple tasks and multiple task identifiers respectively associated with the multiple semantic tokens.

[0284] In yet another possible implementation, the semantic token includes multiple semantic tokens of multiple modalities and multiple modality identifiers respectively associated with the multiple semantic tokens.

[0285] In still another possible implementation, the semantic token includes a first plurality of semantic tokens of multiple tasks, a second plurality of semantic tokens of multiple modalities, multiple task identifiers respectively associated with the first plurality of semantic tokens, and multiple modality identifiers respectively associated with the second plurality of semantic tokens.

[0286] Regarding the transmission of the semantic word, in one possible implementation, the central device can encode the semantic word into a predefined length sequence, and send the predefined length sequence and the wordization configuration to the perception device. The semantic word can be encoded into a predefined length sequence, thus the transmission of the semantic word can be integrated into the physical layer of the wireless communication system for the purpose of short latency. Specifically, the predefined length sequence can be applied or sent in various forms, such as a pilot code, a mask code, an interleaved sequence, etc. In another possible implementation, the central device can encode the semantic word into a payload with a fixed modulation and coding scheme (MCS), and send the payload with the fixed MCS and the wordization configuration to the perception device. The semantic word can be encoded into a payload with a fixed MCS, thus the semantic word can be directly encoded with a specific code, and the transmission of the semantic word can be integrated into the physical layer of the wireless communication system for the purpose of short latency.

[0287] Regarding the specific content of the perception result, in one possible implementation, the perception result comprises one of the following: raw perception data; perception semantics obtained from the raw perception data; semi-raw perception data and perception semantics obtained from the raw perception data; raw perception data and a matching score between the perception data and the semantic word; raw perception data and a distance between the perception data and the semantic word; perception semantics obtained from the raw perception data and a matching score between the perception data and the semantic word; perception semantics obtained from the raw perception data and a distance between the perception data and the semantic word; semi-raw perception data, perception semantics obtained from the raw perception data, and a matching score between the perception data and the semantic word; semi-raw perception data, perception semantics obtained from the raw perception data, and a distance between the perception data and the semantic word. The perception result can adopt various forms related to the perception data, which provides more flexibility and thus can meet different requirements.

[0288] In one possible implementation, the perception result further comprises a task identification or a modality identification. The task identification is used to distinguish a certain task, and the modality identification is used to distinguish a certain modality. Since the task identification or the modality identification is included in the perception result, the central device can easily identify which task or which modality is carried in a certain perception result.

[0289] In a possible implementation, the perception result further includes an identification of one semantic token whose matching score is greater than or equal to the first threshold or whose distance is less than the second threshold. In another possible implementation, the perception result further includes a plurality of identifications of a plurality of semantic tokens whose matching scores are greater than or equal to the first threshold or whose distances are less than the second threshold. The identification of the semantic token is used to distinguish the semantic tokens, and the semantic token whose matching score is greater than or equal to the first threshold or whose distance is less than the second threshold indicates that the perception device has the perception data corresponding to the semantic token. Since the identification is included in the perception result, the center device can easily identify which semantic token is carried in a certain perception result.

[0290] In a possible implementation, as shown in FIG. 7, the communication method includes: Figure 7

[0291] Step 702: The center device receives a semantic from a generative pre-trained transformer (GPT) device.

[0292] Step 704: The center device tokenizes the semantic into semantic tokens and sends the semantic tokens and tokenization configurations to the perception device.

[0293] Step 706: The center device receives a perception result from the perception device, the perception data of the perception device matches the semantic tokens, the tokenization configurations are used to obtain the perception tokens from the perception data, and the perception result indicates the perception data.

[0294] Step 708: The center device outputs the perception result to the GPT device.

[0295] The descriptions about steps 704 and 706 can refer to the descriptions above, which are not repeated here. The center device can perform tokenization processing on the semantic received from the GPT device, and then send the semantic tokens to the perception device. It should be noted that the center device can directly forward the semantic from the GPT device to the perception device without tokenization processing, that is, the tokenization processing can be performed by the perception device, which is not limited here. The center device receives the semantic from the GPT device, provides the semantic tokens and the tokenization configurations for the perception device to make decisions, receives the perception result from the perception device, and outputs the perception result to the GPT device. The center device serves as a bridge between the perception device and the GPT device, thereby facilitating smooth communication between the perception device and the GPT device.

[0296] ​In a possible implementation, the center device can determine a lemmatization configuration, and then send the determined lemmatization configuration to the perception device. The lemmatization processing can have different lemmatization configurations, and a suitable lemmatization configuration can correspond to a specific lemmatization, thereby ensuring high efficiency of data processing. In a possible implementation, the center device can determine a length of a semantic lemma. The length of the semantic lemma can be various, and the center device can select a suitable length of the semantic lemma for the semantic lemma. Determining the length of the semantic lemma can facilitate subsequent processing, for example, the length of the perception lemma can be the same as the length of the semantic lemma.

[0297] In a possible implementation, the lemmatization configuration includes one of the following: a lemmatization model, a lemmatization function, a projection matrix, a graph-based or topology-based pruning, and a compression method. There are various lemmatization configurations to choose from, which provides more flexibility and can meet different requirements.

[0298] In a possible implementation, the center device can send, to the perception device, at least one of the following: a scoring function for determining a matching score between the perception data and the semantic lemma, a first threshold, a function for determining a distance between the perception data and the semantic lemma, or a second threshold. The matching score between the perception data and the semantic lemma can be calculated by the scoring function, and the distance between the perception data and the semantic lemma can be calculated by the corresponding function, thereby providing an efficient implementation. The at least one of the scoring function for determining the matching score, the first threshold, the function for determining the distance, or the second threshold can be sent from the center device to the perception device, which provides more flexibility and can meet different requirements.

[0299] In a possible implementation, the center device can send, to the perception device, the semantic lemma, the lemmatization configuration, and at least one of the following: a scoring function for determining a matching score between the perception data and the semantic lemma, a first threshold, a function for determining a distance between the perception data and the semantic lemma, or a second threshold. The at least one of the scoring function for determining the matching score, the first threshold, the function for determining the distance, or the second threshold can be sent from the center device to the perception device together with the semantic lemma and the lemmatization configuration, which provides more flexibility and can meet different requirements.

[0300] It should be noted that at least one of the scoring function for determining the matching score, the first threshold, the function for determining the distance, and the second threshold can be predefined for the perception device, can be sent to the perception device by the center device together with the semantic wordpiece and wordpiece configuration, or can be sent to the perception device before the center device sends the semantic wordpiece and wordpiece configuration. The scoring function can take the form of an inner product or an Euclidean distance; the function for determining the first distance can take the form of an inner product, a cross-correlation matrix, or a cross-entropy function, and the above implementation manners are only illustrative but not limiting, and it should be noted that there can be other methods to implement the above functions. This provides more flexibility, so that different needs can be met. It should be noted that the threshold in the present disclosure can be predefined or selected according to actual needs.

[0301] In a possible implementation manner, the semantic wordpiece is carried in high layer signaling, or physical layer signaling, or a combination of high layer signaling and physical signaling, the high layer signaling including one of radio resource control (RRC) signaling, medium access control (MAC) layer signaling, and the physical layer signaling including downlink control information (DCI). The semantic wordpiece can be sent from the center device to the perception device through specific signaling, which guarantees high security and high efficiency of data transmission.

[0302] In order to more clearly set forth the communication method of the present disclosure, the following takes a communication system including at least one center device, a plurality of distributed perception devices, and at least one GPT device as an example, and the method will be described in more detail through the following example embodiments.

[0303] Example 1

[0304] In the present disclosure, the wireless system is also referred to as a communication system, or a wireless communication system. Here, the wireless system includes a plurality of devices, for example, the plurality of devices include at least one center device, a plurality of distributed perception devices, and at least one GPT device (for example, a GPT device in the center device) Figure 8 ).

[0305] The GPT device is responsible for encoding or decoding the query message and the perception data. Specifically, it generates a query message containing one or more natural language goals for the hub device; the hub device semantizes the query message into a semantic vector (i.e., the above-mentioned semantics), tokenizes the semantic vector into goal semantic tokens (vectors) (i.e., the above-mentioned semantic tokens), and then broadcasts the goal semantic tokens to the perception devices. The perception devices are triggered by receiving the goal semantic tokens, measure their perception data, and convert the perception data into perception semantic tokens (i.e., the above-mentioned perception tokens). The perception devices compare the goal semantic tokens with the perception semantic tokens and score the correlation between the goal semantic tokens and the perception semantic tokens, and only when the correlation score is higher than a threshold, the perception data is sent in the form of a semantic vector. The hub device fuses the perception data into a semantic vector, and outputs the fused data to the GPT device, which will generate the next query message according to the fused input.

[0306] The hub device can be a BS, such as a gNB or an eNB, or the hub device can be an access point (AP).

[0307] The perception device is responsible for measuring and / or collecting local physical world data. It can be a perception UE, a perception device, an IoT device, a UE, a cell phone, a handset, or other devices. The perception device can be equipped with perception gadgets or components to measure the nearby local physical world data as perception data; the perception device encodes the perception data and sends it to the hub device.

[0308] The GPT device can generate a series of query messages and receive fused perception messages from the hub device. In this disclosure, the GPT device can also be referred to as an AI agent device, a robot device, or a smart control device.

[0309] Specifically, the plurality of perception devices here can be grouped or classified according to the type of perception data. As shown in Figure 9 , the first group of perception devices can measure the first type of perception data (e.g., red, green, blue (RGB) images or videos), while the second group of perception devices can measure the second type of perception data (e.g., radio RF point clouds or laser radar point clouds). It should be noted that some perception devices can be divided into more than one group, i.e., some perception devices can measure more than one type of perception data.

[0310] The hub device actively requests or triggers the perception devices to send their recent perception data (as Figure 10 correspondingly, the perception devices will send their perception data.

[0311] The central device can send one or more first query messages to one or some of the sensing devices in one or more DL broadcast, groupcast or unicast channels, which can be in a physical broadcast channel, a shared channel or a dedicated channel.

[0312] Upon receiving the first query message, the sensing device decides whether to send its sensing data. Specifically, the sensing device decodes the first query message, measures its data, and decides whether to send its sensing data, which is referred to as responding to the first query message. If the sensing device decides to respond to the first query message, the sensing device can encode / encapsulate the sensing data into a payload, and then send it to the central device in one or more UL channels, which can be a physical UL shared channel or a dedicated UL channel.

[0313] Upon receiving all the payloads (i.e. the sensing results described above) from the sensing devices that respond to the first query message, the central device of the wireless system can fuse all or part of the payloads into a fused payload. Optionally, the central device can input the fused payload into a GPT device that can process them, and then generate a second query message.

[0314] The central device can send one or more second query messages to one or some of the sensing devices in one or more DL broadcast, groupcast or unicast channels.

[0315] The GPT device sends a query message to the hub device to inform and configure the hub device to schedule the sensing devices to sense at a time, in a manner, with a content, and select which sensing devices, and send their sensing data to the hub device. The GPT device can be implemented / located with the hub device to achieve a shorter latency, or the GPT device can be implemented in a remote data center that the hub device can access through a core network, or the GPT device can be on other connected devices in the same wireless system as the hub device. It should be noted that in this disclosure, the query message (downlink message) from the hub device to the sensing devices can be carried in a high layer signaling such as radio resource control (RRC) signaling or medium access control (MAC) layer signaling. Or, the query message can be carried in a physical layer signaling such as downlink control information (DCI). Or, the query message can be carried in a combination of high layer signaling and physical signaling. The same applies to other downlink messages / data sent from the hub device to the sensing devices. Similarly, in this disclosure, for uplink messages / data, they can be carried in a high layer signaling such as RRC signaling or MAC layer signaling. Or, they can be carried in a physical layer signaling such as uplink control information (UCI). Or, they can be carried in a combination of high layer signaling and physical signaling. It should be noted that the messages in this disclosure can be replaced by information, and the information can be carried in one single message or more than one single message.

[0316] The GPT device can generate a semantic and send the semantic to the hub device. The hub device can lexicalize the semantic into semantic lexical units and send the semantic lexical units and lexicalization configuration to the sensing devices. In response to the semantic lexical units from the hub device, the sensing devices can collect sensing data, translate the sensing data into sensing semantics, and lexicalize the sensing semantics into sensing lexical units according to the lexicalization configuration. Then, the sensing devices can determine whether the sensing lexical units match the semantic lexical units. In the case that the sensing lexical units match the semantic lexical units, the sensing devices can send the sensing results (e.g., the sensing data) to the hub device. It should be noted that the semantic sent by the GPT device to the hub device can be replaced by other forms, and the semantic lexical units sent by the hub device to the sensing devices can also be replaced by other forms such as the query message mentioned above or other forms. Further, the semantic and the semantic lexical units can include a single query message or more than one query message. Similarly, the fused sensing message is a specific form of the fused sensing result, and the fused sensing result can include a single fused sensing message or more than one fused sensing message.

[0317] As Figure 11 illustrated, a wireless system including a central device, a perception device, and a GPT device can form a series of interactions in which the GPT device generates a stream of query messages for the perception device, the perception device collects and feeds back perception data, and the central device fuses the perception data and inputs it to the GPT device.

[0318] In some cases, some perception devices can proactively send their perception data without receiving any query messages from the central device. The perception devices sending perception data can respond to some emergency queries, such as a fire alarm or a car accident. In a sense, some query messages are pre-defined and configured into the system by default.

[0319] Example 2

[0320] The GPT device in embodiment 1 can generate a stream of query messages based on previous perception messages, where the previous perception messages are received and / or fused by the central device. The GPT device can infer one or more generative AI models. The generative AI model or models infer deep neural network outputs one or more query messages. The GPT device generates a stream of query messages, referred to as a “thought chain,” by interacting with a stream of fused perception messages, to which the central device fuses perception data sent by responding perception devices; as Figure 12 illustrated.

[0321] The query messages generated by the GPT device can convey semantic goals, tasks, or purposes. For example, the query message “locate an oncoming pedestrian” explicitly establishes a semantic goal for the perception device, causing the perception device to focus on nearby pedestrians and preventing the perception device from being distracted by other things. Since the query message conveys one or more semantic goals, the query messages sent by the central device to the perception devices can trigger goal-oriented perception tasks at each responding perception device that receives and responds to the query message. It should be noted that a message can convey multiple goals. For example, the message “find a moving pedestrian wearing a white coat” conveys two semantic goals or tasks: a moving pedestrian and a pedestrian wearing a white coat.

[0322] In one implementation, the central device can broadcast a series of query messages, as it can be too costly, or even prohibited, to individually schedule the sensing devices in a wireless system that includes such a high density of sensing devices. Thus, upon receiving a query message, a sensing device can wake up, but has little idea whether its sensing data is sufficiently relevant to the target conveyed by the query message. As such, the sensing device can enable its sensing widget to sense the environment around it into sensing data, and compare the sensing data with the query message. If the sensing device tells that the sensing data is sufficiently relevant to the query message, the sensing device encodes the sensing data and sends the sensing data to the central device (sensing device #1 in Figure 13 ). Otherwise, the sensing can not respond to the query message at all (sensing device #2 in Figure 13 ). In this sense, the wireless system does not schedule individual sensing devices, but a common task in a collection of sensing devices.

[0323] The central device can receive multiple sensing data from some or all of the sensing devices that respond to the query message at the end of a predefined response time interval. The central device can fuse all the sensing data into one sensing message, and input the sensing message to the GPT device, which will generate the next query message according to the sensing message, as shown in Figure 14 .

[0324] As only those sensing devices that respond to the query message will send sensing data, a large amount of wireless resources can be saved compared to one-to-one scheduling algorithms.

[0325] Example 3

[0326] The series of query messages generated by the GPT device in embodiment 2 and broadcast by the central device are in natural language, i.e., human-readable. The GPT device can use a large language model (LLM) to infer the input of the fused sensing message (also in natural language) to generate a new query message. The LLM model can be a “standard” base model such as transformer, or a “customized” model built for a narrow vocabulary and specific scenarios. For example, a customized LLM for handling industrial 4.0 or a customized LLM for handling wireless communication signaling and protocols. The GPT device can change, update, downsize, upsize, replace one or more of its LLMs at any time as needed. It should be noted that broadcasting, multicasting or unicasting can be implemented.

[0327] The query message generated by the GPT device in Example 2 is in natural language. Due to randomness of generation, two different query messages can convey very similar semantic goal(s). For example, “find pedestrian” and “locate person walking” can have the same semantic goal. Thus, the GPT device can semantize the query message into query semantics, referred to as “embedding,” “semantizing,” “encoding,” “natural language to machine translation,” and the like. The GPT device can translate the query message into query semantics, which can include a scalar vector, a matrix, or a tensor. The translation can be implemented by a deep neural network or other classical function. The query semantics can preserve all key semantic goals conveyed by the query message so that the query semantics can be properly translated (desemantized) back to the query message. Optionally, the GPT device can send the query semantics to the hub device instead of the query message, as shown in Figure 15 As shown in Figure 16 The query semantics are invertible, meaning that the query message can be recovered from the query semantics. It should be noted that if all LLMs output in a common natural language (e.g., English), these LLMs are said to be aligned in natural language; as such, no matter what LLM is used, it can be seamlessly connected to the GPT device and work properly in the wireless system.

[0328] In one implementation, the hub device can also tokenizes the query semantics into query tokens. The query tokens are fixed-length semantics but include scalar vectors, which are more convenient for transmission and comparison purposes. The wireless system can pre-specify multiple lengths for the query tokens. Thus, when the hub device tokenizes the query semantics, it can select an appropriate token length according to the size range of the query semantics. The tokenization can be a very strict function to prevent the perception device from recovering the complete query message from the query tokens. The tokenization can provide certain privacy protection for the query message. The tokenization can be implemented by a deep neural network or other classical function, as shown in Figure 17 .

[0329] Optionally, the hub device receives the query semantics from the GPT device, and then the hub device converts the query semantics into fixed-length query tokens; the hub device can broadcast the query tokens of this length to all perception devices; the hub device can save the query semantics in its own memory or storage device for checking the feedback perception data.

[0330] As in embodiment 2, the perception device can compare its perception data with the query message; the perception device is woken up after receiving the query token (with its length or its length indication) so that its perception widget can measure the nearby physical environment as perception data; the perception device can be equipped with one or more LLMs as semanticization models, input the perception data into the semanticization models to output perception semantics; optionally, the perception device can select a suitable length and format of the perception semantics; the perception device can continue to tokenize the perception semantics into perception tokens of the same length as the query tokens the perception device has received; the perception device compares the query message with the perception data or scores the relevance between the query message and the perception data according to the content it has received.

[0331] Alternative #1 (query token vs. perception token) Figure 18 and Figure 19 ): the perception device receives the query token and the scoring function; compares the query token with the perception token and scores the relevance between the query token and the perception token; if the relevance score is greater than or equal to a pre-defined threshold, the perception device will inform that the perception data is sufficiently relevant to the query message from the center device.

[0332] If the relevance score is greater than or equal to a pre-defined threshold, the perception device can send information including the perception data and the relevance score (i.e. the above-mentioned perception result) to the center device ( Figure 20 ). Here are some alternatives for the content in the sent information:

[0333] Alternative #1: raw perception data + relevance score

[0334] Alternative #2: perception semantics + relevance score

[0335] Alternative #3: semi-raw perception data (e.g. exact value or number) + perception semantics + relevance score.

[0336] The perception device can be equipped with one or more semanticization models to generate perception semantics from perception (raw) data, can be equipped with tokenization models to generate perception tokens from perception semantics, and can be configured with scoring functions; unlike the GPT device, the LLMs, tokenization models and scoring functions that the perception device can use are configured by the center device; the center device can configure the public LLMs and / or tokenization models and scoring functions from the beginning or at runtime and inform the perception device.

[0337] Example 4

[0338] The scoring function for scoring the relevance between the query token and the perception token in embodiment 3 can be implemented by a scoring function. The scoring function can be an inner product, or a dot product, Euclidean distance or other scoring function.

[0339] Regardless of which scoring function is used, the central device must explicitly or implicitly inform and configure all sensing devices to use the same scoring function so that the correlation scores in Embodiment 3 from different sensing devices can be compared at the central device; the central device can inform and configure the sensing devices of the common scoring function at the beginning of the DL message, or specify the scoring function or a list of scoring functions in the specification, or indicate the scoring function using query semantics in the DL.

[0340] Example 5

[0341] Multiple sensing devices of one or more types can serve one or more tasks simultaneously; sensing devices can be triggered once in an efficient manner to serve as many tasks as possible.

[0342] The wireless system can include two GPT devices, or one GPT device that can perform two independent tasks; in the following disclosure, two GPT devices are taken as an example. Two GPT devices can be easily extended to one GPT device that performs two independent tasks.

[0343] Although the two GPT devices have independent tasks, the two GPT devices can trigger the same sensing device at the same time; for example, the GPT device of the self-driving car and the GPT device of the traffic light can trigger the same roadside camera sensing device; however, although the same sensing device can be triggered by the two GPT devices at the same time interval, the query message from the first GPT device can be different from the query message from the second GPT device; for example, the GPT device of the self-driving car can broadcast a query message about "obstacles in motion", and the GPT device of the traffic light can broadcast a query message about "vehicle flow density", both of which can be related to some extent but not similar.

[0344] The first GPT device generates a first query semantics to the central device, and the second GPT device generates a second query semantics to the central device. There are two options, as follows.

[0345] Alternative #1: The center device can tokenizes the first query semantics into a first query token, and the second query semantics into a second query token; the center device can use a first tokenization model to tokenize the first query semantics, and a second tokenization model to tokenize the second query semantics, or the center device can use a common tokenization model to tokenize the first query semantics and the second query semantics; then the center device can broadcast the first query token, the length of the first query token, the first scoring function related to the first query token and the first threshold value related to the first scoring function, and the second query token, the length of the second query token, the second scoring function related to the second query token and the second threshold value related to the second scoring function in a multiplexing manner in the DL channel.

[0346] In another example, as shown in FIG. 8A, the first GPT device generates a first query token for the center device, and the second GPT device generates a second query token for the center device. At this time, the center device can directly send the first query token and the second query token to the perception device. The perception device can receive the first query token and the second query token, wake up to make the perception gadget perceive the physical world around itself as perception data. There are two options, as follows. Figure 21

[0347] Alternative 1: The perception device can convert the perception data into a common perception semantics through one or more LLMs; then the perception device can tokenize the perception semantics into a first perception token according to the length of the first query token, and tokenize the perception semantics into a second perception token according to the length of the second query token; wherein the perception device can use a first tokenization model to tokenize the perception semantics into the first perception token, and use a second tokenization model to tokenize the perception semantics into the second perception token ( Figure 22 ), or can use a common tokenization model to tokenize the perception semantics into the first perception token and the second perception token ( Figure 23 ); the perception device can score the relevance between the first query token and the first perception token, and the relevance between the second query token and the second perception token; if the first relevance score is greater than or equal to the first threshold value, the perception device can inform that the perception data provides sufficient relevance to the first query token, and if the second relevance score is greater than or equal to the second threshold value, the perception device can inform that the perception data provides sufficient relevance to the second query token; if it is determined that the first relevance score is high enough, the perception device can send at least one of the perception data, the perception semantics or the first relevance score (Embodiment 3); if it is determined that the second relevance score is high enough, the perception device can send at least one of the perception data, the perception semantics or the second relevance score (Embodiment 3).

[0348] ​Alternative #2: The perception device can convert the perception data into a first perception semantics through one or more LLMs, and convert the same perception data into a second perception semantics through one or more LLMs; the perception device can then tokenizes the first perception semantics into first perception tokens according to the length of the first query tokens, and tokenizes the second perception semantics into second perception tokens according to the length of the second query tokens; wherein the perception device can use a first tokenization model to tokenize the first perception semantics into the first perception tokens, and use a second tokenization model to tokenize the second perception semantics into the second perception tokens, or can use a common tokenization model to tokenize the perception semantics into the first perception tokens and the second perception tokens; the perception device can score the relevance between the first query tokens and the first perception tokens, and the relevance between the second query tokens and the second perception tokens; if the first relevance score is greater than or equal to a first threshold, the perception device can inform that the perception data provides sufficient relevance to the first query tokens, and if the second relevance score is greater than or equal to a second threshold, the perception device can inform that the perception data provides sufficient relevance to the second query tokens; if it is determined that the first relevance score is high enough, the perception device can send at least one of the perception data, the first perception semantics, or the first relevance score (Embodiment 3); if it is determined that the second relevance score is high enough, the perception device can send at least one of the perception data, the second perception semantics, or the second relevance score (Embodiment 3). Figure 24 ), or can use a common tokenization model to tokenize the perception semantics into the first perception tokens and the second perception tokens ( Figure 25 ); the perception device can score the relevance between the first query tokens and the first perception tokens, and the relevance between the second query tokens and the second perception tokens; if the first relevance score is greater than or equal to a first threshold, the perception device can inform that the perception data provides sufficient relevance to the first query tokens, and if the second relevance score is greater than or equal to a second threshold, the perception device can inform that the perception data provides sufficient relevance to the second query tokens; if it is determined that the first relevance score is high enough, the perception device can send at least one of the perception data, the first perception semantics, or the first relevance score (Embodiment 3); if it is determined that the second relevance score is high enough, the perception device can send at least one of the perception data, the second perception semantics, or the second relevance score (Embodiment 3).

[0349] If the center device receives multiple first perception semantics plus first relevance scores and multiple second perception semantics plus second relevance scores, the center device can fuse the first perception semantics into a first fused perception semantics according to the first relevance scores of the first perception semantics as described in Embodiment 4, and the center device can fuse the second perception semantics into a second fused perception semantics according to the second relevance scores of the second perception semantics as described in Embodiment 4; the center device can score the first fused perception semantics by measuring the relevance between the first fused perception semantics and the first query semantics as described in Embodiment 4, and score the second fused perception semantics by measuring the relevance between the second fused perception semantics and the second query semantics as described in Embodiment 4; the center device can send the first fused perception semantics with the first relevance score to the first GPT device, and send the second fused perception semantics with the second relevance score to the second GPT device, as shown in Figure 26 .

[0350] Example 6

[0351] The transmission of the query tokens (or semantics) in embodiment 3 can be integrated into the physical layer of the wireless communication system for the purpose of short latency, i.e. fast physical world search. For example, the query tokens (or semantics) can be carried in the physical layer signaling.

[0352] As described in embodiment 3, the query tokens can be fixed length scalar vectors, while the perception devices can generate perception tokens of the same length as their query tokens; the query tokens can be encoded in the following possible ways.

[0353] Alternative #1: The query tokens can be encoded as pre-defined length sequences; in a wireless system, the pre-defined length sequences can be applied (transmitted) in various forms, such as a pilot (preamble, postamble, or midamble), a mask (on a pilot, on a pilot, on a spreading code), an interleaving sequence, or other sequences.

[0354] Alternative #2: The query tokens can be encoded as special payloads with a fixed MCS scheme; for example, the length of the tokens can be a power of 2, so that the query tokens can be encoded directly using polar codes with puncturing.

[0355] Alternative #3: If the inner product is used as the scoring function, the long query tokens can be divided into several sub-blocks for incremental transmission; if the inner product on the first sub-block has already given a strong correlation, the transmission of the next sub-block can be skipped.

[0356] The transmission of the token scores (or semantics) in embodiment 5 can be integrated into the physical layer of the wireless communication for the purpose of short latency, i.e. fast physical world search. For example, the query tokens (or semantics) can be carried in the physical layer signaling.

[0357] The token scores mentioned in embodiment 5 can be scalars, while the perception devices can generate token scalars; the token scalars can be encoded.

[0358] In a wireless system, the scalars can be applied (transmitted) in various forms, such as a pilot (preamble, postamble, or midamble), a mask (on a pilot, on a pilot, on a spreading code), an interleaving sequence, or other sequences.

[0359] Next, example embodiments of a product related to the communication method will be described.

[0360] Figure 27 A block diagram of a communication apparatus 2700 is shown. As shown, the apparatus 2700 includes: Figure 27

[0361] A first receiving module 2702 configured to receive semantic tokens and tokenization configuration from a central device;

[0362] ​The determining module 2704 is configured to determine whether the perception data matches the semantic token, and the tokenization configuration is configured to obtain a perception token from the perception data.

[0363] The sending module 2706 is configured to send, to the central device, a perception result when the perception data matches the semantic token, the perception result indicating the perception data.

[0364] In a possible implementation, the first receiving module is specifically configured to receive the semantic token and the tokenization configuration broadcast or multicasted by the central device.

[0365] In a possible implementation, the semantic token includes a semantic token of one task or a semantic token of one modality.

[0366] In a possible implementation, the apparatus further includes a second receiving module configured to receive an indication of a task identity or an indication of a modality identity.

[0367] In a possible implementation, the semantic token includes a plurality of semantic tokens of a plurality of tasks and a plurality of task identities respectively associated with the plurality of semantic tokens.

[0368] In a possible implementation, the semantic token includes a plurality of semantic tokens of a plurality of modalities and a plurality of modality identities respectively associated with the plurality of semantic tokens.

[0369] In a possible implementation, the semantic token includes a first plurality of semantic tokens of a plurality of tasks, a second plurality of semantic tokens of a plurality of modalities, a plurality of task identities respectively associated with the first plurality of semantic tokens, and a plurality of modality identities respectively associated with the second plurality of semantic tokens.

[0370] In a possible implementation, the perception result includes one of the following: the original perception data; a perception semantic obtained from the original perception data; the semi-original perception data and the perception semantic obtained from the original perception data; the original perception data and a matching score between the perception data and the semantic token; the original perception data and a distance between the perception data and the semantic token; the perception semantic obtained from the original perception data and the matching score between the perception data and the semantic token; the perception semantic obtained from the original perception data and the distance between the perception data and the semantic token; the semi-original perception data, the perception semantic obtained from the original perception data, and the matching score between the perception data and the semantic token; and the semi-original perception data, the perception semantic obtained from the original perception data, and the distance between the perception data and the semantic token.

[0371] In a possible implementation, the perception result further includes the task identity or the modality identity.

[0372] In a possible implementation, the perception result further includes an identification of one semantic token whose matching score is greater than or equal to the first threshold or whose distance is less than the second threshold.

[0373] In a possible implementation, the perception result further includes a plurality of identifications of a plurality of semantic tokens whose matching scores are greater than or equal to the first threshold or whose distances are less than the second threshold.

[0374] In a possible implementation, the first receiving module is specifically configured to receive, from the central device, a sequence of a predefined length and a tokenization configuration.

[0375] In a possible implementation, the sequence of the predefined length includes a pilot code, a mask code, or an interleaving sequence.

[0376] In a possible implementation, the first receiving module is specifically configured to receive, from the central device, a payload and a tokenization configuration using a fixed modulation and coding scheme (MCS).

[0377] In a possible implementation, the apparatus further includes an obtaining module, a translating module, and a tokenization module, the obtaining module is configured to obtain perception data, the translating module is configured to translate the perception data into perception semantics according to a semanticization configuration, and the tokenization module is configured to tokenize the perception semantics into perception tokens according to a tokenization configuration; and the determining module is specifically configured to determine whether the perception tokens match the semantic tokens.

[0378] In a possible implementation, the determining module is specifically configured to determine whether a matching score between the perception tokens and the semantic tokens is greater than or equal to a first threshold, or to determine whether a distance between the perception tokens and the semantic tokens is less than or equal to a second threshold.

[0379] In a possible implementation, the tokenization configuration includes one of the following: a tokenization model, a tokenization function, a projection matrix, a graph-based or topology-based pruning, a compression method.

[0380] In a possible implementation, the apparatus further includes a third receiving module configured to receive, from the central device, at least one of the following: a scoring function for determining a matching score between the perception data and the semantic tokens, a first threshold, a function for determining a distance between the perception data and the semantic tokens, or a second threshold.

[0381] In a possible implementation, the first receiving module is specifically configured to receive, from the central device, the semantic tokens, the tokenization configuration, and at least one of the following: a scoring function for determining a matching score between the perception data and the semantic tokens, a first threshold, a function for determining a distance between the perception data and the semantic tokens, or a second threshold.

[0382] In a possible implementation, the semantic word piece is carried in high layer signaling, or physical layer signaling, or a combination of high layer signaling and physical signaling, the high layer signaling including one of radio resource control (RRC) signaling, medium access control (MAC) layer signaling, and the physical layer signaling including downlink control information (DCI).

[0383] In a possible implementation, the semantic word piece includes a query word piece.

[0384] The communication apparatus can be applied to or can be the perception device described in the method examples. Those skilled in the art should understand that, in combination with the related description of the communication method in the embodiments of the present disclosure, the related description of the modules in the examples of the present disclosure can be understood.

[0385] As shown in Figure 28 The present disclosure provides a communication apparatus 2800, including:

[0386] A first sending module 2802 is configured to send a semantic word piece and a word piece configuration to a perception device.

[0387] A first receiving module 2804 is configured to receive a perception result from the perception device, the perception data of the perception device matching the semantic word piece, the word piece configuration being used to obtain a perception word piece from the perception data, and the perception result indicating the perception data.

[0388] In a possible implementation, the first sending module is specifically configured to broadcast or groupcast the semantic word piece and the word piece configuration to a plurality of perception devices.

[0389] In a possible implementation, the semantic word piece includes a semantic word piece of one task or a semantic word piece of one modality.

[0390] In a possible implementation, the apparatus further includes a second sending module configured to send an indication of a task identity or an indication of a modality identity.

[0391] In a possible implementation, the semantic word piece includes a plurality of semantic word pieces of a plurality of tasks and a plurality of task identities respectively associated with the plurality of semantic word pieces.

[0392] In a possible implementation, the semantic word piece includes a plurality of semantic word pieces of a plurality of modalities and a plurality of modality identities respectively associated with the plurality of semantic word pieces.

[0393] In a possible implementation, the semantic tokens include a first plurality of semantic tokens of a plurality of tasks, a second plurality of semantic tokens of a plurality of modalities, a plurality of task identifiers respectively associated with the first plurality of semantic tokens, and a plurality of modality identifiers respectively associated with the second plurality of semantic tokens.

[0394] In a possible implementation, the perception result includes one of the following: the raw perception data; the perception semantics obtained from the raw perception data; the semi-raw perception data and the perception semantics obtained from the raw perception data; the raw perception data and a matching score between the perception data and the semantic token; the raw perception data and a distance between the perception data and the semantic token; the perception semantics obtained from the raw perception data and the matching score between the perception data and the semantic token; the perception semantics obtained from the raw perception data and the distance between the perception data and the semantic token; the semi-raw perception data, the perception semantics obtained from the raw perception data, and the matching score between the perception data and the semantic token; the semi-raw perception data, the perception semantics obtained from the raw perception data, and the distance between the perception data and the semantic token.

[0395] In a possible implementation, the perception result further includes the task identifier or the modality identifier.

[0396] In a possible implementation, the perception result further includes one identifier of one semantic token with a matching score greater than or equal to a first threshold or a distance less than a second threshold.

[0397] In a possible implementation, the perception result further includes a plurality of identifiers of a plurality of semantic tokens with matching scores greater than or equal to a first threshold or distances less than a second threshold.

[0398] In a possible implementation, the apparatus further includes: a first encoding module, configured to encode the semantic token into a sequence of a predefined length; and a first sending module, specifically configured to send, to the perception device, the sequence of the predefined length and the tokenization configuration.

[0399] In a possible implementation, the sequence of the predefined length includes a pilot code, a mask code, or an interleaving sequence.

[0400] In a possible implementation, the apparatus further includes: a second encoding module, configured to encode the semantic token into a payload by using a fixed modulation and coding scheme (MCS); and a first sending module, specifically configured to send, to the perception device, the payload with the fixed MCS and the tokenization configuration.

[0401] In a possible implementation, the semantic token includes a query token.

[0402] In a possible implementation, the apparatus further includes a second receiving module, a tokenization module, and an output module, the second receiving module is configured to receive semantics from a generative pre-trained transformer (GPT) device, the tokenization module is configured to tokenize the semantics into semantic tokens, and the output module is configured to output the perception result to the GPT device.

[0403] In a possible implementation, the apparatus further includes a first determining module configured to determine a tokenization configuration.

[0404] In a possible implementation, the apparatus further includes a second determining module configured to determine a token length of the semantic token.

[0405] In a possible implementation, the tokenization configuration includes one of the following: a tokenization model, a tokenization function, a projection matrix, a graph-based or topology-based pruning, and a compression method.

[0406] In a possible implementation, the method further includes a third sending module configured to send, to the perception device, at least one of the following: a scoring function used to determine a matching score between the perception data and the semantic token, a first threshold, a function used to determine a distance between the perception data and the semantic token, or a second threshold.

[0407] In a possible implementation, the first sending module is specifically configured to send, to the perception device, the semantic token, the tokenization configuration, and at least one of the following: the scoring function used to determine the matching score between the perception data and the semantic token, the first threshold, the function used to determine the distance between the perception data and the semantic token, or the second threshold.

[0408] In a possible implementation, the semantic token is carried in high-layer signaling, or physical-layer signaling, or a combination of high-layer signaling and physical signaling, the high-layer signaling includes one of the following: radio resource control (RRC) signaling, medium access control (MAC) layer signaling, and the physical-layer signaling includes downlink control information (DCI).

[0409] The communication apparatus can be applied to or can be the central device described in the method examples described above. Those skilled in the art should understand that, in combination with the related description of the communication method in the embodiments of the present disclosure, the related description of the modules in the examples of the present disclosure can be understood.

[0410] The present disclosure provides a perception device comprising processing circuitry for performing any of the above communication methods. It should be appreciated that the perception device is capable of performing the steps performed by the perception device in the above method examples, which are not repeated here.

[0411] The present disclosure provides a center device comprising processing circuitry for performing any of the above communication methods. It should be appreciated that the center device is capable of performing the steps performed by the center device in the above method examples, which are not repeated here.

[0412] The present disclosure provides a communication system comprising a center device and a perception device. The perception device is configured to perform the steps performed by the perception device in any of the communication methods, and the center device is configured to perform the steps performed by the center device in any of the communication methods.

[0413] The present disclosure provides a communication system comprising a perception device and at least one of a center device and a GPT device. The perception device is configured to perform the steps performed by the perception device in any of the communication methods, and the center device / GPT device is configured to perform the steps performed by the center device in any of the communication methods.

[0414] Embodiments of the present disclosure provide a chip comprising an input / output (I / O) interface and a processor, the processor being configured to invoke and run computer-executable instructions stored in a memory, so that a device installed with the chip is capable of performing any of the above communication methods.

[0415] Embodiments of the present disclosure provide a computer-readable medium storing computer-executable instructions, which, when executed by a processor, cause the processor to perform any of the above communication methods.

[0416] Embodiments of the present disclosure provide a computer program product comprising computer-executable instructions, which, when executed by a processor, cause the processor to perform any of the above communication methods.

[0417] Embodiments of the present disclosure provide a computer program comprising computer-executable instructions, which, when executed by a processor, cause the processor to perform any of the above communication methods.

[0418] In the present disclosure, a method for semantic / task query and response based on word pieces is provided. Some aspects of the present disclosure relate to a scheme for managing and scheduling semantic-based communication of a large number of perception devices, wherein the perception devices can belong to different types. The query semantics are target-oriented, and only the perception devices whose perception data have sufficient correlation with the semantic message will respond and send their perception data, which are preferably also in semantic form.

[0419] Some aspects of the present disclosure relate to a scheme that schedules based on collective semantic tokens on a large number of perception devices instead of one-to-one individual scheduling.

[0420] Some aspects of the present disclosure relate to a scheme that converts query data and perception data into a common semantic domain using a large language model (LLM) on which query data and perception data can be easily compared and fused with each other.

[0421] The above one or more aspects of the present disclosure can have at least one of the following benefits:

[0422] The scheduling can be task-oriented or goal-oriented; only the perception devices that contribute to the scheduled task or goal will respond and send their perception data;

[0423] Privacy can be protected: tasks, goals, or query and perception data are well protected; no raw data is sent over the air or raw data or messages sent over the air are minimized;

[0424] Forward compatibility: the semantic-based perception system in the present disclosure can be forward compatible, i.e., can support any new perception mechanism.

[0425] In some aspects of the present disclosure, a computer program including instructions is provided. The instructions, when executed by a processor, can cause the processor to implement the method of the present disclosure.

[0426] In some aspects of the present disclosure, a non-transitory computer readable medium storing instructions is provided, the instructions, when executed by a processor, can cause the processor to implement the method of the present disclosure.

[0427] In some aspects of the present disclosure, an apparatus / chipset system is provided, including means for implementing the method implemented by the perception device of the present disclosure.

[0428] In some aspects of the present disclosure, an apparatus / chipset system is provided, including means for implementing the method implemented by the center device of the present disclosure.

[0429] In some aspects of the present disclosure, an apparatus / chipset system is provided, including means for implementing the method implemented by the GPT device of the present disclosure.

[0430] In some aspects of the present disclosure, a system is provided, including at least two of the apparatus in the perception device of the present disclosure, the apparatus in the center device of the present disclosure, and the apparatus in the GPT device of the present disclosure.

[0431] In some aspects of the disclosure, an apparatus / chipset system is provided that includes at least one processor that executes instructions stored in a computer-readable medium to implement the method implemented by the perception device of the disclosure.

[0432] In some aspects of the disclosure, an apparatus / chipset system is provided that includes at least one processor that executes instructions stored in a computer-readable medium to implement the method implemented by the center device of the disclosure.

[0433] In some aspects of the disclosure, an apparatus / chipset system is provided that includes at least one processor that executes instructions stored in a computer-readable medium to implement the method implemented by the GPT device of the disclosure.

[0434] It should be noted that different examples can be implemented individually or in combination. Although combinations of features are shown in the illustrated embodiments, not all features need to be combined in order to achieve the advantages of various examples of the disclosure. In other words, a system or method designed according to an example need not necessarily include all the features shown in any single figure, or all the portions of any figure. Moreover, selected features of one example embodiment can be combined with selected features of other example embodiments.

[0435] Although the disclosure has been described with reference to the illustrative embodiments, the specification is not intended to be construed in a limiting sense. Various modifications and combinations of the illustrative embodiments, as well as other examples of the disclosure, will be apparent to those skilled in the art from the description of the disclosure. Therefore, the appended claims should not be limited to the description of the disclosure contained herein but should be construed to include any such modifications or examples.

[0436] Although the disclosure describes methods and processes with steps performed in a certain order, one or more steps of the methods and processes can be omitted or changed to suit the needs of the situation. In appropriate cases, one or more steps can be performed in an order other than the order described.

[0437] It should be noted that the expression "at least one of A or B" as used herein can be interchangeable with the expression "A and / or B." It refers to a list from which one can choose A or B or both A and B. Similarly, "at least one of A, B, or C" as used herein can be interchangeable with "A and / or B and / or C" or "A, B, and / or C." It refers to a list from which one can choose: A or B or C, or A and B, or A and C, or B and C, or all of A, B, and C. The same principle applies to longer lists with the same format.

[0438] Although the present disclosure is described in terms of methods, one of ordinary skill in the art will appreciate that the present disclosure is also directed to various components for performing at least some of the aspects and features of the described methods, whether by hardware components, software, or any combination of the two. Accordingly, the technical solutions of the present disclosure can be embodied in the form of a software product. The appropriate software product can be stored in a pre-recorded storage device or other similar non-volatile or non-transitory computer-readable medium, including DVDs, CD-ROMs, U disks, removable hard disks, or other storage media. The software product includes instructions tangibly stored thereon, which enable a processing device (e.g., a personal computer, a server, or a network device) to perform the method examples disclosed herein. The machine-executable instructions can be in the form of code sequences, configuration information, or other data, which, when executed, cause a machine (e.g., a processor or other processing device) to perform the steps in the methods provided by the examples of the present disclosure.

[0439] The present disclosure can be implemented in other specific forms without departing from the subject matter of the claims. The examples are considered in all respects to be only illustrative and not restrictive. Selected features from one or more of the above-described examples can be combined to create alternative examples not explicitly described, and all such combinations are understood to fall within the scope of the present disclosure.

[0440] All values and subranges within the disclosed ranges are also disclosed. In addition, although the systems, devices, and processes disclosed and shown herein can include particular numbers of elements / components, the systems, devices, and components can be modified to include more or less such elements / components. For example, although any of the disclosed elements / components can be referred to as a singular number, the examples disclosed herein can be modified to include a plurality of such elements / components. The subject matter described herein is intended to cover and encompass all suitable variations.

[0441] While examples have been described above with reference to the drawings, those skilled in the art will appreciate that changes and modifications can be made without departing from the scope of the appended claims.

Claims

1. A communication method characterized by comprising: Comprising: receiving, from a central device, a semantic token and a tokenization configuration; determining whether perception data matches the semantic token, the tokenization configuration being used to obtain a perception token from the perception data; when the perception data matches the semantic token, sending, to the central device, a perception result, the perception result indicating the perception data.

2. The method of claim 1, wherein, The receiving, from the central device, the semantic token and the tokenization configuration comprises: receiving the semantic token and the tokenization configuration broadcasted or groupcast by the central device.

3. The method according to claim 1 or 2, characterized in that, The semantic token comprises a semantic token of one task or a semantic token of one modality.

4. The method of claim 3, wherein, The method further comprises: receiving an indication of a task identity or an indication of a modality identity.

5. The method according to claim 1 or 2, characterized in that, The semantic token comprises a plurality of semantic tokens of a plurality of tasks and a plurality of task identities respectively associated with the plurality of semantic tokens.

6. The method of claim 1 or 2, wherein, The semantic token comprises a plurality of semantic tokens of a plurality of modalities and a plurality of modality identities respectively associated with the plurality of semantic tokens.

7. The method according to claim 1 or 2, characterized in that, The semantic token comprises a first plurality of semantic tokens of a plurality of tasks, a second plurality of semantic tokens of a plurality of modalities, a plurality of task identities respectively associated with the first plurality of semantic tokens, and a plurality of modality identities respectively associated with the second plurality of semantic tokens.

8. The method according to any one of claims 1 to 7, characterized in that, The perception result comprises one of: raw perception data; perception semantics obtained from raw perception data; semi-raw perception data and perception semantics obtained from raw perception data; raw perception data and a matching score between the perception data and the semantic token; a distance between the perception data and the semantic token; perception semantics obtained from raw perception data and a matching score between the perception data and the semantic token; perception semantics obtained from raw perception data and a distance between the perception data and the semantic token; semi-raw perception data, perception semantics obtained from raw perception data, and a matching score between the perception data and the semantic token; semi-raw perception data, perception semantics obtained from raw perception data, and a distance between the perception data and the semantic token.

9. The method of claim 8, wherein, The perception result further comprises a task identity or a modality identity.

10. The method according to claim 8 or 9, characterized in that, The perception result further comprises one identity of one semantic token having a matching score greater than or equal to a first threshold or a distance less than a second threshold.

11. The method according to claim 8 or 9, characterized in that, The perception result further comprises a plurality of identities of a plurality of semantic tokens having a matching score greater than or equal to a first threshold or a distance less than a second threshold.

12. The method according to any one of claims 1 to 11, characterized in that, The receiving, from the central device, the semantic token and the tokenization configuration comprises: receiving, from the central device, a sequence of a predefined length and the tokenization configuration.

13. The method of claim 12, wherein, The sequence of the predefined length comprises a pilot sequence, a mask sequence, or an interleaving sequence.

14. The method according to any one of claims 1 to 11, characterized in that, The receiving, from the central device, the semantic token and the tokenization configuration comprises: receiving, from the central device, a payload with a fixed modulation and coding scheme (MCS) and the tokenization configuration.

15. The method according to any one of claims 1 to 14, characterized in that, The method further comprises: obtaining the perception data; translating the perception data into perception semantics according to a semanticization configuration; tokenizing the perception semantics into the perception token according to the tokenization configuration; determining whether the perception data matches the semantic token comprises: determining whether the perception token matches the semantic token.

16. The method of claim 15, wherein, the determining whether the perception token matches the semantic token comprises: determining whether a matching score between the perception token and the semantic token is greater than or equal to a first threshold; or determining whether a distance between the perception token and the semantic token is less than or equal to a second threshold.

17. The method of any one of claims 1 to 16, wherein, the tokenization configuration comprises one of: a tokenization model; a tokenization function; a projection matrix; a graph-based or topology-based pruning; or a compression method.

18. The method of any one of claims 1 to 17, wherein, the method further comprises: receiving, from the central device, at least one of a scoring function for determining a matching score between the perception data and the semantic token, a first threshold, a function for determining a distance between the perception data and the semantic token, or a second threshold.

19. The method of any one of claims 1 to 17, wherein, the receiving, from the central device, the semantic token and the tokenization configuration comprises: receiving, from the central device, the semantic token, the tokenization configuration, and at least one of a scoring function for determining a matching score between the perception data and the semantic token, a first threshold, a function for determining a distance between the perception data and the semantic token, or a second threshold.

20. The method of any one of claims 1 to 19, wherein, the semantic token is carried in high layer signaling, or physical layer signaling, or a combination of high layer signaling and physical signaling, the high layer signaling comprising one of radio resource control (RRC) signaling, medium access control (MAC) layer signaling, the physical layer signaling comprising downlink control information (DCI).

21. The method of any one of claims 1 to 20, wherein, the semantic token comprises a query token.

22. A method of communication, comprising: comprises: sending, to a perception device, a semantic token and a tokenization configuration; receiving, from the perception device, a perception result, wherein perception data of the perception device matches the semantic token, the tokenization configuration is used to obtain a perception token from the perception data, and the perception result indicates the perception data.

23. The method of claim 22, wherein, the sending, to the perception device, the semantic token and the tokenization configuration comprises: broadcasting or multicasting the semantic token and the tokenization configuration to a plurality of perception devices.

24. The method of claim 22 or 23, wherein, the semantic token comprises a semantic token of one task or a semantic token of one modality.

25. The method of claim 24, wherein, the method further comprises: sending an indication of a task identity or an indication of a modality identity.

26. The method of claim 22 or 23, wherein, the semantic token comprises a plurality of semantic tokens of a plurality of tasks and a plurality of task identities respectively associated with the plurality of semantic tokens.

27. The method of claim 22 or 23, wherein, the semantic token comprises a plurality of semantic tokens of a plurality of modalities and a plurality of modality identities respectively associated with the plurality of semantic tokens.

28. The method of claim 22 or 23, wherein, the semantic token comprises a first plurality of semantic tokens of a plurality of tasks, a second plurality of semantic tokens of a plurality of modalities, a plurality of task identities respectively associated with the first plurality of semantic tokens, and a plurality of modality identities respectively associated with the second plurality of semantic tokens.

29. The method of any one of claims 22-28, wherein, the perception result comprises one of: raw perception data; perception semantics obtained from raw perception data; semi-raw perception data and perception semantics obtained from raw perception data; raw perception data and a matching score between the perception data and the semantic token; raw perception data and a distance between the perception data and the semantic token; perceptual semantics obtained from raw perceptual data and a matching score between the perceptual data and the semantic token; perceptual semantics obtained from raw perceptual data and a distance between the perceptual data and the semantic token; semi-raw perceptual data, perceptual semantics obtained from raw perceptual data, and a matching score between the perceptual data and the semantic token; semi-raw perceptual data, perceptual semantics obtained from raw perceptual data, and a distance between the perceptual data and the semantic token.

30. The method of claim 29, wherein, The perceptual result further comprises a task identification or a modality identification.

31. The method of claim 29 or 30, wherein, The perceptual result further comprises an identification of one semantic token having a matching score greater than or equal to a first threshold or a distance less than a second threshold.

32. The method of claim 29 or 30, wherein, The perceptual result further comprises a plurality of identifications of a plurality of semantic tokens having a matching score greater than or equal to a first threshold or a distance less than a second threshold.

33. The method of any one of claims 22-32, wherein, The method further comprises: encoding the semantic token into a sequence of a predefined length; The sending the semantic token and the tokenization configuration to the perceptual device comprises: sending the sequence of the predefined length and the tokenization configuration to the perceptual device.

34. The method of claim 33, wherein, The sequence of the predefined length comprises a pilot code, a masking code, or an interleaving sequence.

35. The method of any one of claims 22-32, wherein, The method further comprises: encoding the semantic token into a payload using a fixed modulation coding scheme (MCS); The sending the semantic token and the tokenization configuration to the perceptual device comprises: sending the payload using the fixed MCS and the tokenization configuration to the perceptual device.

36. The method of any one of claims 22-35, wherein, The semantic token comprises a query token.

37. The method of any one of claims 22-36, wherein, The method further comprises: receiving semantics from a generative pre-trained transformer (GPT) device; tokenizing the semantic token into the semantic token; outputting the perceptual result to the GPT device.

38. The method of any one of claims 22-37, wherein, The method further comprises: determining the tokenization configuration.

39. The method of any one of claims 22-38, wherein, The method further comprises: determining a token length of the semantic token.

40. The method of any one of claims 22-39, wherein, The tokenization configuration comprises one of: a tokenization model; a tokenization function; a projection matrix; a graph-based or topology-based pruning; or a compression method.

41. The method of any one of claims 22-40, wherein, The method further comprises sending, to the perceptual device, at least one of a scoring function for determining a matching score between the perceptual data and the semantic token, a first threshold, a function for determining a distance between the perceptual data and the semantic token, or a second threshold.

42. The method of any one of claims 22-40, wherein, The sending the semantic token and the tokenization configuration to the perceptual device comprises: sending, to the perceptual device, the semantic token, the tokenization configuration, and at least one of a scoring function for determining a matching score between the perceptual data and the semantic token, a first threshold, a function for determining a distance between the perceptual data and the semantic token, or a second threshold.

43. The method of any one of claims 22-42, wherein, The semantic token is carried in high layer signaling, or physical layer signaling, or a combination of high layer signaling and physical signaling, the high layer signaling comprising one of radio resource control (RRC) signaling, medium access control (MAC) layer signaling, the physical layer signaling comprising downlink control information (DCI).

44. A communications device, characterized by comprising means for performing the method according to any one of claims 1 to 21, or means for performing the method according to any one of claims 22 to 43.

45. An electronic device, comprising: comprising processing circuitry for performing the method according to any one of claims 1 to 21, or performing the method according to any one of claims 22 to 43.

46. A chip, comprising: comprising an input / output, I / O, interface and a processor configured to invoke and run a computer program stored in a memory, such that a device in which the chip is installed is enabled to perform the method according to any one of claims 1 to 21, or perform the method according to any one of claims 22 to 43.

47. A sensing device, comprising: comprising: one or more processors; a non-transitory computer-readable storage medium coupled to the one or more processors and storing programming for execution by the processors, wherein the perception device, when the programming is executed by the processors, is configured to perform the method according to any one of claims 1 to 21.

48. A central device, comprising: comprising: one or more processors; a non-transitory computer-readable storage medium coupled to the processors and storing programming for execution by the processors, wherein the central device, when the programming is executed by the processors, is configured to perform the method according to any one of claims 22 to 43.

49. A communication system, characterized by comprising the perception device according to claim 47 and the central device according to claim 48.

50. A non-transitory computer readable medium, characterized in that, carrying program code which, when executed by a computer device, causes the computer device to perform the method according to any one of claims 1 to 21 or the method according to any one of claims 22 to 43.

51. A computer program product, characterised in that, comprising program code for performing the method according to any one of claims 1 to 21 or the method according to any one of claims 22 to 43 when the program code is executed on a computer or a processor.

Images