First relay device, second relay device, and communication method

Abstract

Disclosed in an embodiment of the present application are a first relay device, a second relay device and a communication method, the communication method being applied to the first relay device, comprising: receiving a service request of a first task sent by a first user equipment; determining a second relay device and a second user device according to task information of the first task included in the service request; sending, via the first transceiver, a first confirmation request to the second relay device, the first confirmation request being used to confirm whether the second relay device is deployed with a recovery model regarding the first task; and sending, via the first transceiver, a second model confirmation request to the second user equipment, the second confirmation request to confirm whether the second user equipment is deployed with a decoding model regarding the first task.

Description

Technical Field

[0001] The present application relates to, but is not limited to, the field of communication technology, and in particular to a first relay device, a second relay device, and a communication method. Background Art

[0002] Semantic communication significantly improves communication efficiency and robustness. However, in real-world communication environments, signal fading is severe due to factors such as path loss and obstacles in wireless channels. This results in low reliability and severe distortion in end-to-end semantic information transmission, making it difficult to meet communication requirements. Codebook-based semantic communication technologies utilize codebooks as a shared knowledge base and explore the effective mapping between codebook sequences and semantic feature vectors. However, these codebook-based semantic communication solutions all separate source semantic coding from channel coding, making them ineffective in combating the effects of channel fading. Summary of the Invention

[0003] The embodiments of the present application at least provide a first relay device, a second relay device, and a communication method.

[0004] The technical solution of the embodiment of the present application is implemented as follows:

[0005] An embodiment of the present application provides a first relay device, the first relay device comprising: a first transceiver; and

[0006] A first processor is coupled to the first transceiver; the first processor is configured to:

[0007] receiving, via the first transceiver, a service request for a first task sent by a first user equipment;

[0008] determining a second relay device and a second user device according to task information of the first task included in the service request;

[0009] Sending a first confirmation request to the second relay device via the first transceiver, where the first confirmation request is used to confirm whether the second relay device has deployed a recovery model related to the first task;

[0010] A second model confirmation request is sent to the second user equipment via the first transceiver, where the second confirmation request is used to confirm whether the second user equipment is deployed with a decoding model related to the first task.

[0011] In some embodiments, the first processor is configured to:

[0012] receiving, via the first transceiver, a first model confirmation response sent by the second relay device, confirming that the recovery model is deployed;

[0013] Receiving, via the first transceiver, a second model confirmation response sent by the second user equipment, confirming that the decoding model is deployed;

[0014] A model deployment confirmation message is sent to the first user equipment via the first transceiver.

[0015] In some embodiments, the first processor is configured to:

[0016] receiving, via the first transceiver, a third model confirmation response sent by the second relay device, confirming that no recovery model is deployed;

[0017] The recovery model is sent to the second relay device via the first transceiver; or the second relay device is updated.

[0018] In some embodiments, the first processor is configured to:

[0019] Receiving, via the first transceiver, a fourth model confirmation response sent by the second user equipment, confirming that no decoding model is deployed;

[0020] The decoding model is sent to the second user equipment via the first transceiver.

[0021] In some embodiments, the first processor is configured to:

[0022] A recovery model is deployed to the second relay device, and / or a decoding model is deployed to the second user equipment.

[0023] In some embodiments, the first processor is configured to:

[0024] According to the model deployment information, a recovery model is deployed to the second relay device, and / or a decoding model is deployed to the second user equipment.

[0025] In some embodiments, the first processor is configured to:

[0026] Sending, via the first transceiver, a first model acquisition message to the first user equipment and a second model acquisition message to the second user equipment when the model deployment information indicates that the second relay device has not deployed a recovery model and the second user equipment has deployed a decoding model;

[0027] Receiving, via the first transceiver, the encoding model for the first task in response to the first model acquisition message sent by the first user equipment, and the decoding model in response to the second model acquisition message sent by the second user equipment;

[0028] Determine a recovery model based on the encoding model and the decoding model;

[0029] The recovery model is sent to the second relay device via the first transceiver.

[0030] In some embodiments, the first processor is configured to:

[0031] Sending, via the first transceiver, a first model acquisition message to the first user device and a third model acquisition message to the second relay device when the model deployment information indicates that the second relay device is deployed with a recovery model and the second user device is not deployed with a decoding model;

[0032] Receiving, via the first transceiver, the encoding model for the first task in response to the first model acquisition message sent by the first user equipment, and the recovery model in response to the third model acquisition message sent by the second relay device;

[0033] Determine a decoding model based on the encoding model and the recovery model;

[0034] The decoding model is sent to the second user equipment via the first transceiver.

[0035] In some embodiments, the first processor is configured to:

[0036] Sending, via the first transceiver, a first model acquisition message to the first user equipment when the model deployment information indicates that the second relay device is not deployed with a recovery model and the second user equipment is not deployed with a decoding model;

[0037] Receiving, via the first transceiver, a coding model for the first task in response to a first model acquisition message sent by the first user equipment;

[0038] Determine the recovery model and decoding model according to the encoding model;

[0039] The recovery model is sent to the second relay device via the first transceiver, and the decoding model is sent to the second user equipment.

[0040] In some embodiments, the first processor is configured to:

[0041] locally finding the recovery model and the encoding model for the first task;

[0042] sending the searched local search model to the second relay device via the first transceiver;

[0043] The found decoding model is sent to the second user equipment via the first transceiver.

[0044] In some embodiments, the first processor is configured to:

[0045] Determining an encoding model, a recovery model, and a decoding model for the first task according to model requirement information of the first task included in the service request;

[0046] The encoding model is sent to the first user equipment via the first transceiver, the recovery model is sent to the second relay device, and the decoding model is sent to the second user equipment.

[0047] In some embodiments, the first processor is configured to:

[0048] Sending a third confirmation request to the first user equipment via the first transceiver, where the third confirmation request is used to confirm whether the first user equipment is deployed with the coding model;

[0049] A third model confirmation response is received via the first transceiver and is sent by the first user equipment, confirming that the coding model is deployed.

[0050] In some embodiments, the first processor is configured to:

[0051] Determining, according to the service request, whether the first user equipment is deployed with a coding model related to the first task;

[0052] Wherein, when it is determined that the first user equipment is deployed with a coding model, a recovery model is deployed to the second relay device, and / or a decoding model is deployed to the second user equipment; and / or,

[0053] When it is determined that the first user equipment is not deployed with a coding model, a coding model, a recovery model, and a decoding model are determined according to model requirement information of the first task included in the service request.

[0054] An embodiment of the present application provides a second relay device, the second relay device including a second transceiver; and

[0055] a second processor coupled to the second transceiver; the second processor configured to:

[0056] A first confirmation request is received via the second transceiver from the first relay device when the first relay device receives the service request sent by the first user equipment. The first confirmation request is used to confirm whether the second relay device has deployed a recovery model related to the first task.

[0057] In some embodiments, the second processor is configured to:

[0058] Receiving, via a second transceiver, a first codebook index sent by a first user equipment, where the first codebook index is obtained by quantizing first coded data based on a codebook, and the first codebook index is obtained by semantically encoding data of a first task by a coding model;

[0059] Based on the recovery model, recover the first codebook index as the second codebook index;

[0060] A second codebook index is sent to a second user equipment via a second transceiver; the second codebook index is used to restore the second coded data based on the codebook, and the second coded data is decoded by a decoding model to obtain data of the first task.

[0061] In some embodiments, the second processor is configured to:

[0062] receiving, via the second transceiver, a first transmission request sent by the first user equipment;

[0063] Sending, via the second transceiver, a second transmission request to the second user equipment based on the triggering of the first transmission request;

[0064] receiving, via the second transceiver, a first transmission confirmation message sent by the second user equipment in response to the second transmission request;

[0065] A second transmission confirmation message is sent to the first user equipment via the second transceiver based on the triggering of the first transmission confirmation message, where the second transmission message indicates that the second relay device and the second user equipment meet the data transmission condition.

[0066] In some embodiments, the second processor is configured to:

[0067] A first model confirmation response is received via the second transceiver and is sent to the first relay device to confirm that the recovery model is deployed.

[0068] In some embodiments, the second processor is configured to:

[0069] receiving, via the second transceiver, a third model confirmation response sent to the first relay device to confirm that no recovery model is deployed;

[0070] The recovery model sent by the first relay device is received via the second transceiver.

[0071] In some embodiments, the second processor is configured to:

[0072] Receiving, via the second transceiver, a third model acquisition message sent by the first relay device; the third model acquisition message is sent by the first relay device when the model deployment information indicates that the second relay device is deployed with a recovery model and the second user equipment is not deployed with a decoding model;

[0073] The recovery model in response to the third model acquisition message is sent to the first relay device via the second transceiver.

[0074] In some embodiments, the second processor is configured to:

[0075] The recovery model sent by the first relay device is received via the second transceiver.

[0076] An embodiment of the present application provides a communication method, applied to a first relay device, including:

[0077] receiving a service request for a first task sent by a first user device;

[0078] determining a second relay device and a second user device according to task information of the first task included in the service request;

[0079] Sending a first confirmation request to the second relay device, where the first confirmation request is used to confirm whether the second relay device has deployed a recovery model related to the first task;

[0080] A second model confirmation request is sent to the second user equipment, where the second confirmation request is used to confirm whether the second user equipment is deployed with a decoding model related to the first task.

[0081] In an embodiment of the present application, the first relay device determines the second relay device and the second user device based on the task information of the first task included in the service request sent by the first user device, and sends a first confirmation request for confirming whether the second relay device has deployed a recovery model for the first task and a second confirmation request for confirming whether the second user device has deployed a decoding model for the first task to the second relay device and the second user device, respectively. In this way, when the second relay device participates, the first relay device sends a first confirmation request and a second confirmation request to the second relay device and the second user device, respectively, to determine whether the relevant models of the second relay device and the second user device have been successfully deployed. By confirming the model deployment status in advance, communication failure or performance degradation caused by model mismatch can be effectively avoided while better adapting to more severe channel environments and improving the performance of the system in low signal-to-noise ratio environments.

[0082] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and do not limit the technical solutions of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The drawings herein are incorporated into and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and, together with the specification, are used to illustrate the technical solutions of the present application.

[0084] Figure 1 A schematic diagram of an optional flow chart of the communication method provided in an embodiment of the present application;

[0085] Figure 2 A schematic diagram of an optional flow chart of the communication method provided in an embodiment of the present application;

[0086] Figure 3 A schematic diagram of an optional flow chart of the communication method provided in an embodiment of the present application;

[0087] Figure 4 Schematic diagram of a codebook-based semantic relay system provided in an embodiment of the present application;

[0088] Figure 5A schematic flow chart of a codebook-based semantic relay solution provided in an embodiment of the present application;

[0089] Figure 6 Schematic diagram of the service request and model matching process of the codebook-based semantic relay solution provided in an embodiment of the present application;

[0090] Figure 7 A schematic diagram of the interaction between a control node, a relay node, user 1, and user 2 in a communication system provided in an embodiment of the present application;

[0091] Figure 8 A schematic diagram of the codebook-based semantic relay information transmission process provided in an embodiment of the present application;

[0092] Figure 9 A schematic diagram of the codebook rearrangement algorithm flow provided in an embodiment of the present application;

[0093] Figure 10 An optional structural diagram of the device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0094] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions of this application are further elaborated in detail below with reference to the accompanying drawings and embodiments. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0095] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0096] It should be pointed out that the terms "first\second\third" involved in the embodiments of the present application are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.

[0097] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as generally understood by those skilled in the art in the art to which the embodiments of the present application belong. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0098] The embodiments of the present application provide a first relay device, a second relay device, and a communication method.

[0099] In a first aspect, an embodiment of the present application provides a communication method, which is applied to a first relay device, such as Figure 1 Shown, including:

[0100] S101: Receive a service request for a first task sent by a first user equipment;

[0101] S102: Determine a second relay device and a second user equipment according to the task information of the first task included in the service request;

[0102] S103: Send a first confirmation request to the second relay device, where the first confirmation request is used to confirm whether the second relay device has deployed a recovery model for the first task;

[0103] S104: Send a second model confirmation request to the second user equipment, where the second confirmation request is used to confirm whether the second user equipment is deployed with a decoding model related to the first task.

[0104] In a second aspect, an embodiment of the present application provides a communication method, which is applied to a second relay device, such as Figure 2 Shown, including:

[0105] S201. Receive a first confirmation request sent by a first relay device when receiving a service request sent by a first user equipment, where the first confirmation request is used to confirm whether the second relay device is deployed with a recovery model related to the first task.

[0106] In a third aspect, an embodiment of the present application provides a communication method, which is applied to a communication system including a first relay device, a second relay device, a first user device, and a second user device, such as Figure 3 Shown, including:

[0107] S301: A first user device sends a service request for a first task to a first relay device.

[0108] S302: The first relay device determines a second relay device and a second user equipment according to the task information of the first task included in the service request;

[0109] S303: The first relay device sends a first confirmation request to the second relay device, where the first confirmation request is used to confirm whether the second relay device has deployed a recovery model for the first task.

[0110] S304. The first relay device sends a second model confirmation request to the second user equipment, where the second confirmation request is used to confirm whether the second user equipment is deployed with a decoding model related to the first task.

[0111] Next, Figure 1 、 Figure 2 or Figure 3 The communication method shown is described.

[0112] Here, the first user device and the second user device are two devices used for communication in a communication network, wherein the first user device is the data transmitter and the second user device is the data receiver. It can be understood that a terminal device can act as both the data transmitter (the first user device) and the data receiver (the second user device) in a communication.

[0113] The first user equipment and the second user equipment communicate through the relay network.

[0114] The first relay device is a control device in the relay network for network resource management, control or service, that is, the first relay device can be called a control node. The second relay device is a relay node in the relay network for implementing information relay between the first user equipment and the second user equipment.

[0115] When the first user device has data to be sent to the second user device, it sends a service request to the first relay device. The service request is a communication request initiated by the first user device to the first relay device, and includes task information of the first task to be performed. The user device can be a terminal.

[0116] Here, the first user device may be the transmitter of data related to the first task, the second user device may be the receiver of data related to the first task, and the second relay device may be a relay node for data transmission between the first user device and the second user device. The first user device and the second user device may perform semantic communication. In this case, the service request may be a semantic communication scheduling request, which is used to request semantic communication with the second user device.

[0117] The relay network includes relay nodes that can perform relay transmission, and control nodes. In some embodiments, the control node can communicate directly with the user equipment; in some embodiments, the control node is connected to the user equipment through the relay node to achieve communication between the control node and the user equipment.

[0118] When the first user device is able to communicate directly with the first relay device, the first user device may directly send a service request to the first relay device. When the first user device is unable to communicate directly with the first relay device, the first user device may send a service request to the first relay device through a third relay device in the relay network that can be directly connected to the first user device. The third relay device may be the second relay device or a relay device other than the second relay device.

[0119] In some embodiments, the task information may include at least one of the following information: device information of the first user device, device information of the second user device, a service type of the first task, and computing capability requirements of the relay node for the first task.

[0120] In some embodiments, the relay network includes multiple relay devices, and not all relay devices may meet the transmission requirements of the first task. Therefore, the first relay device selects a relay device from the multiple relay devices that meets the transmission requirements of the first task as a relay device based on the task information in the service request. For example, a feasible second relay device may be selected that meets the computing power requirements of the first task for the second relay device. The selected second relay device can ensure the reliability of the execution of the first task.

[0121] In some embodiments, the relay network may select a second relay device from connectable intermediate devices that can connect to the first user device and the second user device, ensuring determinism of the communication link from the first user device, the second relay device, and the second user device.

[0122] In some embodiments, when there are multiple optional second relay devices, the first relay device can select a second relay device from the multiple optional second relay devices based on one or more information such as channel conditions and load to ensure the processing efficiency of the first task.

[0123] The first relay device determines the second user equipment according to the device information of the second user equipment included in the task information.

[0124] After receiving the service request, the first relay device sends a first confirmation request and a second confirmation request to the second relay device and the second user device, respectively. The first confirmation request is used to confirm whether the second relay device has deployed a recovery model related to the first task, and the second confirmation request is used to confirm whether the second user device has deployed a decoding model related to the first task. In this embodiment of the present application, the first confirmation request and the second confirmation request may include task information representing the first task.

[0125] The second relay device confirms whether it has deployed a recovery model related to the first task in response to the received first model confirmation request; the second user device confirms whether it has deployed a decoding model related to the first task in response to the received second model confirmation request.

[0126] In an embodiment of the present application, when the first user device confirms that the second relay device is deployed with a recovery model related to the first task, and the second user device is deployed with a decoding model related to the first task, and confirms that it is deployed with an encoding model related to the first task, a data transmission request is sent to the second relay device.

[0127] In an embodiment of the present application, the first user device may deploy an encoding model, the second relay device may deploy a recovery model, and the second user device may deploy a decoding model. In the case of semantic communication, the encoding model is deployed on the first user device to encode the semantic information to be transmitted and obtain the encoded first codebook index. The recovery model is a neural network model deployed on the second relay device, which is used to recover the first codebook index sent by the first user device on the second relay device to obtain the recovered second codebook index. The decoding model is deployed on the second user device to decode according to the second codebook index sent by the second relay device to obtain the decoded data of the first task.

[0128] During a communication process, in a first user device, the data of the first task is semantically encoded using a coding model to obtain first coded data, the first coded data is quantized into a first codebook index based on a codebook, and the first codebook index is sent by the first user device to the relay device; in a second relay device, the first codebook index received by the second relay device is recovered using a recovery model to obtain a second codebook index, and the second codebook index is sent by the second relay device to the second user device; in the second user device, the received second codebook index is restored based on the codebook to obtain second coded data, and the second coded data is semantically decoded using a decoding model to obtain data of the first task.

[0129] In an embodiment of the present application, the first relay device determines the second relay device and the second user device based on the task information of the first task included in the service request sent by the first user device, and sends a first confirmation request for confirming whether the second relay device has deployed a recovery model for the first task and a second confirmation request for confirming whether the second user device has deployed a decoding model for the first task to the second relay device and the second user device, respectively. In this way, when the second relay device participates, the first relay device sends a first confirmation request and a second confirmation request to the second relay device and the second user device, respectively, to determine whether the relevant models of the second relay device and the second user device have been successfully deployed. By confirming the model deployment status in advance, communication failure or performance degradation caused by model mismatch can be effectively avoided while better adapting to more severe channel environments and improving the performance of the system in low signal-to-noise ratio environments.

[0130] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving a first model confirmation response sent by the second relay device to confirm that the recovery model is deployed; receiving a second model confirmation response sent by the second user device to confirm that the decoding model is deployed; and sending a model deployment confirmation message to the first user device.

[0131] Accordingly, based on Figure 2 The communication method shown, the communication method provided by the embodiment of the present application also includes: sending a first model confirmation response to confirm that the recovery model is deployed.

[0132] Correspondingly, the second user equipment sends a second model confirmation response confirming that the decoding model is deployed.

[0133] Correspondingly, the first user equipment receives the model deployment confirmation message sent by the first relay device.

[0134] Here, the first model confirmation response refers to a message sent by the second relay device, used to confirm that the second relay device has successfully deployed the recovery model for the first task.

[0135] The first model confirmation response may include information such as the model version number, deployment status, and deployment time corresponding to the recovery model. By receiving the first model confirmation response, the first relay device can determine that the second relay device has deployed the recovery model, i.e., the second relay device has the ability to perform data recovery related to the first task.

[0136] The second model confirmation response is a signal sent by the second user device confirming that the second user device has successfully deployed the decoding model for the first task. The second model confirmation response may include key information such as the model version number, deployment status, and deployment time of the decoding model. By receiving the second model confirmation response, the first relay device can determine that the second user device has deployed the decoding model, meaning that the second user device is capable of decoding data related to the first task.

[0137] Upon receiving the first model confirmation response and the second model confirmation response, the first user device sends a model deployment confirmation message to the first user device. The model deployment confirmation message is generated by the first relay device and sent to the first user device to inform the first user device of the current model matching status and deployment status. The model deployment confirmation message is used to synchronize the model deployment status in the communication link with the first user device, allowing the first user device to confirm that the remaining communication terminals have models that meet the data transmission requirements.

[0138] In an embodiment of the present application, the second relay device sends a first model confirmation response to the first relay device when confirming that it has deployed a recovery model related to the first task; the first relay device receives the first model confirmation response and confirms that the second relay device has deployed a recovery model related to the first task; the second user device sends a second model confirmation response to the first relay device when confirming that it has deployed a decoding model related to the first task; the first relay device receives the second model confirmation response and confirms that the second user device has deployed a decoding model related to the first task; when the first relay device confirms that the first user device has deployed the encoding model, the second relay device has deployed the recovery model, and the second user device has deployed the decoding model, it sends a model deployment confirmation message to the first user device; the first user device receives the model deployment confirmation message, initiates a data transmission request to the second relay device, and enters the data transmission phase.

[0139] In an embodiment of the present application, by receiving a first model confirmation response sent by a second relay device and a second model confirmation response sent by a second user device, the first relay device confirms that the second relay device is deployed with a recovery model and that the second user device is deployed with a decoding model, and synchronizes the model deployment status to the first user device. In this way, each node in the communication link can be deployed with relevant model support, thereby effectively avoiding communication failures caused by model mismatch, thereby improving the overall stability and robustness of the semantic communication system.

[0140] In some embodiments, based on Figure 1The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving a third model confirmation response sent by the second relay device confirming that the recovery model is not deployed; sending the recovery model to the second relay device; or updating the second relay device.

[0141] Accordingly, based on Figure 2 The communication method shown, the communication method provided by the embodiment of the present application also includes: sending a third model confirmation response confirming that the recovery model is not deployed; and receiving the recovery model sent by the first relay device.

[0142] After receiving the first confirmation request, the second relay device determines whether it has deployed a recovery model based on the first confirmation request. If it determines that it has not deployed a recovery model, it returns a third model confirmation response to the first relay device. The third model confirmation response is sent by the second relay device to indicate that the second relay device does not currently have a recovery model deployed to perform the first task. The third model confirmation response can be returned to the first relay device in the form of a data packet via a communication link.

[0143] In the embodiment of the present application, upon receiving the third model confirmation response, the first relay device confirms that the second relay device has not deployed the recovery model.

[0144] When confirming that the second relay device has not deployed the recovery model, the first relay device may send the recovery model to the second relay device to redeploy the recovery model in the second relay device. The second relay device receives the recovery model sent by the first relay device and deploys the recovery model locally.

[0145] When the first relay device confirms that the second relay device has not deployed the recovery model, the second relay device can be replaced. The first relay device can send a first confirmation request to the replaced second relay device. The replaced second relay device responds to the first confirmation request and sends a corresponding model confirmation response to the first relay device based on whether it has deployed the recovery model.

[0146] In the embodiment of the present application, when confirming that the second relay device has not deployed the recovery model, the first relay device may return a task cancellation message to the first user device to indicate the cancellation of the execution of the local first task.

[0147] In the embodiment of the present application, upon receiving a third model confirmation response from the second relay device, the first relay device sends a recovery model to the second relay device or updates the second relay device. This enables the second relay device to have the ability to recover the codebook index for the first task, thereby supporting data transmission between the first user device and the second user device.

[0148] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving a fourth model confirmation response sent by the second user equipment to confirm that the decoding model is not deployed; and sending the decoding model to the second user equipment.

[0149] Correspondingly, the second user equipment sends a fourth model confirmation response confirming that the decoding model is not deployed; and receives the decoding model sent by the first relay device.

[0150] After receiving the second confirmation request, the second user equipment determines whether it has deployed the decoding model based on the second confirmation request. If it is determined that it has not deployed the decoding model, it returns a fourth model confirmation response to the first relay device.

[0151] The fourth model confirmation response is sent by the second user device to the first relay device, indicating that the second user device has not currently deployed a decoding model related to the first task. The fourth model confirmation response can be returned to the first relay device in the form of a data packet via the communication link.

[0152] In the embodiment of the present application, upon receiving the fourth model confirmation response, the first relay device confirms that the second user equipment has not deployed the decoding model.

[0153] If the first relay device confirms that the second user device does not have a decoding model deployed, it can send the decoding model to the second user device to redeploy the decoding model in the second user device. The second user device receives the decoding model sent by the first relay device and locally deploys the decoding model. In this way, the second user device can have the decoding capability for the first task, thereby supporting data transmission between the first user device and the second user device.

[0154] In the embodiment of the present application, when confirming that the second relay device has not deployed the recovery model, the first relay device may return a task cancellation message to the first user device to indicate the cancellation of the execution of the local first task.

[0155] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: deploying the recovery model to the second relay device, and / or deploying the decoding model to the second user device.

[0156] Accordingly, based on Figure 2 The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving the recovery model sent by the first relay device.

[0157] Correspondingly, the second user equipment receives the decoding model sent by the first relay device.

[0158] In an embodiment of the present application, the first relay device may send a recovery model to the second relay device and / or send a decoding model to the second user device based on the model deployment status of the second relay device and the second user device.

[0159] After receiving the recovery model, the second relay device deploys the recovery model. Here, the second relay device can load the recovery model onto the second relay device according to the operating environment of the configured recovery model, and verify and monitor the recovery model, so that the second relay device has the ability to execute the recovery model.

[0160] After receiving the decoding model, the second user device deploys the decoding model. Here, the second user device can load the decoding model onto the second user device according to the operating environment of the configured decoding model, and verify and monitor the decoding model, so that the second user device has the ability to execute the decoding model.

[0161] In an embodiment of the present application, when determining that a coding model related to the first task is deployed on the first user device, the first relay device may send a recovery model to the second relay device and / or send a decoding model to the second user device based on the model deployment status of the second relay device and the second user device.

[0162] The first relay device deploys the recovery model to the second relay device so that the second relay device can be deployed with the recovery model. The first relay device deploys the decoding model to the second user device so that the second user device can be deployed with the decoding model.

[0163] The first relay device sends a recovery model deployment instruction to the second relay device, and the second relay device deploys the recovery model on the second relay device in response to the recovery model deployment instruction. The recovery model deployment instruction may include a recovery model. If the recovery model deployment is successful, the first relay device receives a recovery model deployment success message sent by the second relay device; if the recovery model deployment is unsuccessful and the second relay device does not receive a successful recovery model, the first relay device may resend the recovery model to the second relay device; if the recovery model deployment is unsuccessful and the second relay device is unsuitable, the first relay device may update the second relay device.

[0164] The first relay device sends a decoding model deployment instruction to the second user device, and the second user device deploys the decoding model on the second user device in response to the decoding model deployment instruction. The decoding model deployment instruction includes a decoding model. If the decoding model is successfully deployed, the first relay device receives a decoding model deployment success message sent by the second user device; if the decoding model is not successfully deployed, and the second user device does not receive a successful decoding model, the first relay device re-sends the decoding model to the second user device; if the decoding model is not successfully deployed, and the decoding model is not suitable for the second user device, the first relay device re-trains the encoding model, decoding model and recovery model, and sends the encoding model, decoding model and recovery model to the first user device, the second user device and the second relay device respectively.

[0165] In an embodiment of the present application, a recovery model is deployed to the second relay device, and / or a decoding model is deployed to the second user device, so that the second relay device has the ability to execute the recovery model, and / or the second user device has the ability to execute the decoding model.

[0166] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: deploying the recovery model to the second relay device according to the model deployment information, and / or deploying the decoding model to the second user device.

[0167] The model deployment information is used to characterize the model deployment situation of the second relay device and the second user device. In the embodiment of the present application, the first relay device can adopt different model deployment solutions according to different model deployment situations.

[0168] In the embodiment of the present application, the model deployment information represents the model deployment status of the second relay device and the second user device, including one of the following:

[0169] Case 1: The second relay device does not deploy a recovery model;

[0170] Case 2: The second user equipment does not deploy a decoding model;

[0171] Case 3: The second relay device is not deployed with a recovery model, and the second user equipment is not deployed with a decoding model.

[0172] In an embodiment of the present application, different model deployment schemes are determined based on different model deployment information, so that in the communication link, the second user device and the second relay device can successfully deploy corresponding models to support data transmission between the first user device and the second user device.

[0173] In some embodiments, based on Figure 1The communication method shown, the communication method provided by the embodiment of the present application also includes: when the model deployment information indicates that the second relay device has not deployed the recovery model, and the second user device has deployed the decoding model, sending a first model acquisition message to the first user device, and sending a second model acquisition message to the second user device; receiving the encoding model of the first task sent by the first user device in response to the first model acquisition message, and the decoding model sent by the second user device in response to the second model acquisition message; determining the recovery model based on the encoding model and the decoding model; and sending the recovery model to the second relay device.

[0174] Accordingly, based on Figure 2 The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving the recovery model sent by the first relay device.

[0175] Correspondingly, the first user equipment receives the first model acquisition message sent by the first relay device; and sends the coding model for the first task in response to the first model acquisition message to the first relay device.

[0176] Correspondingly, the second user equipment receives the second model acquisition message sent by the first relay device; and sends information about the decoding model in response to the second model acquisition message to the first relay device.

[0177] Here, the first model acquisition message and the second model acquisition message are messages sent to the first user device and the second user device respectively for obtaining corresponding models, wherein the first model acquisition message is used to request the encoding model of the first user device, and the second model acquisition message is used to request the decoding model of the second user device.

[0178] Upon receiving the first model acquisition message, the first user device sends the model parameters of the encoding model for the first task to the first relay device. Upon receiving the second model acquisition message, the second user device sends the model parameters of the decoding model for the first task to the first relay device. The first relay device trains a recovery model based on the model parameters of the encoding model and the model parameters of the decoding model, and sends the trained recovery model to the second relay device.

[0179] In an embodiment of the present application, for situation one, when the first relay device determines based on the model deployment information that the second relay device has not deployed the recovery model and the second user device has deployed the decoding model, the first relay device sends a first model acquisition message to the first user device and sends a second model acquisition message to the second user device; the first user device sends the encoding model of the first task to the first relay device in response to the first model acquisition message; the second user device sends the decoding model of the first task to the first relay device in response to the second model acquisition message; the first relay device trains the recovery model based on the encoding model and the decoding model, and sends the trained recovery model to the second relay device.

[0180] In an embodiment of the present application, when the recovery model is not deployed on the second relay device and the decoding model is deployed on the second user device, the encoding model deployed on the first user device and the decoding model deployed on the second user device are obtained, the recovery model is trained according to the encoding model and the decoding model, and the trained recovery model is sent to the second relay device, so that the recovery model can be successfully deployed on the second relay device.

[0181] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: when the model deployment information indicates that the second relay device is deployed with the recovery model and the second user device is not deployed with the decoding model, sending a first model acquisition message to the first user device and sending a third model acquisition message to the second relay device; receiving the encoding model of the first task sent by the first user device in response to the first model acquisition message, and the recovery model sent by the second relay device in response to the third model acquisition message; determining the decoding model based on the encoding model and the recovery model; and sending the decoding model to the second user device.

[0182] Accordingly, based on Figure 2 The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving a third model acquisition message sent by the first relay device; and sending the recovery model in response to the third model acquisition message to the first relay device.

[0183] Correspondingly, the first user equipment receives the first model acquisition message sent by the first relay device; and sends the coding model for the first task in response to the first model acquisition message to the first relay device.

[0184] Correspondingly, the second user equipment receives the decoding model sent by the first relay device.

[0185] Here, the third model acquisition message is a message for acquiring a recovery model sent to the second relay device, and is used to request the second relay device to send a recovery model for the first task.

[0186] Upon receiving the first model acquisition message, the first user device sends the model parameters of the encoding model for the first task to the first relay device. Upon receiving the third model acquisition message, the second relay device sends the model parameters of the recovery model for the first task to the first relay device. The first relay device trains a decoding model based on the model parameters of the encoding model and the model parameters of the recovery model, and sends the trained decoding model to the second user device.

[0187] In an embodiment of the present application, for situation two, when the first relay device determines based on the model deployment information that the second relay device is deployed with a recovery model and the second user device is not deployed with a decoding model, the first relay device sends a first model acquisition message to the first user device and sends a third model acquisition message to the second relay device; the first user device sends the encoding model of the first task to the first relay device in response to the first model acquisition message; the second relay device sends the recovery model of the first task to the first relay device in response to the third model acquisition message; the first relay device trains the decoding model based on the encoding model and the recovery model, and sends the trained decoding model to the second user device.

[0188] In an embodiment of the present application, when a recovery model is deployed on the second relay device and a decoding model is not deployed on the second user device, the decoding model is trained according to the encoding model and the recovery model by obtaining the encoding model deployed on the first user device and the recovery model deployed on the second relay device, and the trained decoding model is sent to the second user device, so that the decoding model can be successfully deployed on the second user device.

[0189] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: when the model deployment information indicates that the second relay device is not deployed with the recovery model and the second user device is not deployed with the decoding model, sending a first model acquisition message to the first user device; receiving the encoding model of the first task sent by the first user device in response to the first model acquisition message; determining the recovery model and the decoding model based on the encoding model; sending the recovery model to the second relay device, and sending the decoding model to the second user device.

[0190] Accordingly, based on Figure 2 The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving the recovery model sent by the first relay device.

[0191] Correspondingly, the second user equipment receives the decoding model sent by the first relay device.

[0192] Correspondingly, the first user equipment receives the decoding model sent by the first relay device.

[0193] Correspondingly, the first user equipment receives the first model acquisition message sent by the first relay device; and sends the coding model for the first task in response to the first model acquisition message to the first relay device.

[0194] Upon receiving the first model acquisition message, the first user device sends the model parameters of the encoding model for the first task to the first relay device. The first relay device trains a decoding model and a recovery model based on the model parameters of the encoding model, sends the trained recovery model to the second relay device, and sends the trained decoding model to the second user device.

[0195] For situation three, when the first relay device determines based on the model deployment information that the second relay device has not deployed the recovery model and the second user device has not deployed the decoding model, the first relay device sends a first model acquisition message to the first user device; the first user device sends the encoding model for the first task to the first relay device in response to the first model acquisition message; the first relay device trains the decoding model and the recovery model based on the encoding model, and sends the trained decoding model to the second user device, and sends the trained recovery model to the second relay device.

[0196] In an embodiment of the present application, when the recovery model is not deployed on the second relay device and the decoding model is not deployed on the second user device, the encoding model deployed on the first user device is obtained, the decoding model and the recovery model are determined based on the encoding model, and the decoding model is sent to the second user device, so that the decoding model can be successfully deployed on the second user device; and the recovery model is sent to the second relay device, so that the recovery model can be successfully deployed on the second relay device.

[0197] In some embodiments, based on Figure 1 The communication method shown, the communication method provided in an embodiment of the present application also includes: locally searching for the recovery model and the decoding model for the first task; sending the found recovery model to the second relay device; and sending the found decoding model to the second user device.

[0198] In an embodiment of the present application, upon receiving a service request, the first relay device may search for a recovery model and a decoding model locally; if the recovery model and the decoding model are found, the recovery model and the decoding model may be sent to the second relay device and the second user device respectively; if the recovery model and the decoding model are not found, the recovery model may be deployed to the second relay device according to the model deployment information, and / or the decoding model may be deployed to the second user device.

[0199] The first relay device locally searches for a recovery model and a decoding model. If a recovery model for the first task is found, the first relay device sends the found recovery model to the second relay device. If a decoding model for the first task is found, the first relay device sends the found decoding model to the second user device. This reduces signaling interactions between the first relay device, the second relay device, and the second user device, thereby lowering signaling overhead.

[0200] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: determining the encoding model, the recovery model and the decoding model for the first task according to the model requirement information of the first task included in the service request; sending the encoding model to the first user device, sending the recovery model to the second relay device, and sending the decoding model to the second user device.

[0201] Accordingly, based on Figure 2 The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving the recovery model sent by the first relay device.

[0202] Correspondingly, the first user equipment receives the coding model for the first task sent by the first relay device.

[0203] Correspondingly, the second user equipment receives the decoding model sent by the first relay device.

[0204] Here, the model requirement information may include information representing the model requirements, such as the model type and the performance requirements of the model.

[0205] In an embodiment of the present application, the first relay device may determine the encoding model, the recovery model and the decoding model for the first task based on the model requirement information of the first task included in the service request when it is determined that the first user device has not deployed the encoding model.

[0206] The first relay device may first search for an encoding model, a recovery model, and a decoding model that meet the model type requirements and the performance requirements of the first task on the model based on the model requirement information of the first task included in the service request.

[0207] In an embodiment of the present application, the first relay device can directly train the encoding model, recovery model and decoding model that meet the model requirements, send the trained encoding model to the first user device, send the trained recovery model to the second relay device, and send the trained decoding model to the second user device.

[0208] In some embodiments, the first relay device can locally search for the encoding model, recovery model, and decoding model based on the model requirement information of the first task. When the encoding model, recovery model, and decoding model are found, the first relay device sends the found encoding model to the first user device, sends the found recovery model to the second relay device, and sends the found decoding model to the second user device. If the encoding model, recovery model, and decoding model that meet the model requirements are not found, the encoding model, recovery model, and decoding model that meet the model requirements are trained, and the trained encoding model is sent to the first user device, the trained recovery model is sent to the second relay device, and the trained decoding model is sent to the second user device.

[0209] In an embodiment of the present application, the encoding model, recovery model and decoding model that meet the model requirements corresponding to the first task are determined based on the model requirement information of the first task included in the service request; the encoding model is sent to the first user device, the recovery model is sent to the second relay device, and the decoding model is sent to the second user device, so as to respectively deploy the encoding model on the first user device, the recovery model on the second relay device and the decoding model on the second user device.

[0210] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: sending a third confirmation request to the first user equipment, the third confirmation request being used to confirm whether the first user equipment is deployed with the coding model; receiving a third model confirmation response sent by the first user equipment confirming that the coding model is deployed.

[0211] Correspondingly, the first user equipment receives the third confirmation request sent by the first relay device; and sends a third model confirmation response to the first relay device, confirming that the coding model is deployed.

[0212] In an embodiment of the present application, the first relay device sends a third confirmation request to the first user device; the first user responds to the third confirmation request, and when it determines that it has successfully deployed the coding model, sends a third model confirmation response to the first relay device; when the first relay device receives the third model confirmation response, it confirms that the first user device has deployed the coding model.

[0213] When the first relay device confirms that the first user equipment has deployed the encoding model, the second relay device has deployed the recovery model, and the second user equipment has deployed the decoding model, a model deployment confirmation message is sent to the first user equipment.

[0214] In some embodiments, based on Figure 1 The communication method shown, the communication method provided by the embodiment of the present application also includes: determining whether the first user device is deployed with a coding model for the first task according to the service request; wherein, when it is determined that the first user device is deployed with the coding model, the recovery model is deployed to the second relay device, and / or, the decoding model is deployed to the second user device; and / or, when it is determined that the first user device is not deployed with the coding model, the coding model, the recovery model and the decoding model are determined according to the model requirement information of the first task included in the service request.

[0215] In an embodiment of the present application, the first relay device determines whether the first user device is deployed with a coding model based on a service request; when the first user device is deployed with a coding model, the second relay device is not deployed with a recovery model, and the second user device is deployed with a decoding model, the first relay device deploys the recovery model to the second relay device; when the first user device is deployed with a coding model, the second relay device is deployed with a recovery model, and the second user device is not deployed with a decoding model, the first relay device deploys a decoding model to the second user device; when the first user device is deployed with a coding model, the second relay device is not deployed with a recovery model, and the second user device is not deployed with a decoding model, the first relay device deploys a recovery model to the second relay device and the first relay device deploys a decoding model to the second user device.

[0216] In a case where the first user equipment does not deploy a coding model, the first relay device determines a coding model, a recovery model, and a decoding model according to model requirement information of the first task included in the service request.

[0217] In an embodiment of the present application, comprehensive consideration is given to whether the encoding model, recovery model and decoding model are respectively deployed on the first user device, the second relay device and the second user device, so that the model requirements corresponding to the first task can be met under different conditions in the communication link to achieve data transmission.

[0218] In some embodiments, based on Figure 2The communication method shown, the communication method provided by the embodiment of the present application also includes: receiving a first codebook index sent by the first user equipment, the first codebook index is the first coded data quantized based on the codebook, and the first codebook index is obtained by semantically encoding the data of the first task by the coding model; based on the recovery model, restoring the first codebook index to a second codebook index; sending the second codebook index to the second user equipment; the second coding index is used to restore the second coded data based on the codebook, and the second coded data is decoded by the decoding model to obtain the data of the first task.

[0219] Correspondingly, the first user equipment sends the first codebook index.

[0220] Correspondingly, the second user equipment receives the second codebook index sent by the second relay device.

[0221] In the embodiment of the present application, the codebook can be stored in a shared knowledge base, including codebook indexes and codebook vectors, with a mapping relationship between the codebook indexes and the codebook vectors. The originally trained codebook is an unordered codebook, and adjacent codebook vectors lack correlation. Therefore, a reordering algorithm is used to reorder the codebook to obtain a codebook with correlation between adjacent codebooks.

[0222] The data of the first task may be data to be transmitted by the first user equipment. In one example, the data of the first task may be image data.

[0223] In an embodiment of the present application, the first user device obtains semantic features by extracting semantic features from the data of the first task, and quantizes the semantic features into codebook vectors through a codebook-based vector quantization technology, and represents the mapping relationship between the semantic features and the codebook vectors through codebook indexes.

[0224] At the transmitting end, the first user device uses the coding model to perform semantic extraction on the data of the first task to obtain first coded data with semantic features; and quantizes the first coded data into a codebook vector based on the rearranged codebook, and determines the first codebook index corresponding to the codebook vector based on the mapping relationship between the codebook vector and the codebook index. The first user device modulates the first codebook index and sends it to the second relay device. In one example, Figure 4 As shown, the data of the first task to be sent by the first user device is image X. The image X is mapped from the original color space to the semantic latent space through the potential transformation, and the semantic feature obtained is L=LE(X;α), where LE(·;α) is a semantic encoder with a learnable parameter α. M represents the number of feature vectors, and N represents the dimension of the feature vector. The codebook is represented as K is the number of codebook vectors, and N is the codebook vector dimension. Through vector quantization operation, based on the codebook (shared semantic knowledge base), each semantic feature vector l m quantized into a codebook vector e k and mapped to the first codebook index k of the codebook vector, where m∈[1,M],k∈[1,K]. The vector quantization operation is expressed as k=VQ(L;E), so the first user equipment only transmits a corresponding set of codebook indices There is no need to transmit the complete semantic features. Then, after channel modulation, the transmission signal is obtained and sent to the second relay device.

[0225] In the embodiment of the present application, the first user equipment restores the first codebook index k to the semantic feature L according to the codebook. vq .

[0226] In the embodiment of the present application, at the relay end, the second relay device receives the first codebook index, uses a recovery model to restore the first codebook index to a second codebook index, and sends the second codebook index to the second user equipment.

[0227] The second relay device receives the first codebook index, calculates the transition probability corresponding to the first codebook index, obtains the confidence of the first codebook index, and corrects the first codebook index with low confidence to obtain a more accurate second codebook index. Figure 4 As shown, the first codebook index received by the second relay device is Estimate the posterior distribution through the semantic restorer (i.e., recovery model) at the relay end β is a learnable parameter of the semantic restorer, E is a shared semantic codebook, and the posterior distribution is used as the confidence of the first codebook index. According to the confidence of the first codebook index, the first codebook index is restored to the second codebook index through maximum a posteriori decoding, which is expressed as This avoids restoring a first codebook index with low confidence when restoring the first codebook index.

[0228] In this embodiment of the present application, based on the semantic codebook, a recovery model can be used to directly correct decision errors in the modulated signal indexed by the first codebook, without the need to fully decode the semantic information, thereby reducing the computational complexity of the second relay device. Furthermore, due to the influence of channel noise introduced during the training process, the encoding model of the first user device, the recovery model of the second relay device, and the decoding model of the second user device can better adapt to more severe channel environments, thereby improving the system's performance in low signal-to-noise ratio environments.

[0229] In the embodiment of the present application, at the receiving end, the second user equipment receives the second codebook index, decodes the second codebook index using the decoding model to obtain second encoded data, and decodes the second encoded data using the decoding model to obtain data related to the first task. In one example, Figure 4 As shown, the second codebook index received by the second user equipment is expressed as The second codebook index is restored to the quantized semantic features according to the codebook of the shared knowledge base (i.e. the second encoded data), using the decoding model to decode the semantic features Decode and get the decoded image data Here, LD(·;γ,E) is a semantic decoder with learnable parameters γ, and the codebook E assists semantic decoding.

[0230] It can be understood that the first user device uses the encoding model to semantically encode the first coded data to obtain a first codebook index, and sends the first codebook index to the second relay device; the second relay device uses the recovery model to recover the first codebook index to obtain a second codebook index, and the second relay device sends the second codebook index to the second user device; the second user device uses the decoding model to restore the second codebook index to obtain second coded data, and the second coded data is the decoded data of the first task.

[0231] In this embodiment of the present application, a second relay device recovers the first codebook index sent by the first user device as a second codebook index based on a recovery model, and sends the second codebook index to the second user device. The first codebook index is obtained by quantizing the first coded data, and the second codebook index is the second coded data restored based on the codebook. In this way, by transmitting the codebook index and recovering and decoding the codebook index using the recovery model and decoding model at the relay and receiving ends to achieve data transmission, the amount of transmitted data is significantly increased, thereby improving communication efficiency.

[0232] In some embodiments, based on Figure 2 The communication method shown, the communication method provided in the embodiment of the present application also includes: receiving a first transmission request sent by the first user device; sending a second transmission request to the second user device based on the triggering of the first transmission request; receiving a first transmission confirmation message sent by the second user device in response to the second transmission request; based on the triggering of the first transmission confirmation message, sending a second transmission confirmation message to the first user device, the second transmission message indicating that the second relay device and the second user device meet the data transmission conditions.

[0233] In an embodiment of the present application, upon confirming that the first user device has deployed the encoding model, the second relay model has deployed the recovery model, and the second user device has deployed the decoding model, the first user device sends a first transmission request to the second relay device, thereby entering the data transmission phase. In an embodiment of the present application, upon receiving a model deployment confirmation message, the first user device may confirm that the first user device has deployed the encoding model, the second relay model has deployed the recovery model, and the second user device has deployed the decoding model.

[0234] Upon receiving the first transmission request, the second relay device triggers a second transmission request to the second user device based on the first transmission request. In response to the second transmission request, upon confirming that the data transmission conditions are met, the second user device sends a first transmission confirmation message to the second relay device. In response to the first transmission confirmation message, the second relay device triggers a second transmission confirmation message to the first user device. Upon receiving the second transmission confirmation message, the first user device confirms that the data transmission conditions are met between the second relay device and the second user device.

[0235] In the embodiment of the present application, the second relay device responds to the first transmission request sent by the first user device through a confirmation message, if it confirms that the second relay device and the second user device both meet the transmission conditions. In this way, data transmission is performed when all nodes in the communication system meet the transmission conditions, thereby improving the accuracy of data transmission.

[0236] The following describes the application of the embodiments of the present application in actual scenarios.

[0237] Semantic communication is a new communications paradigm. By integrating artificial intelligence with communications and organically mapping task requirements with information transmission, it can significantly improve communication efficiency and enhance the user experience. Traditional communication systems focus on metrics unrelated to data content, such as channel capacity, bit error rate, and outage probability. Semantic communication, on the other hand, focuses on the content and meaning of data and customizes performance metrics for different data types, such as sentence similarity for text, peak signal-to-noise ratio and structural similarity for images, and distortion rate for voice data. Semantic communication leverages the deep integration of intelligence with communications, networking, and other technologies. Through semantic feature extraction, semantic information transmission, and semantic information recovery, it enables efficient and accurate information transmission and precise control.

[0238] With the tremendous success of artificial intelligence (AI) technology, especially deep learning, in multiple fields such as computer vision, natural language processing, and speech recognition, semantic communication utilizes semantic encoders based on deep neural networks to effectively extract and compress semantic information. However, because semantic encoders typically directly output continuously distributed signals, most existing semantic communication system designs use analog modulation, which makes it difficult to deploy semantic communication systems on modern digital communication equipment. To make semantic communication compatible with digital communication systems, semantic features must be converted into bits for transmission. Codebook-based semantic communication technology provides a feasible solution by designing a codebook as a shared knowledge base and exploring the effective mapping relationship between codebook sequences and semantic feature vectors. However, existing codebook-based semantic communication solutions all separate source semantic coding from channel coding, which cannot effectively combat the effects of channel fading.

[0239] At the same time, although semantic communication greatly improves communication efficiency and robustness, in actual communication environments, due to path loss and obstacles in the wireless channel, the signal fades severely, resulting in low reliability and severe distortion of end-to-end semantic information transmission, which is difficult to meet communication needs. It is often necessary to rely on relay communication to improve the reliability of information transmission. Therefore, when designing a semantic relay communication system, the semantic information processing problem at the relay end needs to be considered.

[0240] Based on the above description, the embodiment of the present application proposes a codebook-based semantic relay scheme for semantic communication, which efficiently compresses the source information based on the codebook and encodes it into a digital signal. At the same time, it proposes a relay semantic recovery mechanism for the codebook sequence. By jointly designing a codebook-based semantic codec and a relay recoverer through the joint source channel, efficient relay transmission of digital semantic signals is achieved.

[0241] The explicit image semantic information is encoded into the latent semantic space through a neural network-based autoencoder, and then the features are mapped to indexes based on vector quantization technology. They are then converted into bits and transmitted through a digital communication system. A neural network-based semantic index restorer is deployed at the relay end for the codebook index information, achieving efficient compression of image information and compatibility with digital communication systems, and effectively improving the performance and robustness of semantic relay communication.

[0242] The neural network-based semantic restorer needs to match the semantic encoding and decoding model of the transceiver and conduct communication interaction between the transceiver and the relay system based on the deployment of the model at the transceiver.

[0243] The specific process of the codebook-based semantic relay solution for semantic communication proposed in the embodiment of the present application can be as follows: Figure 5As shown, it includes the following steps S501 to S503: Step S501: service request confirmation; Step S502: model matching; Step S503: information transmission.

[0244] The specific process of service confirmation request and model matching can be as follows: Figure 6 As shown, the process includes the following steps S601 to S620:

[0245] Step S601: User 1 sends a service request;

[0246] Here, user 1 corresponds to the first user equipment in the aforementioned embodiment.

[0247] The relay network includes multiple relay nodes (corresponding to the second relay device in the aforementioned embodiment) and some control nodes for network resource management, control or service (corresponding to the first relay device in the aforementioned embodiment).

[0248] User 1 sends a service request to the control node in the relay network. The service request includes a semantic communication scheduling request, whether the user has a semantic extraction model, and the semantic extraction model parameters.

[0249] Step S602: The control node confirms the relay link;

[0250] The relay network includes multiple relay nodes. The control node selects the relay node with the best channel conditions and the smallest load from multiple relay nodes as a feasible relay node based on the service request. In the subsequent data transmission process, the selected feasible relay node communicates with the user.

[0251] Step S603: Does user 1 have a semantic extraction model? If yes, proceed to step S608; if no, proceed to step S604;

[0252] Here, the semantic extraction model corresponds to the encoding model in the aforementioned embodiment.

[0253] Based on the service request, the control node obtains the model deployment status of user 1. There are two cases: user 1 has deployed a semantic extraction model and user 1 has not deployed a semantic extraction model.

[0254] Step S604: the control node searches for a task requirement model;

[0255] When user 1 has not deployed a semantic extraction model, the control node searches for the corresponding semantic extraction model, decoding model, and recovery model based on the task requirements included in the service request, such as the transmission image size and performance requirements.

[0256] Step S605: Is the task requirement model found? If so, proceed to step S607; if not, proceed to step S606;

[0257] Step S606: The control node trains the model and proceeds to step S607;

[0258] If the corresponding semantic extraction model, decoding model, and recovery model are not found, the control node trains the corresponding semantic extraction model, decoding model, and recovery model according to the service request.

[0259] Step S607: Send the task requirement model to user 1, the relay node, and user 2 respectively; and proceed to step S619;

[0260] Step S608: The control node searches for a model that matches user 1;

[0261] In the case that a semantic extraction model already exists for user 1, the control node searches for a model that matches user 1. Here, the model that matches user 1 includes a recovery model and a decoding model.

[0262] Step S609: Is a model matching user 1 found? If so, proceed to step S6010; if not, proceed to step S611;

[0263] Step S610: The control node sends the found model to user 2 or the relay node;

[0264] Step S611: Determine the model matching situation; proceed to one of steps S612 to S615;

[0265] Step S612: Match the models of user 1, user 2, and the relay node, and proceed to step S619;

[0266] If both the relay node and user 2 have models matching user 1, user 1 sends a transmission request. The control node sends confirmation requests to the relay node and user 2, and the relay node and user 2 each reply to the control node to confirm the transmission instruction. After confirming that the relay node and user 2 have deployed matching models, the control node sends a confirmation instruction to user 1. After user 1 receives the confirmation instruction from the control node, the data transmission phase begins.

[0267] Step S613: Model matching between user 1 and relay node; proceed to step S616;

[0268] If user 2 doesn't have a model that matches user 1, but the relay node does, the control node sends a model acquisition request to user 1. User 1 sends the semantic extraction model to the control node, which then trains a decoding model based on the existing model. Once training is complete, the control node transmits the decoding model to user 2.

[0269] Step S614: Match the models of user 1 and user 2; proceed to step S617;

[0270] If the relay node does not have a model that matches user 1, but user 2 does, the control node sends a model acquisition request to both users 1 and 2. Users 1 and 2 send their respective models to the control node, which then trains the relay restorer (i.e., the recovery model) based on the existing model. After training, the control node sends the recovery model to the relay node.

[0271] Step S615: The models of user 1, user 2, and the relay node do not match; proceed to step S618;

[0272] If neither the relay node nor user 2 has a model matching user 1, the control node sends a model acquisition request to user 1. User 1 sends the model to the control node, which then trains a decoding model and a recovery model based on the existing models. After training, the control node transmits the decoding model and the recovery model to user 2 and the relay node, respectively.

[0273] Step S616: The control node trains the model and sends it to user 2;

[0274] Step S617: The control node trains the model and sends it to the relay node;

[0275] Step S618: The control node trains the model and sends it to user 1, user 2, and the relay node;

[0276] Step S619: deploying the model;

[0277] Step S620: Start transmission.

[0278] In the embodiment of the present application, the interaction between the control node, the relay node, user 1 and user 2 in the communication system is as follows: Figure 7 As shown, the following steps S701 to S710 may be included:

[0279] Step S701: User 1 sends a service request to the control node;

[0280] Step S702: The control node sends a model confirmation request to the relay node;

[0281] Step S703: The control node sends a model confirmation request to user 2;

[0282] Step S704: The relay node sends a model confirmation instruction to the control node;

[0283] Step S705: User 2 sends a model confirmation instruction to the control node;

[0284] Step S706: The relay node sends a model confirmation instruction to the control node;

[0285] Step S707: User 1 sends a transmission request to the relay node;

[0286] Step S708: The relay node sends a transmission confirmation request to user 2;

[0287] Step S709: User 2 sends a transmission confirmation instruction to the relay node;

[0288] Step S710: The relay node sends a transmission confirmation instruction to user 1.

[0289] The specific implementation process of the semantic relay transmission system is as follows: Figure 8 As shown, the specific information transmission is as follows:

[0290] At the transmitting end 801 (corresponding to the first user equipment in the aforementioned embodiment):

[0291] (1) Through the semantic extraction model, the image information to be transmitted (corresponding to the data of the first task in the aforementioned embodiment) is converted into semantic features (corresponding to the first encoded data in the aforementioned embodiment) and mapped to the latent space.

[0292] (2) Through codebook-based vector quantization technology, the semantic feature vector is quantized into a codebook vector, and the mapping relationship is represented by the codebook index.

[0293] (3) Modulate the transmitted codebook index signal.

[0294] The specific description is as follows:

[0295] like Figure 4 As shown, the transmitted image at the sender is denoted as X, and the original image is mapped from the original color space to the semantic latent space through the latent transformation, denoted as L = LE(X; α), where LE(·; α) is a semantic encoder with a learnable parameter α. M represents the number of feature vectors, and N represents the dimension of the feature vector. The codebook is represented as K represents the number of codebook vectors, and N represents the codebook vector dimension.

[0296] The original training codebook is an unordered codebook, and there is a lack of correlation between adjacent codebook vectors. Therefore, the rearrangement algorithm is used to realize codebook rearrangement. The rearranged codebook is expressed as Where Recom(·) represents the codebook rearrangement algorithm. The codebook rearrangement algorithm process is as follows Figure 9 As shown, e0 is initialized to l m The mean of , i=1, may include the following steps S901 to S906:

[0297] Step S901: Initialization;

[0298] Step S902: Find the distance e in E i-1 Recent e k ;

[0299] Step S903: Delete e from E k ;

[0300] Step S904: i =e k , i=i+1;

[0301] Step S905: Is E empty? If so, proceed to step S906; if not, proceed to step S902;

[0302] Step S906: Output E.

[0303] Through vector quantization operation, with the help of codebook (shared semantic knowledge base), each feature vector l in the latent semantic information m quantized into a codebook vector e k And mapped to the index k of the codebook vector (corresponding to the first codebook index in the aforementioned embodiment), where m∈[1,M],k∈[1,K]. The vector quantization operation is expressed as k=VQ(L;E), so the transmitter only needs to transmit a set of corresponding codebook indices. There is no need to transmit the complete semantic features. Then, after channel modulation, the transmission signal is obtained for transmission.

[0304] At the relay end 803 (corresponding to the second relay device in the aforementioned embodiment):

[0305] (1) The confidence of the received index signal is obtained by calculating the transition probability corresponding to each codebook vector of each received codebook index signal.

[0306] (2) Correct low-confidence indexes.

[0307] The specific description is as follows:

[0308] like Figure 4 As shown, at the relay end, the obtained codebook index is expressed as Estimate the posterior distribution through the semantic restorer at the relay end β is a learnable parameter of the relay semantic restorer, E is the shared semantic codebook, and this posterior distribution is used as the confidence of the received index.

[0309] According to the obtained posterior distribution (confidence), we recover the source index information through maximum a posteriori decoding, which is expressed as This avoids restoring to a low-confidence index when restoring the index.

[0310] By using the above method, the codebook index error generated during the transmission process is corrected and the correct codebook index sequence is recovered, which is expressed as Where R(·; β, E) is a relay semantic restorer with learnable parameters β, and E is a shared semantic codebook. Based on the semantic codebook, the recovery model can directly correct errors in the judgment of the codebook-indexed modulated signal without fully decoding the semantic information, reducing the computational complexity at the relay end. Furthermore, due to the introduction of channel noise during training, the semantic codec at the transceiver end and the relay semantic restorer at the relay end can better adapt to harsher channel environments, improving system performance in low signal-to-noise ratio environments.

[0311] At the receiving end 802 (corresponding to the second user equipment in the aforementioned embodiment):

[0312] (1) Demodulate the received signal.

[0313] (2) The received signal is restored to image information through the corresponding semantic decoder.

[0314] The specific description is as follows:

[0315] like Figure 4 As shown, after the relay end recovers the codebook index, the receiving end obtains the final codebook index Restore to quantized semantic features based on the shared semantic codebook The final image data is recovered by the corresponding semantic decoder and expressed as where LD(·;γ,E) is a semantic decoder with learnable parameters γ, and the codebook E assists semantic decoding.

[0316] In the embodiment of the present application, the semantic coding of the joint source channel and the codebook design enhance the semantic association between codebook indexes, and the semantic index restorer for codebook index information deployed at the relay end further improves the performance and robustness of the relay transmission system.

[0317] In a fourth aspect, in order to implement the above-mentioned wireless communication method, a device 1000 (a first relay device, a second relay device, a first user device or a second user device) in an embodiment of the present application is as follows: Figure 10As shown, the system may include at least one processor 1001 and at least one transceiver 1002 coupled to the at least one processor 1001. The transceiver 1002 may include at least one separate receive circuit system and transmit circuit system, or at least one integrated receive circuit system and transmit circuit system. The at least one processor 1001 may be a central processing unit (CPU), a microprocessor unit (MPU), a digital signal processor (DSP), or a field-programmable gate array (FPGA).

[0318] According to some embodiments of the present application, when the device 1000 is a first relay device, the first relay device includes a first transceiver; and

[0319] A first processor is coupled to the first transceiver; the first processor is configured to:

[0320] receiving, via the first transceiver, a service request for a first task sent by a first user equipment;

[0321] determining a second relay device and a second user equipment according to the task information of the first task included in the service request;

[0322] Sending a first confirmation request to the second relay device via the first transceiver, where the first confirmation request is used to confirm whether the second relay device has deployed a recovery model related to the first task;

[0323] A second model confirmation request is sent to the second user equipment via the first transceiver, where the second confirmation request is used to confirm whether the second user equipment is deployed with a decoding model related to the first task.

[0324] In some embodiments the first processor is configured to:

[0325] receiving, via the first transceiver, a first model confirmation response sent by the second relay device, confirming that the recovery model is deployed;

[0326] receiving, via the first transceiver, a second model confirmation response sent by the second user equipment, confirming that the decoding model is deployed;

[0327] Sending a model deployment confirmation message to the first user equipment via the first transceiver.

[0328] In some embodiments, the first processor is configured to:

[0329] receiving, via the first transceiver, a third model confirmation response sent by the second relay device, confirming that the recovery model is not deployed;

[0330] The recovery model is sent to the second relay device via the first transceiver; or the second relay device is updated.

[0331] In some embodiments, the first processor is configured to:

[0332] receiving, via the first transceiver, a fourth model confirmation response sent by the second user equipment, confirming that the decoding model is not deployed;

[0333] The decoding model is sent to the second user equipment via the first transceiver.

[0334] In some embodiments, the first processor is configured to:

[0335] The recovery model is deployed to the second relay device, and / or the decoding model is deployed to the second user equipment.

[0336] In some embodiments, the first processor is configured to:

[0337] According to the model deployment information, the recovery model is deployed to the second relay device, and / or the decoding model is deployed to the second user equipment.

[0338] In some embodiments, the first processor is configured to:

[0339] Sending, via the first transceiver, a first model acquisition message to the first user equipment and a second model acquisition message to the second user equipment when the model deployment information indicates that the second relay device has not deployed the recovery model and the second user equipment has deployed the decoding model;

[0340] Receiving, via the first transceiver, the encoding model for the first task sent by the first user equipment in response to the first model acquisition message, and the decoding model sent by the second user equipment in response to the second model acquisition message;

[0341] Determining the recovery model according to the encoding model and the decoding model;

[0342] The recovery model is sent to the second relay device via the first transceiver.

[0343] In some embodiments, the first processor is configured to:

[0344] Sending, via the first transceiver, a first model acquisition message to the first user equipment and a third model acquisition message to the second relay device when the model deployment information indicates that the second relay device is deployed with the recovery model and the second user equipment is not deployed with the decoding model;

[0345] Receiving, via the first transceiver, the coding model for the first task sent by the first user equipment in response to the first model acquisition message, and the recovery model sent by the second relay device in response to the third model acquisition message;

[0346] Determining the decoding model according to the encoding model and the recovery model;

[0347] The decoding model is sent to the second user equipment via the first transceiver.

[0348] In some embodiments, the first processor is configured to:

[0349] Sending, via the first transceiver, a first model acquisition message to the first user equipment when the model deployment information indicates that the second relay device is not deployed with the recovery model and the second user equipment is not deployed with the decoding model;

[0350] Receiving, via the first transceiver, the coding model for the first task in response to the first model acquisition message sent by the first user equipment;

[0351] Determining the recovery model and the decoding model according to the encoding model;

[0352] The recovery model is sent to the second relay device via the first transceiver, and the decoding model is sent to the second user equipment.

[0353] In some embodiments, the first processor is configured to:

[0354] locally searching for the recovery model and the decoding model for the first task;

[0355] sending the found recovery model to the second relay device via the first transceiver;

[0356] The found decoding model is sent to the second user equipment via the first transceiver.

[0357] In some embodiments, the first processor is configured to:

[0358] determining, according to the model requirement information of the first task included in the service request, the encoding model, the recovery model, and the decoding model for the first task;

[0359] The encoding model is sent to the first user equipment, the recovery model is sent to the second relay device, and the decoding model is sent to the second user equipment via the first transceiver.

[0360] In some embodiments, the first processor is configured to:

[0361] Sending, via the first transceiver, a third confirmation request to the first user equipment, where the third confirmation request is used to confirm whether the first user equipment is deployed with the coding model;

[0362] A third model confirmation response is received via the first transceiver and is sent by the first user equipment, confirming that the coding model is deployed.

[0363] In some embodiments, the first processor is configured to:

[0364] Determining, according to the service request, whether the first user equipment is deployed with a coding model related to the first task;

[0365] Wherein, when it is determined that the encoding model is deployed on the first user equipment, the recovery model is deployed to the second relay device, and / or the decoding model is deployed to the second user equipment; and / or,

[0366] In a case where it is determined that the first user equipment is not deployed with the encoding model, the encoding model, the recovery model, and the decoding model are determined according to the model requirement information of the first task included in the service request.

[0367] a second processor coupled to the second transceiver; the second processor configured to:

[0368] Sending, via the second transceiver, synchronization signal transmission configuration information to the first device in a first frequency band, where the first frequency band is lower than a millimeter wave frequency band, the second device and at least one of the first devices forming a domain, and the second device being configured to manage communication of the at least one device within the domain;

[0369] The synchronization signal is transmitted via the second transceiver in the millimeter wave frequency band according to the synchronization signal transmission configuration information.

[0370] In some embodiments, the second processor is configured to:

[0371] determining whether to activate a millimeter wave transmission function in the first device according to a service transmission requirement and / or an energy saving requirement of the first device;

[0372] Second indication information is sent to the first device via the second transceiver in the first frequency band, where the second indication information is used to determine whether to activate the millimeter wave transmission function.

[0373] When the device 1000 is a second relay device, the second relay device includes a second transceiver; and

[0374] a second processor coupled to the second transceiver; the second processor configured to:

[0375] A first confirmation request sent by the first relay device upon receiving a service request sent by the first user equipment is received via the second transceiver, wherein the first confirmation request is used to confirm whether the second relay device has deployed a recovery model for the first task.

[0376] In some embodiments, the second processor is configured to:

[0377] Receiving, via the second transceiver, a first codebook index sent by the first user equipment, where the first codebook index is obtained by quantizing first coded data based on a codebook, and the first codebook index is obtained by semantically encoding data of the first task by a coding model;

[0378] Restoring the first codebook index to a second codebook index based on the restoration model;

[0379] The second codebook index is sent to the second user equipment via the second transceiver; the second codebook index is used to restore the second coded data based on the codebook, and the second coded data is decoded by the decoding model to obtain data of the first task.

[0380] In some embodiments, the second processor is configured to:

[0381] receiving, via the second transceiver, a first transmission request sent by the first user equipment;

[0382] sending, via the second transceiver, a second transmission request to the second user equipment based on the triggering of the first transmission request;

[0383] receiving, via the second transceiver, a first transmission confirmation message sent by the second user equipment in response to the second transmission request;

[0384] A second transmission confirmation message is sent to the first user equipment via the second transceiver based on the triggering of the first transmission confirmation message, where the second transmission message indicates that the second relay device and the second user equipment meet a data transmission condition.

[0385] The description of the above device embodiment is similar to the description of the above method embodiment and has similar beneficial effects as the method embodiment. For technical details not disclosed in the device embodiment of this application, please refer to the description of the method embodiment of this application for understanding.

[0386] It should be noted that in the embodiment of the present application, if the above-mentioned wireless communication method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods of each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk. In this way, the embodiment of the present application is not limited to any specific combination of hardware and software.

[0387] In the fifth aspect, in order to implement the above-mentioned wireless communication method, an embodiment of the present application provides an electronic device, including a memory and a processor, the memory stores a computer program that can be run on the processor, and when the processor executes the program, it implements the steps in the wireless communication method provided in the above-mentioned embodiment.

[0388] In a sixth aspect, an embodiment of the present application provides a storage medium, that is, a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the wireless communication method provided in the above embodiment are implemented.

[0389] It should be noted that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the description of the method embodiments of this application for understanding.

[0390] It should be understood that “one embodiment” or “an embodiment” mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, “in one embodiment” or “in some embodiments” appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are for description only and do not represent the advantages and disadvantages of the embodiments.

[0391] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0392] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0393] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0394] In addition, all functional units in the embodiments of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the above-mentioned integrated units can be implemented in the form of hardware or in the form of hardware plus software functional units.

[0395] A person skilled in the art will understand that all or part of the steps of the above-mentioned method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, and other media that can store program codes.

[0396] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods of each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0397] The above are merely embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A first relay device, comprising a first transceiver, and a first processor coupled to the first transceiver; the first processor being configured to: receiving, via the first transceiver, a service request for a first task sent by a first user equipment; determining a second relay device and a second user equipment according to the task information of the first task included in the service request; Sending a first confirmation request to the second relay device via the first transceiver, where the first confirmation request is used to confirm whether the second relay device has deployed a recovery model related to the first task; A second model confirmation request is sent to the second user equipment via the first transceiver, where the second confirmation request is used to confirm whether the second user equipment is deployed with a decoding model related to the first task.

2. The first relay device according to claim 1, wherein the first processor is configured to: receiving, via the first transceiver, a first model confirmation response sent by the second relay device, confirming that the recovery model is deployed; receiving, via the first transceiver, a second model confirmation response sent by the second user equipment, confirming that the decoding model is deployed; Sending a model deployment confirmation message to the first user equipment via the first transceiver.

3. The first relay device according to claim 1, wherein the first processor is configured to: receiving, via the first transceiver, a third model confirmation response sent by the second relay device, confirming that the recovery model is not deployed; The recovery model is sent to the second relay device via the first transceiver; or the second relay device is updated.

4. The first relay device according to claim 1, wherein the first processor is configured to: receiving, via the first transceiver, a fourth model confirmation response sent by the second user equipment, confirming that the decoding model is not deployed; The decoding model is sent to the second user equipment via the first transceiver.

5. The first relay device according to claim 1, wherein the first processor is configured to: The recovery model is deployed to the second relay device, and / or the decoding model is deployed to the second user equipment.

6. The first relay device according to claim 1, wherein the first processor is configured to: determining, according to the model requirement information of the first task included in the service request, the encoding model, the recovery model, and the decoding model for the first task; The encoding model is sent to the first user equipment, the recovery model is sent to the second relay device, and the decoding model is sent to the second user equipment via the first transceiver.

7. The first relay device according to claim 6, wherein the first processor is configured to: Sending, via the first transceiver, a third confirmation request to the first user equipment, where the third confirmation request is used to confirm whether the first user equipment is deployed with the coding model; A third model confirmation response is received via the first transceiver and is sent by the first user equipment, confirming that the coding model is deployed.

8. The first relay device according to any one of claims 1 to 7, wherein the first processor is configured to: Determining, according to the service request, whether the first user equipment is deployed with a coding model related to the first task; in, In the case of determining that the encoding model is deployed on the first user equipment, deploying the recovery model to the second relay device, and / or deploying the decoding model to the second user equipment; and / or, In a case where it is determined that the first user equipment is not deployed with the encoding model, the encoding model, the recovery model, and the decoding model are determined according to the model requirement information of the first task included in the service request.

9. A second relay device, comprising a second transceiver, and a second processor coupled to the second transceiver; the second processor being configured to: A first confirmation request sent by the first relay device upon receiving a service request sent by the first user equipment is received via the second transceiver, wherein the first confirmation request is used to confirm whether the second relay device has deployed a recovery model for the first task.

10. A communication method, applied to a first relay device, comprising: receiving a service request for a first task sent by a first user device; determining a second relay device and a second user equipment according to the task information of the first task included in the service request; Sending a first confirmation request to the second relay device, where the first confirmation request is used to confirm whether the second relay device has deployed a recovery model related to the first task; A second model confirmation request is sent to the second user equipment, where the second confirmation request is used to confirm whether the second user equipment is deployed with a decoding model related to the first task.

Images