Communication method and device

Abstract

The invention discloses a communication method and device. The method comprises the following steps: a first device receives first indication information and first information; wherein the first indication information is used for indicating to activate the first function module; the first information is used for determining that the target AI model to be activated comprises N AI models, each AI model in the N AI models is used for realizing the function of the first function module, and N is a positive integer. The first device may activate the N AI models according to the first information. Through the method, the first device can select the proper N AI models to realize the function of the first function module according to the first information, so that the performance of realizing the function of the first function module can be improved.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a communication method and device. Background Art

[0002] Artificial intelligence (AI) is a technology that simulates the human brain to perform complex calculations. With the improvement of data storage and computing power, AI technology has been increasingly used. The 3rd Generation Partnership Project (3GPP) proposed to apply AI to the new radio (NR) communication system to improve network performance and user experience through intelligent collection and data analysis.

[0003] At present, for a certain functional module, there may be multiple AI models in the terminal to implement the function of the functional module. How to select a suitable AI model to implement the function of the functional module requires further research. Summary of the invention

[0004] The present application provides a communication method and device for selecting a suitable AI model to implement the functions of a functional module.

[0005] In a first aspect, an embodiment of the present application provides a communication method. The method can be applied to a first device. For example, the first device may be a terminal or a module in a terminal (such as a circuit, a chip, a chip system, or a processor), or a logical node, a logical module, or software that can implement all or part of the terminal functions. Among them, the method may include: the first device receives a first indication message and a first information. Among them, the first indication message can be used to indicate the activation of a first functional module; the first information is used to determine that the target AI model to be activated includes N AI models, each of the N AI models is used to implement the function of the first functional module, and N is a positive integer. Then, the first device can activate the N AI models according to the first information.

[0006] Through this method, the first device can select appropriate N AI models according to the first information to implement the function of the first functional module, thereby improving the performance of implementing the function of the first functional module.

[0007] In some possible designs, the first information is used to indicate at least one of the following: input parameters of the target AI model; output parameters of the target AI model; version number corresponding to the target AI model; format of the target AI model; requirements of the target AI model for terminal capabilities; equipment vendors of equipment that can apply the target AI model; application scenarios of the target AI model; requirements for the computational complexity of the target AI model; requirements for the processing power of the target AI model; requirements for the performance of the target AI model; range of target AI model sizes; or whether to allow the output results of multiple target AI models to be merged. The first information indicates the conditions that the target AI model should meet, so that through the first information, the first device can quickly and accurately determine the target AI model to be activated.

[0008] In some possible designs, the first device may also determine the available resources of the first device. The available resources and the first information may be used to determine that the target AI model includes N AI models. Through this design, the first device may determine that the target AI model to be activated includes N AI models based on the available resources of the first device and the first information, so that N AI models adapted to the available resources of the first device may be selected to implement the functions of the first functional module, thereby improving the performance of implementing the functions of the first functional module.

[0009] In some possible designs, the available resources include at least one of the following: available computing power of the first device; available power of the first device; or available storage space of the first device. The design provides multiple possible forms of available resources, and the first device can flexibly determine the target AI model to be activated based on them.

[0010] In some possible designs, the first device may also determine second information, and the second information and the first information are used to determine that the target AI model includes N AI models. The second information is used to indicate at least one of the following of the M AI models: credibility, accuracy, complexity, power consumption, type of input information, size of input information, type of output information, or size of output information. M is a positive integer greater than or equal to N, and the M AI models are AI models included in the first device that can implement the function of the first functional module. The second information indicates the status of the M AI models, so that the first device can select appropriate N AI models to implement the function of the first functional module based on the status of the M AI models and the first information, thereby improving the performance of implementing the function of the first functional module.

[0011] In some possible designs, the first device may send a first output result, which is the output result of some AI models among the N AI models, and the some AI models are AI models among the N AI models that meet the first condition. Wherein, N is greater than or equal to 2. Through this design, the first device may activate multiple AI models corresponding to the first functional module, and the first output result sent by the first device is the output result of the AI ​​model among the multiple AI models that meet the first condition, thereby improving the accuracy and / or credibility of the first output result.

[0012] In some possible designs, the first condition includes at least one of the following: some AI models are the AI ​​models with the highest credibility among the N AI models; some AI models are the AI ​​models with the highest accuracy among the N AI models; or some AI models are the AI ​​models whose output results are the most similar to the last measurement result of the first device among the N AI models. With this design, the first device can quickly and accurately determine the AI ​​model corresponding to the first output result.

[0013] In some possible designs, the first device may send a second output result, and the second output result is related to the output results of N AI models. Where N is greater than or equal to 2. Through this design, the first device can activate multiple AI models corresponding to the first functional module, and the second output result sent by the first device is related to the output results of the multiple AI models, thereby improving the accuracy and / or credibility of the second output result.

[0014] In some possible designs, the second output result is a weighted sum of the output results of the N AI models. This design can reduce the complexity of obtaining the first output result.

[0015] In the second aspect, an embodiment of the present application provides a communication method. The method can be applied to a second device. For example, the second device may also be an access network device or other device capable of managing life cycle management (LCM), or a module applied to an access network device or other device capable of managing LCM, such as a circuit, a chip, a chip system or a processor, or a logical node, a logical module or software that can implement all or part of the functions of an access network device or other device capable of managing LCM. Among them, the method may include: the second device sends a first indication message and a first message. Among them, the first indication message is used to indicate the activation of a first functional module; the first information can be used to determine that the target AI model to be activated includes N AI models, and the first information is used to activate N AI models, each of the N AI models is used to implement the function of the first functional module, and N is a positive integer.

[0016] In some possible designs, the first information is used to indicate at least one of the following: input parameters of the target AI model; output parameters of the target AI model; version number corresponding to the target AI model; format of the target AI model; requirements of the target AI model for terminal capabilities; equipment manufacturers of devices that can apply the target AI model; application scenarios of the target AI model; requirements for the computational complexity of the target AI model; requirements for the processing capabilities of the target AI model; requirements for the performance of the target AI model; range of target AI model sizes; or whether it is allowed to merge the output results of multiple target AI models.

[0017] In some possible designs, the second device may receive a first output result, where the first output result is an output result of some AI models among the N AI models, where the some AI models are AI models among the N AI models that meet the first condition, where N is greater than or equal to 2.

[0018] In some possible designs, the first condition includes at least one of the following: some AI models are the most credible AI models among N AI models; some AI models are the most accurate AI models among N AI models; or some AI models are the AI ​​models whose output results are most similar to the last measurement results of the first device among the N AI models.

[0019] In some possible designs, the second device may receive a second output result, where the second output result is related to output results of N AI models, where N is greater than or equal to 2.

[0020] In some possible designs, the second output result is a weighted sum of the output results of N AI models.

[0021] In a third aspect, the present application provides a communication device, which may be a terminal or a module in a terminal (such as a circuit, a chip, a chip system or a processor), or a logical node, a logical module or software that can implement all or part of the terminal functions. The communication device has the function of implementing the first aspect above. For example, the communication device includes a module or unit or means corresponding to the operation involved in the first aspect above, and the module or unit or means may be implemented by software, or by hardware, or the corresponding software may be implemented by hardware.

[0022] In one possible design, the communication device includes a processing unit and an interface unit. The interface unit can be used to send and receive signals to achieve communication between the communication device and other devices; the processing unit can be used to perform some internal operations of the communication device. The functions performed by the processing unit and the interface unit can correspond to the operations involved in the first aspect above.

[0023] In one possible design, the communication device includes a processor, which can be used to couple with a memory. The memory can store necessary computer programs or instructions for implementing the functions involved in the first aspect. The processor can execute the computer program or instructions stored in the memory, and when the computer program or instructions are executed, the communication device implements the method in any possible design of the first aspect.

[0024] In one possible design, the communication device includes a processor and a memory, and the memory can store necessary computer programs or instructions for implementing the functions involved in the first aspect. The processor can execute the computer program or instructions stored in the memory, and when the computer program or instructions are executed, the communication device implements the method in any possible design of the first aspect.

[0025] In one possible design, the communication device includes a processor and an interface circuit, wherein the processor is used to communicate with other devices through the interface circuit and execute the method in any possible design of the first aspect above.

[0026] In a fourth aspect, the present application provides a communication device, which may be an access network device or a module in the access network device (such as a circuit, a chip, a chip system or a processor), and may also be a logical node, a logical module or software that can implement all or part of the functions of the access network device. The communication device has the function of implementing the second aspect above. For example, the communication device includes a module or unit or means corresponding to the operation involved in the second aspect above, and the module or unit or means may be implemented by software, or by hardware, or the corresponding software may be implemented by hardware.

[0027] In one possible design, the communication device includes a processing unit and an interface unit. The interface unit can be used to send and receive signals to achieve communication between the communication device and other devices; the processing unit can be used to perform some internal operations of the communication device. The functions performed by the processing unit and the interface unit can correspond to the operations involved in the second aspect above.

[0028] In one possible design, the communication device includes a processor, which can be used to couple with a memory. The memory can store necessary computer programs or instructions for implementing the functions involved in the second aspect. The processor can execute the computer program or instructions stored in the memory, and when the computer program or instructions are executed, the communication device implements the method in any possible design of the second aspect.

[0029] In one possible design, the communication device includes a processor and a memory, and the memory can store necessary computer programs or instructions for implementing the functions involved in the second aspect. The processor can execute the computer program or instructions stored in the memory, and when the computer program or instructions are executed, the communication device implements the method in any possible design of the second aspect.

[0030] In one possible design, the communication device includes a processor and an interface circuit, wherein the processor is used to communicate with other devices through the interface circuit and execute the method in any possible design of the second aspect above.

[0031] It can be understood that in the third aspect or the fourth aspect above, the processor can be implemented by hardware or by software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor, which is implemented by reading the software code stored in the memory. In addition, the above processors can be one or more, and the memories can be one or more. The memory can be integrated with the processor, or the memory can be separately set from the processor. In the specific implementation process, the memory can be integrated with the processor on the same chip, or can be set on different chips respectively. The embodiment of the present application does not limit the type of memory and the setting method of the memory and the processor.

[0032] In a fifth aspect, the present application provides a communication system, which may include the communication device described in the third aspect and the communication device described in the fourth aspect. For example, the communication system includes a terminal and an access network device; wherein the terminal is used to execute the communication method provided in the first aspect, and the access network device is used to execute the communication method provided in the second aspect.

[0033] In a sixth aspect, the present application provides a computer-readable storage medium, in which a computer program or instruction is stored. When the computer program or instruction is executed, the method in any possible design of any aspect of the first to second aspects above is implemented.

[0034] In a seventh aspect, the present application provides a computer program product, which includes a computer program code. When the computer program code is executed, the method in any possible design of any one of the first to second aspects above is implemented.

[0035] In an eighth aspect, the present application provides a chip for reading a computer program stored in a memory to execute a method in any possible design of any one of the first to second aspects above.

[0036] The technical effects that can be achieved in any of the second to eighth aspects mentioned above can refer to the description of the technical effects that can be achieved by any possible design in any of the first aspect mentioned above, and the repetitions will not be discussed here. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 An architecture diagram of a communication system provided in an embodiment of the present application;

[0038] Figure 2 An architecture diagram of another communication system provided in an embodiment of the present application;

[0039] Figure 3 A flow chart of a communication method provided in an embodiment of the present application;

[0040] Figure 4 A structural diagram of a communication device provided in an embodiment of the present application;

[0041] Figure 5 A structural diagram of another communication device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The technical solution in the embodiment of the present application will be described below in conjunction with the accompanying drawings in the embodiment of the present application. The technical solution in the embodiment of the present application can be applied to various communication systems, such as universal mobile telecommunications system (UMTS), wireless local area network (WLAN), wireless fidelity (Wi-Fi) system, fourth generation (4G) mobile communication system (such as long term evolution (LTE) system), fifth generation (5G) mobile communication system (such as NR system), and future evolved communication system (such as sixth generation (6G) mobile communication system).

[0043] The present application will present various aspects, embodiments or features around a system that may include multiple devices, components, modules, etc. It should be understood and appreciated that each system may include additional devices, components, modules, etc., and / or may not include all of the devices, components, modules, etc. discussed in conjunction with the figures. In addition, combinations of these schemes may also be used.

[0044] To facilitate understanding of the embodiments of the present application, Figure 1 A possible, non-limiting system schematic is shown. Figure 1As shown, the communication system 10 includes a radio access network (RAN) 100 and a core network (CN) 200. Optionally, the communication system 10 may further include the Internet 300.

[0045] RAN 100 includes at least one RAN node (e.g. Figure 1 110a and 110b, collectively referred to as 110) and at least one terminal (such as Figure 1 RAN 100 may also include other RAN nodes, such as wireless relay equipment and / or wireless backhaul equipment ( Figure 1 The terminal 120 is connected to the RAN node 110 in a wireless manner. The RAN node 110 is connected to the core network 200 in a wireless or wired manner. The core network device in the core network 200 and the RAN node 110 in the RAN 100 may be different physical devices, or may be the same physical device that integrates the core network logical function and the radio access network logical function.

[0046] RAN 100 may be a 3GPP-related cellular system, such as a 4G or 5G mobile communication system, or a future-oriented evolution system (such as a 6G mobile communication system). RAN 100 may also be an open access network (open RAN, O-RAN or ORAN), a cloud radio access network (cloud radio access network, CRAN), or a WiFi system. RAN 100 may also be a communication system that integrates two or more of the above systems.

[0047] The RAN node 110, which may also be sometimes referred to as a RAN entity or an access node, constitutes a part of the communication system to help the terminal achieve wireless access. The multiple RAN nodes 110 in the communication system 10 may be nodes of the same type or nodes of different types. In some scenarios, the roles of the RAN node 110 and the terminal 120 are relative, for example, Figure 1 The network element 120i may be a helicopter or a drone, which may be configured as a mobile base station. For the terminals 120j that access the RAN 100 through the network element 120i, the network element 120i is a base station; but for the base station 110a, the network element 120i is a terminal. The RAN node 110 and the terminal 120 are sometimes referred to as communication devices, for example Figure 1 The network elements 110a and 110b may be understood as communication devices having base station functions, and the network elements 120a-120j may be understood as communication devices having terminal functions.

[0048] The RAN node may also be expressed in different ways, such as access network equipment. Unless otherwise specified in this application, the access network equipment is used for expression.

[0049] In one possible scenario, the access network device may be a base station, an evolved NodeB (eNodeB), an access point (AP), a transmission reception point (TRP), a next generation NodeB (gNB), a next generation base station in a 6G mobile communication system, a base station in a future mobile communication system, or an access node in a WiFi system. The access network device may be a macro base station (e.g. Figure 1 110a in), micro base stations or indoor stations (such as Figure 1 110b in the example), a relay node or a donor node, or a wireless controller in a CRAN scenario. Optionally, the access network device may also be a server, a wearable device, a vehicle or an on-board device, etc. For example, the access network device in the vehicle to everything (V2X) technology may be a road side unit (RSU). All or part of the functions of the access network device in the present application may also be implemented by software functions running on hardware, or by virtualization functions instantiated on a platform (such as a cloud platform). The access network device in the present application may also be a logical node, a logical module or software that can implement all or part of the functions of the access network device.

[0050] In another possible scenario, Figure 2 As shown, multiple access network devices collaborate to assist the terminal in achieving wireless access, and different access network devices respectively implement part of the functions of the base station. For example, the access network device can be a centralized unit (central unit, CU), a distributed unit (distributed unit, DU), a CU-control plane (control plane, CP), a CU-user plane (user plane, UP), or a radio unit (radio unit, RU), etc. The CU and DU can be set separately, or they can also be included in the same network element, such as a baseband unit (baseband unit, BBU). The RU can be included in a radio frequency device or a radio frequency unit, such as a remote radio unit (remote radio unit, RRU), an active antenna unit (active antenna unit, AAU) or a remote radio head (remote radio head, RRH).

[0051] In different systems, CU (or CU-CP and CU-UP), DU or RU may also have different names, but those skilled in the art can understand their meanings. For example, in the ORAN system, CU may also be called an open CU (open CU, O-CU), DU may also be called an open DU (open DU, O-DU), CU-CP may also be called an open CU-CP (open CU-CP, O-CU-CP), CU-UP may also be called an open CU-UP (open CU-UP, O-CU-UP), and RU may also be called an open RU (open RU, O-RU). Any unit in the CU (or CU-CP, CU-UP), DU and RU in this application may be implemented by a software module, a hardware module, or a combination of a software module and a hardware module.

[0052] The terminal may also be referred to as a terminal device, user equipment (UE), mobile station, mobile terminal, etc. The terminal can be widely used in various scenarios, for example, device-to-device (D2D), vehicle-to-everything (V2X) communication, machine-type communication (MTC), Internet of Things (IOT), virtual reality, augmented reality, industrial control, autonomous driving, telemedicine, smart grid, smart furniture, smart office, smart wear, smart transportation, smart city, etc. The terminal may be a mobile phone, a tablet computer, a computer with wireless transceiver function, a wearable device, a vehicle, a drone, a helicopter, an airplane, a ship, a robot, a robotic arm, a smart home device, etc. The embodiments of the present application do not limit the device form of the terminal.

[0053] In the following text of this application, "sending information to a device (such as a terminal)" can be understood as the destination of the information being the device, and can include directly or indirectly sending information to the device. "Receiving information from a device (such as a terminal)" or "receiving information from a device (such as a terminal)" can be understood as the source of the information being the device, and can include directly or indirectly receiving information from the device. The information may be processed as necessary between the source and destination of the information, such as format changes, but the destination can understand the valid information from the source. Similar expressions in this application can be understood similarly and will not be repeated here.

[0054] The communication system and business scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. A person of ordinary skill in the art can appreciate that with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0055] The following first explains the relevant terms involved in the embodiments of the present application. It should be noted that these explanations are intended to make the embodiments of the present application easier to understand and should not be regarded as limiting the scope of protection claimed by the present application.

[0056] (1) AI use case:

[0057] AI application scenarios include but are not limited to at least one of the following: energy saving, load balancing, mobility optimization, channel status information reference signal (CSI-RS) feedback enhancement, beam management enhancement, or positioning accuracy enhancement. These are described below.

[0058] 1) Energy saving:

[0059] The access network equipment can predict the load of the access network equipment based on its own and neighboring cell's load, energy consumption, energy efficiency and other information, as well as the terminal's trajectory, measurement results and other information. Under the premise of not affecting network coverage and user access, the access network equipment can take appropriate energy-saving measures in a timely manner based on the prediction results. Among them, the energy-saving measures may include at least one of the following: deactivating the cell, shutting down the carrier, shutting down the channel, shutting down the time slot, reducing the transmission power, etc.

[0060] 2) Load balancing:

[0061] The access network equipment can predict the load of the access network equipment based on its own and neighboring cell's load, energy consumption, energy efficiency and other information, as well as the terminal's trajectory, measurement results and other information. Based on the prediction results, the access network equipment can switch some terminals to neighboring cells, or switch some terminals served by neighboring cells to the local cell, so that the loads of the access network equipment in the network are close to each other, avoiding the situation where some access network equipment is overloaded, affecting the terminal's business, while the resources of other access network equipment are idle.

[0062] 3) Mobility optimization:

[0063] The access network equipment can predict the future trajectory of the terminal based on the historical trajectory information of the terminal and the measurement information of the terminal. Based on the prediction results, the access network equipment can determine in advance whether the terminal needs to switch, configure the information required for the switch in advance for the terminal that needs to switch, and notify the target cell to prepare access resources for the terminal, thereby reducing the delay of the terminal during the switching process and improving the success rate of the terminal switching.

[0064] 4) CSI-RS feedback enhancement:

[0065] The access network device can provide the terminal with an encoder and quantization tool based on the terminal's capabilities and other information. After the terminal compresses and quantizes the CSI-RS based on the encoder and quantization tool, it can send the CSI-RS to the access network device. The access network device can determine the channel quality of the terminal based on the CSI-RS.

[0066] 5) Enhanced beam management:

[0067] The access network device can obtain full beam scanning results from multiple terminals and train a sparse scanning matrix accordingly. Then, after sending the sparse scanning matrix to terminal a, the access network device can obtain the sparse scanning results from terminal a and determine the optimal CSI-RS beam of terminal a based on the sparse scanning results.

[0068] 6) Positioning enhancement:

[0069] Positioning enhancement is used to improve positioning accuracy. Positioning enhancement may include at least one of the following: positioning enhancement based on access network equipment, positioning enhancement based on positioning management function network elements, and positioning enhancement based on terminals.

[0070] (2) AI Model:

[0071] The AI ​​model can represent the mapping relationship between the input and output of the model. The AI ​​model can be a neural network, linear regression model, decision tree model, support vector machine (SVM), Bayesian network, Q learning model or other machine learning model.

[0072] The AI ​​model can be used to implement AI functions, and the AI ​​functions can be functions of functional modules. Exemplarily, the AI ​​functions can include at least one of the following: data collection (collecting training data and / or reasoning data), data preprocessing, model training (or model learning), model information release (configuring model information), model verification, model reasoning, or reasoning result release. Among them, reasoning can also be called prediction.

[0073] When the AI ​​model is located on the terminal side, for a certain functional module, there may be multiple AI models in the terminal that can implement the function of the functional module. For example, for the CSI channel measurement result prediction function, there are multiple AI models in the terminal that can provide prediction results.

[0074] (3) LCM in the AI ​​model application process:

[0075] During the application of the AI ​​model, due to factors such as the mobility of the terminal, changes in user services, and changes in hardware resources, the terminal side may need to replace the AI ​​model used, that is, the AI ​​model needs to be LCMed to optimize the performance of the terminal. The LCM process may include at least one of the following: activation, deactivation, switching, updating, transmission, or rollback of the AI ​​model and / or functional module.

[0076] LCM is usually functional-based. Specifically, the access network device indicates that the terminal needs to perform LCM processing on a certain functional module. The terminal can independently select an AI model that implements the function of the functional module and perform LCM processing on the AI ​​model. For example, the access network device indicates that the terminal needs to activate the CSI channel measurement result prediction function. The terminal can select an AI model that implements the CSI channel prediction function and activate the AI ​​model.

[0077] At present, for a certain functional module, there may be multiple AI models in the terminal to implement the function of the functional module. How the terminal selects a suitable AI model to implement the function of the functional module requires further research. In addition, the terminal usually only selects one AI model to implement the function of the functional module, and the accuracy of implementing the function of the functional module through the AI ​​model is low.

[0078] In view of this, an embodiment of the present application provides a communication method. Figure 3 A flow chart corresponding to the communication method provided in the embodiment of the present application. Figure 3In the method, the first device and the second device are used as the execution subject of the interaction diagram as an example to illustrate the method, but the present application does not limit the execution subject of the interaction diagram. For example, the first device can be a terminal, or a module applied to the terminal, such as a circuit, a chip, a chip system or a processor, or a logical node, a logical module or software that can realize all or part of the terminal functions; the second device can also be an access network device or other device capable of managing LCM (for example, operation administration and maintenance (OAM) equipment or core network equipment), or a module applied to the access network device or other device capable of managing LCM, such as a circuit, a chip, a chip system or a processor, or a logical node, a logical module or software that can realize all or part of the functions of the access network device or other device capable of managing LCM. For another example, the first device can be an access network device, or a module applied to the access network device, such as a circuit, a chip, a chip system or a processor, or a logical node, a logical module or software that can realize all or part of the functions of the access network device; the second device can be a terminal, or a module applied to the terminal, such as a circuit, a chip, a chip system or a processor, or a logical node, a logical module or software that can realize all or part of the functions of the terminal. like Figure 3 As shown, the method includes:

[0079] S301: The second device sends first indication information; correspondingly, the first device receives the first indication information, wherein the first indication information can be used to indicate activation of a first functional module.

[0080] In some possible ways, when the second device determines to activate the function corresponding to the first functional module on the first device side, the second device may determine to activate the first functional module on the first device side and send a first indication message to the first device. The present application does not limit the specific content of the function corresponding to the first functional module on the first device side that the second device determines to activate. It should be understood that there may be one or more functional modules that the second device determines to activate, and the present application takes the first functional module as an example for explanation.

[0081] The first indication information may indicate activation of the first functional module in a variety of ways. In some examples, the first indication information includes an identification (identity, ID) of the first functional module. Thus, after receiving the first indication information, the first device may determine to activate the first functional module. In other examples, the first indication information includes indication information of the function corresponding to the first functional module. Thus, after receiving the first indication information, the first device may determine to activate the first functional module. For example, the first indication information includes indication information of the CSI measurement result compression function. After receiving the first indication information, the first device may determine that the first functional module is a functional module corresponding to the CSI measurement result compression function, and determine to activate the first functional module. For example, the first indication information includes indication information of the beam prediction function, so that after receiving the first indication information, the first device may determine that the first functional module is a functional module corresponding to the beam prediction function, and determine to activate the first functional module.

[0082] The first indication information may be carried in a traditional message or in a new message, and this application does not impose any limitation on this.

[0083] S302: The second device sends the first information; correspondingly, the first device receives the first information.

[0084] Among them, the first information can be used to determine that the target AI model to be activated includes N AI models, each of the N AI models can realize the function of the first functional module, and N is a positive integer. In other words, the first device can determine that the target AI model to be activated includes N AI models based on the first information. The specific content of the determination will be described in S303 below, and the content of the first information will also be described in S303, which will not be expanded here. The first information can be carried in a traditional message or in a new message. The first information and the first indication information can be carried in the same message or in different messages. In addition, this application does not limit the execution order of S301 and S302.

[0085] S303: The first device activates N AI models according to the first information.

[0086] Optionally, S303 includes steps A1 to A2:

[0087] Step A1: The first device determines, based on the first information, that the target AI models to be activated include N AI models.

[0088] Exemplarily, the first information is used to indicate at least one of the following 1 to 12: 1. Input parameters of the target AI model; 2. Output parameters of the target AI model; 3. Version number corresponding to the target AI model; 4. Format of the target AI model; 5. Requirements of the target AI model for the capabilities of the first device; 6. Equipment vendors of equipment that can apply the target AI model; 7. Application scenarios of the target AI model; 8. Requirements for the computational complexity of the target AI model; 9. Requirements for the processing power of the target AI model; 10. Requirements for the performance of the target AI model; 11. Range of target AI model size; 12. Whether to allow the output results of multiple target AI models to be merged. Among them, points 1 to 12 are conditions that the target AI model should meet, that is, the first information can indicate the conditions that the target AI model should meet, so that through the first information, the first device can quickly and accurately determine the target AI model to be activated.

[0089] In combination with the content of the first information, the first device includes M AI models capable of implementing the function of the first functional module as an example to illustrate how the first device determines, based on the first information, that the target AI models to be activated include N AI models. Where M is a positive integer greater than or equal to N.

[0090] Example 1: The first information is used to indicate the input parameters of the target AI model; the target AI model includes: an AI model whose input parameters include the input parameters indicated by the first information among the M AI models. For example, the first information indicates: the input parameter of the target AI model is input parameter 1. The M AI models are AI model 1 to AI model 3. The input parameters of AI model 1 include input parameter 1, the input parameters of AI model 2 include input parameter 1 and input parameter 2, and the input parameters of AI model 3 include input parameter 2. In this case, the first device can determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0091] Example 2: The first information is used to indicate the output parameter of the target AI model; the target AI model includes: an AI model whose output parameter includes the output parameter indicated by the first information among the M AI models. For example, the first information indicates: the output parameter of the target AI model is output parameter 1. The M AI models are AI model 1 to AI model 3. The output parameter of AI model 1 includes output parameter 1, the output parameter of AI model 2 includes output parameter 1 and output parameter 2, and the output parameter of AI model 3 includes output parameter 2. In this case, the first device can determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0092] Example 3: The first information is used to indicate the version number of the target AI model; the target AI model includes: an AI model whose version number includes the version number indicated by the first information among the M AI models. For example, the first information indicates: the version number of the target AI model is version number 1. The M AI models are AI model 1 to AI model 3. The version numbers of AI model 1 and AI model 2 are version number 1, and the version number of AI model 3 is version number 2. In this case, the first device can determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0093] Example 4: The first information is used to indicate the format of the target AI model; the target AI model includes: an AI model whose format includes the format indicated by the first information among the M AI models. For example, the first information indicates: the format of the target AI model is format 1. The M AI models are AI model 1 to AI model 3. The formats of AI model 1 and AI model 2 are format 1, and the format of AI model 3 is format 2. In this case, the first device can determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0094] Example 5: The first information is used to indicate the requirements of the target AI model on the capabilities of the first device; the target model includes: an AI model among the M AI models that meets the requirements of the capabilities of the first device indicated by the first information. The requirements of the target AI model on the capabilities of the first device may include at least one of the following: the requirements of the target AI model on the available computing power of the first device, the requirements of the target AI model on the available storage space of the first device, or the requirements of the target AI model on the available power of the first device.

[0095] For example, the first information indicates that the target AI model requires the available computing power of the first device to be greater than or equal to 1 Tera (T, i.e., 1012) floating-point operations per second (FLPS). The M AI models are AI model 1 to AI model 3. AI model 1 and AI model 2 require the available computing power of the first device to be greater than or equal to 1T FLPS, and AI model 3 requires the available computing power of the first device to be greater than or equal to 0.5T FLPS. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, i.e., the N AI models include AI model 1 and AI model 2.

[0096] For another example, the first information indicates that the target AI model requires the available storage space of the first device to be greater than or equal to 50 megabytes (MB). The M AI models are AI model 1 to AI model 3. AI model 1 and AI model 2 require the available storage space of the first device to be greater than or equal to 50MB, and AI model 3 requires the available storage space of the first device to be greater than or equal to 30MB. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0097] For another example, the first information indicates that the target AI model requires the available power of the first device to be greater than or equal to 2000 mAh. The M AI models are AI model 1 to AI model 3. AI model 1 and AI model 2 require the available power of the first device to be greater than or equal to 2000 mAh, and AI model 3 requires the available power of the first device to be greater than or equal to 1000 mAh. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0098] For another example, the first information indicates: the target AI model requires the available storage space of the first device to be greater than or equal to 50 megabytes; the target AI model requires the available power of the first device to be greater than or equal to 2000 milliamperes. The M AI models are AI model 1 to AI model 3. AI model 1 requires the available storage space of the first device to be greater than or equal to 50MB, and AI model 2 and AI model 3 require the available storage space of the first device to be greater than or equal to 30MB. AI model 1 and AI model 2 require the available power of the first device to be greater than or equal to 2000 milliamperes, and AI model 3 requires the available power of the first device to be greater than or equal to 1000 milliamperes. In this case, the first device may determine that the target AI model includes AI model 1, that is, the N AI models include AI model 1.

[0099] Example 6: The first information is used to indicate the equipment vendor of the device that can apply the target AI model; the target AI model includes: the AI ​​model of the equipment vendor indicated by the first information among the M AI models. For example, the first information indicates: the equipment vendors of the device that can apply the target AI model include equipment vendor 1. The M AI models are AI model 1 to AI model 3. The equipment vendors of the devices that can apply AI model 1 and AI model 2 include equipment vendor 1, and the equipment vendors of the devices that can apply AI model 3 include equipment vendor 2. In this case, the first device can determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0100] Example 7: The first information is used to indicate the application scenario of the target AI model; the target AI model includes: an AI model among the M AI models that can apply the application scenario indicated by the first information. For example, the first information indicates: the application scenario of the target AI model includes application scenario 1. The M AI models are AI model 1 to AI model 3. The application scenarios of AI model 1 and AI model 2 include application scenario 1, and the application scenarios of AI model 3 include application scenario 2. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0101] Optionally, the application scenario of the target AI model includes at least one of the following: (1) The requirements of the target AI model on the hardware configuration of the first device and the second device, for example, the requirements of the target AI model on the number of receiving antennas of the first device and / or the number of transmission ports of the second device. (2) The requirements of the target AI model on the channel environment, for example, the target AI model requires the channel environment to be an urban micro (UMi) scenario, an urban macro (UMa) scenario, or an indoor hotspot (InH) scenario. (3) The requirements of the target AI model on the network access configuration, for example, the access frequency of the target AI model. (4) The requirements of the target AI model on the local environment of the first device, for example, the requirements of the target AI on the speed of the first device.

[0102] Example 8: The first information is used to indicate the requirement for the computational complexity of the target AI model; the target AI model includes: an AI model among the M AI models that meets the requirement for the computational complexity indicated by the first information. For example, the first information indicates: the computational complexity of the target AI model is lower than the computational complexity threshold. The M AI models are AI model 1 to AI model 3. The computational complexity of AI model 1 and AI model 2 is lower than the computational complexity threshold, and the computational complexity of AI model 3 is higher than the computational complexity threshold. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0103] Example 9: The first information is used to indicate the requirements for the processing capability of the target AI model; the target AI model includes: an AI model among the M AI models that meets the requirements for the processing capability indicated by the first information. For example, the first information indicates: the target AI model provides output results before a first time. The M AI models are AI model 1 to AI model 3. AI model 1 and AI model 2 are capable of providing output results before a first time, and AI model 3 provides output results after the first time. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0104] Example 10: The first information is used to indicate the performance requirements of the target AI model; the target AI model includes: an AI model among the M AI models that meets the performance requirements indicated by the first information. The performance requirements of the target AI model may include requirements for the accuracy and / or credibility of the target AI model.

[0105] For example, the first information indicates that the accuracy of the target AI model is higher than the first accuracy threshold. The M AI models are AI model 1 to AI model 3. The accuracy of AI model 1 and AI model 2 is higher than the first accuracy threshold, and the accuracy of AI model 3 is lower than the first accuracy threshold. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0106] For another example, the first information indicates that the credibility of the target AI model is higher than the first credibility threshold. The M AI models are AI model 1 to AI model 3. The credibility of AI model 1 and AI model 2 is higher than the first credibility threshold, and the credibility of AI model 3 is lower than the first credibility threshold. In this case, the first device may determine that the target AI model includes AI model 1 and AI model 2, that is, the N AI models include AI model 1 and AI model 2.

[0107] For another example: the first information indicates that the accuracy of the target AI model is higher than the first accuracy threshold, and the credibility of the target AI model is higher than the first credibility threshold. The M AI models are AI model 1 to AI model 3. The accuracy of AI model 1 is higher than the first accuracy threshold, and the accuracy of AI model 2 and AI model 3 is lower than the first accuracy threshold. The credibility of AI model 1 and AI model 2 is higher than the first credibility threshold, and the credibility of AI model 3 is lower than the first credibility threshold. In this case, the first device can determine that the target AI model includes AI model 1, that is, the N AI models include AI model 1.

[0108] Example 11: The first information is used to indicate the range of the target AI model size; the target AI model includes: an AI model among the M AI models that meets the size range indicated by the first information. For example, the first information indicates: the target AI model size range is 40 to 60 MB. The M AI models are AI model 1 to AI model 3. The size of AI model 1 is 50 MB, the size of AI model 2 is 80 MB, and the size of AI model 3 is 30 MB. In this case, the first device can determine that the target AI model includes AI model 1, that is, the N AI models include AI model 1.

[0109] Example 12: The first information indicates whether the output results of multiple target AI models are allowed to be merged. When the output results of multiple AI models are allowed to be merged, the target AI model includes: one or more AI models among the M AI models; and / or, when the output results of multiple AI models are not allowed to be merged, the target AI model includes: one AI model among the M AI models. Optionally, when the first information does not indicate whether the output results of multiple target AI models are allowed to be merged, it may be allowed by default to merge the output results of multiple AI models, or it may be not allowed to merge the results of multiple AI models by default.

[0110] It should be understood that at least two of the above examples 1 to 12 can be combined with each other. For example, Example 1 and Example 11 are combined. The first information is used to indicate: the input parameters of the target AI model and the range of the target AI model size; the target AI model includes: among the M AI models, the input parameters include the input parameters indicated by the first information, and the AI ​​model that meets the size range indicated by the first information. For example, the first information indicates: the input parameters of the target AI model include input parameter 1, and the range of the target AI model size is 40 to 60MB. The M AI models are AI model 1 to AI model 3. The input parameters of AI model 1 and AI model 2 include input parameter 1; the input parameters of AI model 3 include input parameter 2. The size of AI model 1 and AI model 3 is 50MB, and the size of AI model 2 is 80MB. In this case, the first device can determine that the target AI model includes AI model 1, that is, the N AI models include AI model 1.

[0111] Step A2: The first device activates N AI models. The present application does not limit the specific content of the first device activating N AI models.

[0112] pass Figure 3According to the method shown, the first device can select appropriate N AI models to implement the function of the first functional module based on the first information, thereby improving the performance of implementing the function of the first functional module. In addition, since the first information comes from the second device, the second device can assist the first device in selecting appropriate N AI models to implement the function of the first functional module, thereby effectively managing the AI ​​models in the first device.

[0113] As mentioned above, the first information can be used to determine that the target AI models to be activated include N AI models. Optionally, the basis for determining that the target AI models to be activated include N AI models may also include other information, which is described below through methods a1 and a2.

[0114] Mode a1: The first device determines the available resources of the first device, and the available resources and the first information can be used to determine that the target AI model to be activated includes N AI models. That is, the first device can determine that the target AI model to be activated includes N AI models based on the available resources of the first device and the first information.

[0115] Optionally, in method a1, the first device may determine through steps B1 to B2 that the target AI models to be activated include N AI models.

[0116] Step B1: The first device may determine, based on the first information, that the candidate target AI models include P AI models, where P is a positive integer greater than or equal to N. The specific content of step B1 may refer to step A1, except that the target AI model to be activated is replaced by the candidate target AI model, and the N AI models are replaced by P AI models, which will not be described in detail here.

[0117] Step B2: The first device may select a target AI model to be activated from P AI models according to the available resources of the first device, that is, select N AI models.

[0118] If the available resources of the first device are greater than or equal to the first available resource threshold, the first device may select the target AI model to be activated from the P AI models in a variety of ways, for example, way b1 or way b2. The first available resource threshold may be preset, for example, specified by a protocol; or may be determined by the first device; or may be notified to the first device by other devices (for example, the second device).

[0119] Mode b1: If the available resources of the first device are more than or equal to the first available resource threshold, the first device may select one or more AI models with the highest complexity from the P AI models as the target AI models to be activated. For example, the P AI models include AI model 1 and AI model 2. The complexity of AI model 1 is higher than the complexity of AI model 2. If the available resources of the first device are more than or equal to the first available resource threshold, the first device may select AI model 1 as the target AI model to be activated. In this way, when the first device has more available resources, the first device may select the AI ​​model with the highest complexity as the target AI model to be activated, thereby improving performance.

[0120] Mode b2: If the available resources of the first device are more than or equal to the first available resource threshold, the first device may select one or more AI models with a complexity higher than the first complexity threshold from the P AI models as the target AI models to be activated. For example, the P AI models include AI model 1 and AI model 2. The complexity of AI model 1 is higher than the first complexity threshold, and the complexity of AI model 2 is lower than the first complexity threshold. If the available resources of the first device are more than or equal to the first available resource threshold, the first device may select AI model 1 as the target AI model to be activated. In this way, when the first device has more available resources, the first device may select an AI model with higher complexity as the target AI model to be activated, thereby improving performance. Among them, the first complexity threshold may be pre-set, for example, specified by the protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0121] If the available resources of the first device are less than the second available resource threshold, the first device can select the target AI model to be activated from P AI models in a variety of ways, for example, way c1 or way c2. The second available resource threshold can be pre-set, for example, specified by the protocol; or it can be determined by the first device; or it can be notified to the first device by other devices (for example, the second device). Optionally, the second available resource threshold is less than or equal to the first available resource threshold.

[0122] Method c1: If the available resources of the first device are less than the second available resource threshold, the first device may select one or more AI models with the lowest complexity from the P AI models as the target AI model to be activated. For example, the P AI models include AI model 1 and AI model 2. The complexity of AI model 1 is higher than the complexity of AI model 2. If the available resources of the first device are less than the second available resource threshold, the first device may select AI model 2 as the target AI model to be activated. In this way, when the available resources of the first device are less, the first device may select the AI ​​model with the lowest complexity as the target AI model to be activated, thereby improving performance.

[0123] Mode c2: If the available resources of the first device are less than the second available resource threshold, the first device may select one or more AI models with a complexity lower than the second complexity threshold from the P AI models as the target AI model to be activated. For example, the P AI models include AI model 1 and AI model 2. The complexity of AI model 1 is higher than the second complexity threshold, and the complexity of AI model 2 is lower than the second complexity threshold. If the available resources of the first device are less than the second available resource threshold, the first device may select AI model 2 as the target AI model to be activated. In this way, when the available resources of the first device are less, the first device may select an AI model with lower complexity as the target AI model to be activated, thereby saving the resources of the first device. Among them, the second complexity threshold may be pre-set, for example, specified by the protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device). Optionally, the second complexity threshold is less than or equal to the first complexity threshold.

[0124] Optionally, the available resources of the first device include at least one of the following: available computing power of the first device, available power of the first device, or available storage space of the first device. The following describes how the first device selects a target AI model to be activated from P AI models based on the available resources of the first device.

[0125] Example 1: The available resources of the first device include the available computing power of the first device. The first available resource threshold includes a first available computing power threshold, and the second available resource threshold includes a second available computing power threshold. The first available computing power threshold is greater than or equal to the second available computing power threshold. If the available computing power of the first device is higher than or equal to the first available computing power threshold, the first device may select one or more AI models with the highest complexity from the P AI models as the target AI models to be activated; or, the first device may select one or more AI models with a complexity higher than the first complexity threshold from the P AI models as the target AI models to be activated. And / or, if the available computing power of the first device is lower than the second available computing power threshold, the first device may select one or more AI models with the lowest complexity from the P AI models as the target AI models to be activated; or, the first device may select one or more AI models with a complexity lower than the second complexity threshold from the P AI models as the target AI models to be activated.

[0126] Example 2: The available resources of the first device include the available power of the first device. The first available resource threshold includes a first available power threshold, and the second available resource threshold includes a second available power threshold. The first available power threshold is greater than or equal to the second available power threshold. If the available power of the first device is greater than or equal to the first available power threshold, the first device may select one or more AI models with the highest complexity from the P AI models as the target AI models to be activated; or, the first device may select one or more AI models with a complexity higher than the first complexity threshold from the P AI models as the target AI models to be activated. And / or, if the available power of the first device is less than the second available power threshold, the first device may select one or more AI models with the lowest complexity from the P AI models as the target AI models to be activated; or, the first device may select one or more AI models with a complexity lower than the second complexity threshold from the P AI models as the target AI models to be activated.

[0127] Example 3: The available resources of the first device include the available storage space of the first device. The first available resource threshold includes a first available storage space threshold, and the second available resource threshold includes a second available storage space threshold. The first available storage space threshold is greater than or equal to the second available storage space threshold. If the available storage space of the first device is greater than or equal to the first available storage space threshold, the first device may select one or more AI models with the highest complexity from the P AI models as the target AI models to be activated; or, the first device may select one or more AI models with a complexity higher than the first complexity threshold from the P AI models as the target AI models to be activated. And / or, if the available storage space of the first device is less than the second available storage space threshold, the first device may select one or more AI models with the lowest complexity from the P AI models as the target AI models to be activated; or, the first device may select one or more AI models with a complexity lower than the second complexity threshold from the P AI models as the target AI models to be activated.

[0128] It should be understood that at least two of the above examples 1 to 3 can be combined with each other. For example, Example 1, Example 2 and Example 3 are combined. The available resources of the first device include: the available computing power of the first device, the available power of the first device and the available storage space of the first device. The first available resource threshold includes the first available computing power threshold, the first available power threshold and the first available storage space threshold, and the second available resource threshold includes the second available computing power threshold, the second available power threshold and the second available storage space threshold. The first available computing power threshold is greater than or equal to the second available computing power threshold; the first available power threshold is greater than or equal to the second available power threshold; the first available storage space threshold is greater than or equal to the second available storage space threshold. If the available computing power of the first device is higher than or equal to the first available computing power threshold, the available power of the first device is greater than or equal to the first available power threshold, and the available storage space of the first device is greater than or equal to the first available storage space threshold, the first device may select one or more AI models with the highest complexity from the P AI models as the target AI model to be activated; or, the first device may select one or more AI models with a complexity higher than the first complexity threshold from the P AI models as the target AI model to be activated. And / or, if the available computing power of the first device is lower than the second available computing power threshold, the available power of the first device is less than the second available power threshold, and the available storage space of the first device is less than the second available storage space threshold, the first device may select one or more AI models with the lowest complexity from the P AI models as the target AI model to be activated; or, the first device may select one or more AI models with complexity lower than the second complexity threshold from the P AI models as the target AI model to be activated. In other words, when the available computing power of the first device is high, the available power is high, and the available storage space is high, the first device may select an AI model with high complexity as the target AI model to be activated, thereby improving performance; when the available computing power of the first device is low, the available power is low, and the available storage space is low, the first device may select an AI model with low complexity as the target AI model to be activated, thereby saving resources.

[0129] Through method a1, the first device can determine, based on the available resources of the first device and the first information, that the target AI model to be activated includes N AI models, so that N AI models adapted to the available resources of the first device can be selected to implement the functions of the first functional module, thereby improving the performance of implementing the functions of the first functional module.

[0130] In some possible methods, method a1 can be replaced by: the first device determines, based on the available resources of the first device, that the target AI model to be activated includes N AI models. Exemplarily, the first device includes M AI models that can implement the functions of the first functional module, where M is a positive integer greater than or equal to N. The first device can select the target AI model to be activated from the M AI models based on the available resources of the first device, that is, select N AI models. For details, please refer to step B2, except that P AI models are replaced by M AI models, which will not be repeated here.

[0131] Method a2: The first device may determine the second information, and the second information and the first information are used to determine that the target AI model includes N AI models. That is, the first device may determine that the target AI model to be activated includes N AI models based on the second information and the first information. Among them, the second information may indicate the status of M AI models. The M AI models are AI models included in the first device that can realize the function of the first functional module, and M is a positive integer greater than or equal to N. Exemplarily, the second information can be used to indicate at least one of the following of the M AI models: credibility, accuracy, complexity, power consumption, type of input information, size of input information, type of output information, or size of output information.

[0132] Optionally, in method a2, the first device may determine through steps C1 to C2 that the target AI models to be activated include N AI models.

[0133] Step C1: The first device may determine, based on the first information, that the candidate target AI models include P AI models, where P is a positive integer greater than or equal to N. The specific content of step C1 may refer to step A1, except that the target AI model to be activated is replaced by the candidate target AI model, and the N AI models are replaced by P AI models, which will not be described in detail here.

[0134] Step C2: The first device may select a target AI model to be activated from the P AI models according to the second information, that is, select N AI models.

[0135] The following describes step C2 in conjunction with the content of the second information.

[0136] Example a1: The second information is used to indicate the credibility of M AI models. In some examples, the target AI model to be activated may be the N AI models with the highest credibility among P AI models. For example, the P AI models include AI model 1 and AI model 2, and the credibility of AI model 1 is higher than the credibility of AI model 2. The target AI model to be activated may be AI model 1. In other examples, the target AI model to be activated may be N AI models among P AI models whose credibility is higher than the second credibility threshold. For example, the P AI models include AI model 1 and AI model 2, the credibility of AI model 1 is higher than the second credibility threshold, and the credibility of AI model 2 is lower than the second credibility threshold. The target AI model to be activated may be AI model 1. Among them, the second credibility threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0137] Example a2: The second information is used to indicate the accuracy of M AI models. In some examples, the target AI model to be activated may be the N AI models with the highest accuracy among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the accuracy of AI model 1 is higher than the accuracy of AI model 2. The target AI model to be activated may be AI model 1. In other examples, the target AI model to be activated may be N AI models among the P AI models whose accuracy is higher than the second accuracy threshold. For example, the P AI models include AI model 1 and AI model 2, the accuracy of AI model 1 is higher than the second accuracy threshold, and the accuracy of AI model 2 is lower than the second accuracy threshold. The target AI model to be activated may be AI model 1. Among them, the second accuracy threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0138] Example a3: The second information is used to indicate the complexity of the M AI models. In some examples, the target AI model to be activated may be the N AI models with the lowest complexity among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the complexity of AI model 1 is higher than the complexity of AI model 2. The target AI model to be activated may be AI model 2. In other examples, the target AI model to be activated may be the N AI models with complexity lower than the third complexity threshold among the P AI models. For example, the P AI models include AI model 1 and AI model 2, the complexity of AI model 1 is higher than the third complexity threshold, and the complexity of AI model 2 is lower than the third complexity threshold. The target AI model to be activated may be AI model 2. Among them, the third complexity threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0139] Example a4: The second information indicates the power consumption of M AI models. In some examples, the target AI models to be activated may be the N AI models with the lowest power consumption among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the power consumption of AI model 1 is higher than that of AI model 2. The target AI model to be activated may be AI model 2. In other examples, the target AI model to be activated may be the N AI models with power consumption lower than the power consumption threshold among the P AI models. For example, the P AI models include AI model 1 and AI model 2, the power consumption of AI model 1 is higher than the power consumption threshold, and the power consumption of AI model 2 is lower than the power consumption threshold. The target AI model to be activated may be AI model 2. The power consumption threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0140] Example a5: The second information indicates the type of input information of the M AI models.

[0141] In some examples, the target AI model to be activated may be the N AI models with the least types of input information among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the number of input information types of AI model 1 is greater than the number of input information types of AI model 2. The target AI model to be activated may be AI model 2.

[0142] In other examples, the target AI model to be activated may be N AI models among the P AI models whose number of input information types is less than the first input information type number threshold. For example, the P AI models include AI model 1 and AI model 2. The number of input information types of AI model 1 is greater than the first input information type number threshold, and the number of input information types of AI model 2 is less than the first input information type number threshold. The target AI model to be activated may be AI model 2. Among them, the first input information type number threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0143] In some other examples, the target AI model to be activated may be the N AI models with the most types of input information among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the number of input information types of AI model 1 is greater than the number of input information types of AI model 2. The target AI model to be activated may be AI model 1.

[0144] In some further examples, the target AI model to be activated may be N AI models among the P AI models whose number of input information types is greater than or equal to the second input information type number threshold. For example, the P AI models include AI model 1 and AI model 2. The number of input information types of AI model 1 is greater than the second input information type number threshold, and the number of input information types of AI model 2 is less than the second input information type number threshold. The target AI model to be activated may be AI model 1. Among them, the second input information type number threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device). Optionally, the second input information type number threshold is less than or equal to the first input information type number threshold.

[0145] Example a6: The second information indicates the data volume of input information of M AI models.

[0146] In some examples, the target AI model to be activated may be the N AI models with the least amount of input information data among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the amount of input information data of AI model 1 is greater than the amount of input information data of AI model 2. The target AI model to be activated may be AI model 2.

[0147] In other examples, the target AI model to be activated may be N AI models among P AI models whose input information data volume is less than the first input information data volume threshold. For example, the P AI models include AI model 1 and AI model 2. The input information data volume of AI model 1 is greater than the first input information data volume threshold, and the input information data volume of AI model 2 is less than the first input information data volume threshold. The target AI model to be activated may be AI model 2. Among them, the first input information data volume threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0148] In some other examples, the target AI model to be activated may be the N AI models with the largest amount of input information data among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the amount of input information data of AI model 1 is greater than the amount of input information data of AI model 2. The target AI model to be activated may be AI model 1.

[0149] In some further examples, the target AI model to be activated may be N AI models among the P AI models whose input information data volume is greater than or equal to the second input information data volume threshold. For example, the P AI models include AI model 1 and AI model 2. The input information data volume of AI model 1 is greater than the second input information data volume threshold, and the input information data volume of AI model 2 is less than the second input information data volume threshold. The target AI model to be activated may be AI model 1. Among them, the second input information data volume threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device). Optionally, the second input information data volume threshold is less than or equal to the first input information data volume threshold.

[0150] Example a7: The second information indicates the type of output information of the M AI models.

[0151] In some examples, the target AI model to be activated may be the N AI models with the least types of output information among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the number of output information types of AI model 1 is greater than the number of output information types of AI model 2. The target AI model to be activated may be AI model 2.

[0152] In other examples, the target AI model to be activated may be N AI models among P AI models whose number of output information types is less than the first output information type number threshold. For example, the P AI models include AI model 1 and AI model 2. The number of output information types of AI model 1 is greater than the first output information type number threshold, and the number of output information types of AI model 2 is less than the first output information type number threshold. The target AI model to be activated may be AI model 2. Among them, the first output information type number threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0153] In some other examples, the target AI model to be activated may be the N AI models with the most types of output information among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the number of output information types of AI model 1 is greater than the number of output information types of AI model 2. The target AI model to be activated may be AI model 1.

[0154] In some further examples, the target AI model to be activated may be N AI models among the P AI models whose number of output information types is greater than or equal to the second output information type number threshold. For example, the P AI models include AI model 1 and AI model 2. The number of output information types of AI model 1 is greater than the second output information type number threshold, and the number of output information types of AI model 2 is less than the second output information type number threshold. The target AI model to be activated may be AI model 1. Among them, the second output information type number threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device). Optionally, the second output information type number threshold is less than or equal to the first output information type number threshold.

[0155] Example a8: The second information indicates the data volume of the output information of M AI models.

[0156] In some examples, the target AI model to be activated may be the N AI models with the least amount of output information data among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the amount of output information data of AI model 1 is greater than the amount of output information data of AI model 2. The target AI model to be activated may be AI model 2.

[0157] In other examples, the target AI model to be activated may be N AI models among P AI models whose output information data volume is less than the first output information data volume threshold. For example, the P AI models include AI model 1 and AI model 2. The output information data volume of AI model 1 is greater than the first output information data volume threshold, and the output information data volume of AI model 2 is less than the first output information data volume threshold. The target AI model to be activated may be AI model 2. Among them, the first output information data volume threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device).

[0158] In some other examples, the target AI model to be activated may be the N AI models with the largest amount of output information data among the P AI models. For example, the P AI models include AI model 1 and AI model 2, and the amount of output information data of AI model 1 is greater than the amount of output information data of AI model 2. The target AI model to be activated may be AI model 1.

[0159] In some further examples, the target AI model to be activated may be N AI models among the P AI models whose output information data volume is greater than or equal to the second output information data volume threshold. For example, the P AI models include AI model 1 and AI model 2. The output information data volume of AI model 1 is greater than the second output information data volume threshold, and the output information data volume of AI model 2 is less than the second output information data volume threshold. The target AI model to be activated may be AI model 1. Among them, the second output information data volume threshold may be pre-set, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by other devices (for example, the second device). Optionally, the second output information data volume threshold is less than or equal to the first output information data volume threshold.

[0160] Example a8: The second information indicates the matching operators of the M AI models. The target AI models to be activated may be the N AI models among the P AI models that match the current operator of the first device. For example, the P AI models include AI model 1 and AI model 2. The operator matched by AI model 1 is operator 1, and the operator matched by AI model 2 is operator 2. If the current operator of the first device is operator 1, the target AI model to be activated may be AI model 1.

[0161] Example a9: The second information indicates the power consumption and accuracy of the M AI models. The target AI models to be activated may be the N AI models with the smallest function values ​​of power consumption and accuracy among the P AI models. Exemplarily, the function values ​​of power consumption and accuracy may be weighted function values ​​of power consumption and accuracy. The weights of power consumption and accuracy may be pre-configured, for example, specified by a protocol; or, may be determined by the first device; or, may be notified to the first device by other devices (for example, the second device). In this way, the first device may choose to run an AI model with low energy consumption and high accuracy as the target AI model to be activated, thereby improving performance.

[0162] Through method a2, the first device can determine that the target AI models to be activated include N AI models based on the status of M AI models and the first information, so that appropriate N AI models can be selected to implement the function of the first functional module, thereby improving the performance of implementing the function of the first functional module.

[0163] In some possible methods, method a2 can be replaced by: the first device determines, based on the second information, that the target AI model to be activated includes N AI models. Exemplarily, the first device includes M AI models that can implement the function of the first functional module, where M is a positive integer greater than or equal to N. The first device can select N AI models from the M AI models based on the second information. For details, please refer to step C2, except that P AI models are replaced by M AI models, which will not be repeated here.

[0164] In some possible ways, after activating N AI models, the first device may send the output results of the AI ​​models, and there may be multiple ways of sending, for example, way d1, way d2 or way d3.

[0165] Mode d1: N is 1; that is, the first device activates an AI model. The first device may send the output result of the AI ​​model; correspondingly, the second device receives the output result of the AI ​​model. The output result of the AI ​​model may be carried in a traditional message or in a new message, and this application does not limit this.

[0166] Mode d2: N is greater than or equal to 2; that is, the first device activates multiple AI models. The first device sends a first output result; correspondingly, the second device receives the first output result. The first output result is the output result of some AI models among the N AI models, and the some AI models are AI models among the N AI models that meet the first condition.

[0167] Exemplarily, the first condition includes at least one of conditions 1 to 3:

[0168] Condition 1: The part of the AI ​​models is the most credible AI model among the N AI models. For example, the N AI models include AI model 1 and AI model 2. The credibility of AI model 1 is higher than that of AI model 2. The part of the AI ​​models is AI model 1. In this way, the first output result sent by the first device is the output result of AI model 1, so that the second device can obtain the output result of the AI ​​model with high credibility.

[0169] Condition 2: The part of the AI ​​models is the AI ​​model with the highest accuracy among the N AI models. For example, the N AI models include AI model 1 and AI model 2. The accuracy of AI model 1 is higher than the accuracy of AI model 2. The part of the AI ​​models is AI model 1. In this way, the first output result sent by the first device is the output result of AI model 1, so that the second device can obtain the output result of the AI ​​model with high accuracy.

[0170] Condition 3: This part of the AI ​​models is the AI ​​model whose output result is the most similar to the last measurement result of the first device among the N AI models. For example, the N AI models include AI model 1 and AI model 2. The last measurement result of the first device is measurement result 1. If the similarity between the output result of AI model 1 and measurement result 1 is higher than the similarity between the output result of AI model 2 and measurement result 1, then this part of the AI ​​models is AI model 1. When the similarity between the output result of an AI model and the last measurement result is high, the accuracy of the output result of the AI ​​model is usually high. In this way, the first output result sent by the first device is the output result of AI model 1, so that the second device can obtain the output result of an AI model with high accuracy.

[0171] The similarity can also be called the degree of coincidence. There are multiple evaluation criteria for similarity. For example, the smaller the mean square error between the output result of a certain AI model and the last measurement result of the first device, the higher the similarity between the output result of the AI ​​model and the last measurement result of the first device.

[0172] Through this method d2, the first device can activate multiple AI models corresponding to the first functional module, and the first output result sent by the first device is the output result of the AI ​​model among the multiple AI models that meets the first condition, thereby improving the accuracy and / or credibility of the first output result.

[0173] Mode d3: The first device sends a second output result; correspondingly, the second device receives the second output result. The second output result is related to the output results of the N AI models.

[0174] Optionally, the second output result is a weighted sum of the output results of N AI models. For example:

[0175] output=w_1*output_1+w_2*output_2+…+w_N*output_N.

[0176] Wherein, output is the second output result, w_i is the weight corresponding to the i-th AI model among the N AI models, output_i is the output result of the i-th AI model among the N AI models, and i is an integer ranging from 1 to N. Wherein, w_i may be preset, for example, specified by a protocol; or, it may be determined by the first device; or, it may be notified to the first device by another device (for example, the second device).

[0177] Through this method d3, the first device can activate multiple AI models corresponding to the first functional module, and the second output result sent by the first device is related to the output results of the multiple AI models, thereby improving the accuracy and / or credibility of the second output result.

[0178] It should be understood that Figure 3 The method shown is explained using the activation of AI models and functional modules as an example. Figure 3 The method shown can also be applied to other operations in the LCM process, for example, it can be applied to at least one of the following operations in the LCM process: deactivation, switching, updating, transmission or rollback of AI models and / or functional modules. In this case, Figure 3 The activation in the method shown can be replaced by other operations in the LCM process.

[0179] It should also be understood that Figure 3 The method shown is illustrated using a scenario involving an AI model as an example. Figure 3 The method shown can also be applied to the following scenario: the second device instructs the first device to implement the first function, and the first device has multiple ways to implement the first function. In this case, Figure 3 The activation of the first functional module in the method shown can be replaced by implementing the first function, and the AI ​​model can be replaced by a way to implement the first function.

[0180] Based on the same technical concept as the above method embodiment, the embodiment of the present application provides a corresponding communication device, which can be used to perform the functions of the relevant steps in the above method embodiment. The function can be implemented by hardware, can be implemented by software, or can be implemented by hardware executing the corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The communication device can be a terminal, or can be a module in a terminal (such as a circuit or a chip), or can be a logical node, a logical module or software that can realize all or part of the functions of a terminal or an access network device; or the communication device can be an access network device or a module in an access network device (such as a circuit or a chip), or can be a logical node, a logical module or software that can realize all or part of the functions of an access network device.

[0181] In a possible implementation, the structure of the communication device provided in the embodiment of the present application is as follows: Figure 4 As shown, the communication device 400 includes a processing unit 402. Optionally, the communication device also includes an interface unit 401. The functions of each unit in the communication device 400 are introduced below.

[0182] The interface unit 401 is used to input and / or output information. The input information can be replaced by receiving information, and the output information can be replaced by sending information. When outputting information, the interface unit 401 can output information to other devices outside the communication device 400, or it can output information to other units in the communication device 400. In some embodiments, the interface unit 401 can be implemented by at least one of a physical interface, a communication module, a communication interface, and an input and output interface. In other embodiments, the interface unit 401 can be implemented by an interface circuit, for example, a mobile communication module. Among them, the mobile communication module may include one or more of at least one antenna, at least one filter, a switch, a power amplifier, a low noise amplifier (LNA), etc.

[0183] The processing unit 402 can be used to support the communication device 400 to perform the processing actions in the above method embodiment. The processing unit 402 can be implemented by one or more processors. For example, the processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor can be a microprocessor or any conventional processor.

[0184] In one embodiment, the communication device 400 is applied to Figure 3 The first device in the embodiment of the present application is shown. The specific functions of the processing unit 402 in this embodiment are introduced below.

[0185] Processing unit 402 is used to: receive first indication information through interface unit 401, the first indication information is used to indicate activation of the first functional module; receive first information through interface unit 401, the first information is used to determine that the target AI model to be activated includes N AI models, each of the N AI models is used to implement the function of the first functional module, N is a positive integer; activate the N AI models according to the first information.

[0186] In some possible embodiments, the processing unit 402 is further used to: determine the available resources of the communication device 400, the available resources and the first information are used to determine that the target AI model includes N AI models, and the communication device 400 is a device for activating N AI models.

[0187] In some other possible ways, the processing unit 402 is also used to: determine the second information, and the second information and the first information are used to determine that the target AI model includes N AI models. The second information is used to indicate at least one of the following of the M AI models: credibility, accuracy, complexity, power consumption, type of input information, size of input information, type of output information, or size of output information. Wherein, M is a positive integer greater than or equal to N, and the M AI models are AI models included in the communication device 400 that can realize the function of the first functional module.

[0188] In some examples, the processing unit 402 is also used to: send a first output result through the interface unit 401, the first output result is the output result of some AI models among N AI models, the some AI models are AI models among the N AI models that meet the first condition, and N is greater than or equal to 2.

[0189] In other examples, the processing unit 402 is further used to: send a second output result through the interface unit 401, and the second output result is related to the output results of N AI models, where N is greater than or equal to 2.

[0190] In another embodiment, the communication device 400 is applied to Figure 3 The second device in the embodiment of the present application is shown. The specific functions of the processing unit 402 in this embodiment are introduced below.

[0191] Processing unit 402 is used to: send first indication information through interface unit 401, the first indication information is used to indicate the activation of the first functional module; send first information through interface unit 401, the first information is used to determine that the target artificial intelligence AI model to be activated includes N AI models, the first information is used to activate the N AI models, each of the N AI models is used to implement the function of the first functional module, and N is a positive integer.

[0192] In some possible embodiments, the processing unit 402 is also used to: receive a first output result through the interface unit 401, the first output result being the output result of some AI models among the N AI models, the some AI models being the AI ​​models among the N AI models that meet the first condition, and N is greater than or equal to 2.

[0193] In some other possible embodiments, the processing unit 402 is further used to: receive a second output result through the interface unit 401, and the second output result is related to the output results of N AI models, where N is greater than or equal to 2.

[0194] For more detailed description of the processing unit 402 and the interface unit 401, please refer to Figure 3 The relevant description in the method embodiment shown is directly obtained and will not be repeated here.

[0195] It should be noted that the division of modules in the above embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional unit in each embodiment of the present application may be integrated into a processing unit, or may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0196] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0197] In a possible implementation, the communication device provided in the embodiment of the present application refers to Figure 5 As shown, the communication device 500 includes: a processor 502. Optionally, the communication device 500 also includes: an interface circuit 501 and a memory 503. The interface circuit 501, the processor 502 and the memory 503 are coupled to each other.

[0198] Optionally, the interface circuit 501, the processor 502 and the memory 503 are coupled to each other via a bus 504. The bus 504 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0199] The interface circuit 501 is used to input and / or output information. The input information can be replaced by receiving information, and the output information can be replaced by sending information. When outputting information, the interface circuit 501 can output information to other devices outside the communication device 500, or can output information to other units in the communication device 500. Exemplarily, the interface circuit 501 can be implemented by at least one of a physical interface, a communication module, a communication interface, an input-output interface, and a mobile communication module. Among them, the mobile communication module may include one or more of at least one antenna, at least one filter, a switch, a power amplifier, an LNA, etc.

[0200] The processor 502 may be used to support the communication device 500 in executing the processing actions in the above method embodiment. When the communication device 500 is used to implement the above method embodiment, the processor 502 may also be used to implement the functions of the above processing unit 402. The processor 502 may be a CPU, or other general-purpose processors, DSPs, ASICs, FPGAs, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor may be a microprocessor, or any conventional processor.

[0201] In one embodiment, the communication device 500 is applied to Figure 3 The first device in the embodiment of the present application is shown. The specific functions of the processor 502 in this implementation are introduced below.

[0202] Processor 502 is used to: receive first indication information through interface circuit 501, the first indication information is used to indicate activation of a first functional module; receive first information through interface circuit 501, the first information is used to determine that the target AI model to be activated includes N AI models, each of the N AI models is used to implement the function of the first functional module, N is a positive integer; activate the N AI models according to the first information.

[0203] In another embodiment, the communication device 500 is applied to Figure 3 The second device in the embodiment of the present application is shown. The specific functions of the processor 502 in this implementation are introduced below.

[0204] Processor 502 is used to: send first indication information through interface circuit 501, the first indication information is used to indicate activation of a first functional module; send first information through interface circuit 501, the first information is used to determine that the target artificial intelligence AI model to be activated includes N AI models, the first information is used to activate the N AI models, each of the N AI models is used to implement the function of the first functional module, and N is a positive integer.

[0205] The specific functions of the processor 502 can refer to the description of the communication method provided in the above embodiments and examples of the present application, and Figure 4 The specific functional description of the communication device 400 in the embodiment of the present application is shown and will not be repeated here.

[0206] The memory 503 is used to store program instructions and / or data, etc. Specifically, the program instructions may include a program code, which includes a computer operation instruction. The memory 503 may include a RAM, and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. The processor 502 executes the program instructions stored in the memory 503, and uses the data stored in the memory 503 to implement the above functions, thereby implementing the communication method provided in the above embodiment of the present application. The memory 503 can be integrated with the processor 502, or it can be a memory outside the communication device.

[0207] It is understandable that this application Figure 5 The memory 503 in can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a ROM, a programmable read-only memory (programmable ROM, PROM), an erasable programmable read-only memory (erasable PROM, EPROM), an electrically erasable programmable read-only memory (electrically EPROM, EEPROM) or a flash memory. The volatile memory can be a RAM, which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (static RAM, SRAM), dynamic random access memory (dynamic RAM, DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (doubledata rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhancedSDRAM, ESDRAM), synchronous connection dynamic random access memory (synchlink DRAM, SLDRAM) and direct memory bus random access memory (direct rambus RAM, DR RAM). It should be noted that memory of the systems and methods described herein is intended to comprise, without being limited to, these and any other suitable types of memory.

[0208] Based on the above embodiments, the embodiments of the present application further provide a computer program product including computer executable instructions. When the computer program product is run, the method provided in the above embodiments is executed.

[0209] Based on the above embodiments, the embodiments of the present application further provide a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a computer, the computer executes the method provided in the above embodiments.

[0210] The storage medium may be any available medium that can be accessed by a computer. For example, but not limited to, a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0211] Based on the above embodiments, an embodiment of the present application further provides a chip, which is used to read a computer program stored in a memory to implement the method provided in the above embodiments.

[0212] Based on the above embodiments, the embodiments of the present application provide a chip system, which includes a processor for supporting a computer device to implement the functions involved in each device in the above embodiments. In a possible design, the chip system also includes a memory for storing the necessary programs and data for the computer device. The chip system can be composed of a chip, or it can include a chip and other discrete devices.

[0213] In the various embodiments of the present application, unless otherwise specified or provided for in any logical conflict, the terms and / or descriptions between the different embodiments are consistent and may be referenced to each other, and the technical features in the different embodiments may be combined to form new embodiments according to their inherent logical relationships.

[0214] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0215] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0216] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0217] In this application, "at least one" means one or more, and "more" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the text description of this application, the character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0218] It is understood that the various numbers involved in the embodiments of the present application are only for the convenience of description and are not used to limit the scope of the embodiments of the present application. The size of the sequence number of the above-mentioned processes does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic.

[0219] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A communication method, characterized in that: include: receiving first indication information, where the first indication information is used to indicate activation of a first functional module; Receive first information, where the first information is used to determine that the target artificial intelligence (AI) model to be activated includes N AI models, each of the N AI models is used to implement the function of the first functional module, and N is a positive integer; According to the first information, the N AI models are activated.

2. The method according to claim 1, characterized in that The first information is used to indicate at least one of the following: Input parameters of the target AI model; Output parameters of the target AI model; The version number corresponding to the target AI model; The format of the target AI model; The target AI model’s requirements for terminal capabilities; Equipment manufacturers of equipment that can apply the target AI model; Application scenarios of the target AI model; Requirements for the computational complexity of the target AI model; Requirements for the processing capabilities of the target AI model; The performance requirements of the target AI model; The range of target AI model sizes; or Whether to allow merging the output results of multiple target AI models.

3. The method according to claim 1 or 2, characterized in that Also includes: Determine available resources of the first device, where the available resources and the first information are used to determine that the target AI model includes the N AI models, and the first device is a device for activating the N AI models.

4. The method according to claim 3, characterized in that The available resources include at least one of the following: the available computing capacity of the first device; The available amount of power of the first device; or The available storage space of the first device.

5. The method according to any one of claims 1 to 4, characterized in that Also includes: Determine second information, where the second information and the first information are used to determine that the target AI model includes the N AI models, and the second information is used to indicate at least one of the following of the M AI models: Reliability, accuracy, complexity, power consumption, type of input information, size of input information, type of output information, or size of output information; Among them, M is a positive integer greater than or equal to N, the M AI models are AI models included in the first device that can realize the function of the first functional module, and the first device is a device that activates the N AI models.

6. The method according to any one of claims 1 to 5, characterized in that The method further comprises: Send a first output result, where the first output result is the output result of some AI models among the N AI models, where the some AI models are AI models among the N AI models that meet the first condition, and N is greater than or equal to 2.

7. The method according to claim 6, characterized in that The first condition includes at least one of the following: The partial AI models are the AI ​​models with the highest credibility among the N AI models; The partial AI models are the AI ​​models with the highest accuracy among the N AI models; or The partial AI models are AI models whose output results are most similar to the last measurement result of the first device among the N AI models.

8. The method according to any one of claims 1 to 5, characterized in that The method further comprises: Send a second output result, where the second output result is related to the output results of the N AI models, where N is greater than or equal to 2.

9. The method according to claim 8, characterized in that The second output result is a weighted sum of the output results of the N AI models.

10. A communication method, characterized in that: include: Sending first indication information, where the first indication information is used to indicate activation of a first functional module; Send first information, where the first information is used to determine that the target artificial intelligence AI model to be activated includes N AI models, and the first information is used to activate the N AI models, where each of the N AI models is used to implement the function of the first functional module, and N is a positive integer.

11. The method according to claim 10, characterized in that The first information is used to indicate at least one of the following: Input parameters of the target AI model; Output parameters of the target AI model; The version number corresponding to the target AI model; The format of the target AI model; The target AI model’s requirements for terminal capabilities; Equipment manufacturers of equipment that can apply the target AI model; Application scenarios of the target AI model; Requirements for the computational complexity of the target AI model; Requirements for the processing capabilities of the target AI model; The performance requirements of the target AI model; The range of target AI model sizes; or Whether to allow merging the output results of multiple target AI models.

12. The method according to claim 10 or 11, characterized in that The method further comprises: A first output result is received, where the first output result is the output result of some AI models among the N AI models, where the some AI models are AI models among the N AI models that meet a first condition, and N is greater than or equal to 2.

13. The method according to claim 12, characterized in that The first condition includes at least one of the following: The partial AI models are the AI ​​models with the highest credibility among the N AI models; The partial AI models are the AI ​​models with the highest accuracy among the N AI models; or The partial AI models are AI models whose output results are most similar to the last measurement result of the first device among the N AI models.

14. The method according to claim 10 or 11, characterized in that: The method further comprises: Receive a second output result, where the second output result is related to the output results of the N AI models, where N is greater than or equal to 2.

15. The method according to claim 14, characterized in that The second output result is a weighted sum of the output results of the N AI models.

16. A communication device, characterized in that: The method comprises a unit for executing the method according to any one of claims 1 to 9, or comprises a unit for executing the method according to any one of claims 10 to 15.

17. A communication device, characterized in that: The device comprises a processor, wherein the processor executes instructions so that the device executes the method according to any one of claims 1 to 9, or the device executes the method according to any one of claims 10 to 15.

18. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program or instructions. When the computer program or instructions are executed, the method according to any one of claims 1 to 15 is implemented.

19. A computer program product, characterized in that The computer program product comprises: a computer program code, and when the computer program code is executed, the method according to any one of claims 1 to 15 is implemented.

20. A communication system, characterized in that: include: A first device, configured to implement the method according to any one of claims 1 to 9; A second device is used to implement the method according to any one of claims 10-15.

Images