Information interaction method and device, and communication device

Abstract

The application discloses an information interaction method and device and communication equipment, and belongs to the technical field of communication. The information interaction method of the application comprises the following steps: a first communication equipment acquires target algorithm information, wherein the target algorithm comprises at least one of a neural network algorithm, an artificial intelligence algorithm and a machine learning algorithm; and the first communication equipment performs information interaction with at least one target communication equipment according to the target algorithm information.

Description

[0001] This application is a divisional application of the invention application filed on November 30, 2021, with application number 202111447307.2 and title "Information Interaction Method, Apparatus and Communication Equipment". Technical Field

[0002] This application belongs to the field of communication technology, specifically relating to an information interaction method, apparatus, and communication equipment. Background Technology

[0003] When Artificial Intelligence (AI) is applied to wireless communication systems, corresponding neural networks need to be run on the terminal. However, the network-side nodes do not know the terminal's neural network computing power, the time the terminal takes to run the neural network, etc., making it difficult to configure a neural network that matches the terminal's operating resources. This makes it difficult for the neural network model used by the terminal to match its operating resources. Summary of the Invention

[0004] This application provides an information interaction method, apparatus, and communication device that can solve the problem of how to ensure that the neural network model used by the terminal is compatible with its operating resources.

[0005] Firstly, an information exchange method is provided, including:

[0006] The first communication device acquires target algorithm information, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm; and interacts with at least one target communication device based on the target algorithm information.

[0007] Secondly, an information exchange method is provided, including:

[0008] The second communication device sends target algorithm information to the first communication device, the target algorithm including at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm; the second communication device interacts with the target communication device according to the target algorithm information, the target communication device including at least one of the first communication device and the third communication device.

[0009] Thirdly, an information exchange method is provided, including:

[0010] A third communication device acquires first interactive information sent by a target communication device. The first interactive information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the resource information of the target algorithm. The target communication device includes at least one of a second communication device and a first communication device.

[0011] The third communication device interacts with the target communication device based on the first interaction information.

[0012] Fourthly, an information interaction device is provided, comprising:

[0013] The first transceiver module is used to acquire target algorithm information, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm;

[0014] The first interaction module is used to interact with at least one target communication device based on the target algorithm information.

[0015] Fifthly, an information interaction device is provided, comprising:

[0016] The second transceiver module is used to send target algorithm information to the first communication device, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm;

[0017] The second interaction module is used to interact with the target communication device based on the target algorithm information, wherein the target communication device includes at least one of the first communication device and the third communication device.

[0018] Sixthly, an information interaction device is provided, comprising:

[0019] The first acquisition module is used to acquire first interaction information sent by the target communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the target algorithm. The target communication device includes at least one of a second communication device and a first communication device.

[0020] The third interaction module is used to interact with the target communication device based on the first interaction information.

[0021] A seventh aspect provides a communication device including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method as described in the first, second, or third aspect.

[0022] Eighthly, a communication device is provided, including a processor and a communication interface, wherein the communication interface is used to acquire target algorithm information, the target algorithm including at least one of a neural network algorithm, an artificial intelligence algorithm, and a machine learning algorithm; to interact with at least one target communication device based on the target algorithm information; or, to send target algorithm information to a first communication device, the target algorithm including at least one of a neural network algorithm, an artificial intelligence algorithm, and a machine learning algorithm; to interact with a target communication device based on the target algorithm information, the target communication device including at least one of a first communication device and a third communication device; or, to acquire first interaction information sent by the target communication device, the first interaction information including at least one of first processing resource information and target algorithm recommendation information, the first processing resource information being used to instruct the first communication device to run the target algorithm, the target communication device including at least one of a second communication device and a first communication device; and to interact with the target communication device based on the first interaction information.

[0023] Ninth aspect, a communication system is provided, comprising: a first communication device, a second communication device, and a third communication device, wherein the first communication device is configured to perform the steps of the information interaction method as described in the first aspect, the second communication device is configured to perform the steps of the information interaction method as described in the second aspect, and the third communication device is configured to perform the steps of the information interaction method as described in the third aspect.

[0024] In a tenth aspect, a readable storage medium is provided, on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method as described in the first, second, or third aspect.

[0025] Eleventhly, a chip is provided, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the steps of the methods described in the first, second, or third aspects.

[0026] In a twelfth aspect, a computer program / program product is provided, the computer program / program product being stored in a storage medium, the computer program / program product being executed by at least one processor to perform the steps of the method as described in the first aspect, the second aspect, or the third aspect.

[0027] In this embodiment of the application, the first communication device obtains target algorithm information, which includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm. The first communication device can determine the target algorithm used by the first communication device by interacting with at least one communication device based on the target algorithm information, such as the target neural network model used by the first communication device, thereby achieving the purpose of the neural network model used by the first communication device being a neural network model adapted to its resource capabilities. Attached Figure Description

[0028] Figure 1 This diagram illustrates the structure of a communication system to which embodiments of this application can be applied.

[0029] Figure 2 One of the flowcharts illustrating the information interaction method according to an embodiment of this application;

[0030] Figure 3 A second flowchart illustrating the information interaction method according to an embodiment of this application;

[0031] Figure 4 The third flowchart illustrating the information interaction method of an embodiment of this application;

[0032] Figure 5 One of the schematic diagrams of the information interaction device according to an embodiment of this application;

[0033] Figure 6 This is a structural block diagram of a communication device according to an embodiment of this application;

[0034] Figure 7 A second structural block diagram illustrating the communication device according to an embodiment of this application;

[0035] Figure 8 A second schematic diagram of the information interaction device according to an embodiment of this application;

[0036] Figure 9 The third schematic diagram of the information interaction device according to an embodiment of this application;

[0037] Figure 10 This is one of the structural block diagrams of the network-side device according to an embodiment of this application;

[0038] Figure 11 This is a second structural block diagram illustrating an embodiment of the network-side device of this application. Detailed Implementation

[0039] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0040] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0041] It is worth noting that the technologies described in this application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency Division Multiple Access (SC-FDMA), and other systems. The terms "system" and "network" in this application are often used interchangeably, and the described technologies can be used in the systems and radio technologies mentioned above, as well as in other systems and radio technologies. The following description describes New Radio (NR) systems for illustrative purposes, and NR terminology is used in most of the following description; however, these technologies can also be applied to applications beyond NR systems, such as 6th generation (6G) radio systems. th Generation 6G communication system.

[0042] Figure 1This diagram illustrates a block diagram of a wireless communication system applicable to embodiments of this application. The wireless communication system includes a terminal 11 and a network-side device 12. Terminal 11 can be a mobile phone, tablet computer, laptop computer (also known as a notebook computer), personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, vehicle-mounted device (VUE), pedestrian terminal (PUE), smart home (home devices with wireless communication capabilities, such as refrigerators, televisions, washing machines, or furniture), game console, personal computer (PC), ATM, or self-service machine, etc. Wearable devices include: smartwatches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart chains, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. It should be noted that the specific type of terminal 11 is not limited in this embodiment. Network-side equipment 12 may include access network equipment or core network equipment. Access network equipment 12 may also be referred to as radio access network equipment, radio access network (RAN), radio access network function, or radio access network unit. Access network equipment 12 may include base stations, WLAN access points, or WiFi nodes, etc. Base stations may be referred to as Node B, evolved Node B (eNB), access point, base transceiver station (BTS), radio base station, radio transceiver, Basic Service Set (BSS), Extended Service Set (ESS), home B node, home evolved B node, Transmitting Receiving Point (TRP), or any other suitable term in the field, as long as the same technical effect is achieved. The base station is not limited to specific technical terms. It should be noted that this application embodiment only uses a base station in an NR system as an example for description and does not limit the specific type of base station.Core network equipment may include, but is not limited to, at least one of the following: core network node, core network function, Mobility Management Entity (MME), Access and Mobility Management Function (AMF), Session Management Function (SMF), User Plane Function (UPF), Policy Control Function (PCF), Policy and Charging Rules Function (PCRF), Edge Application Server Discovery Function (EASDF), Unified Data Management (UDM), Unified Data Repository (UDR), Home Subscriber Server (HSS), Centralized network configuration (CNC), Network Repository Function (NRF), Network Exposure Function (NEF), Local NEF (or L-NEF), Binding Support Function (BSF), and Application Function. Function (AF), etc. It should be noted that the embodiments of this application only use the core network equipment in the NR system as an example for introduction, and do not limit the specific type of core network equipment.

[0043] The information interaction method provided in this application will be described in detail below with reference to the accompanying drawings, through some embodiments and application scenarios.

[0044] like Figure 2 As shown in the figure, this application provides an information interaction method, including:

[0045] Step 201: The first communication device acquires target algorithm information, which includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm.

[0046] The target algorithm information is obtained through a pre-agreed method, or the target algorithm information is configured by at least one of the communication devices. The target algorithm may specifically include training algorithms, inference algorithms, etc.

[0047] Optionally, the target algorithm information includes at least one of the following:

[0048] The target algorithm requires the type of activation function;

[0049] The target algorithm requires a specific type of operation.

[0050] The conversion relationship of available resources used between different operations required by the target algorithm;

[0051] The target algorithm requires that it support different specific operations or super-operations within the same time unit.

[0052] The target algorithm requires a combination relationship between different specific operations;

[0053] The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit.

[0054] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0055] The target algorithm requires the time needed to complete a specific operation to achieve a specific objective.

[0056] The target algorithm requires supercomputing information;

[0057] The target algorithm requires information about the neural network model.

[0058] Model size;

[0059] The number of quantization bits in the target algorithm model;

[0060] Number of model layers;

[0061] The maximum number of neurons in each layer of the model;

[0062] The target algorithm requires the runtime of each neural network algorithm.

[0063] The target algorithm requires the runtime of a combination of multiple neural network algorithms;

[0064] The target algorithm requires a specific time.

[0065] The target algorithm requires the target parameters of the target algorithm;

[0066] At least one expected time to run the target algorithm;

[0067] The input type of the target algorithm;

[0068] The preprocessing information corresponding to the input information of the target algorithm;

[0069] Different target parameters corresponding to neural network models of the same target algorithm type;

[0070] A neural network model library. This neural network model library may include at least one neural network model.

[0071] Optionally, the target parameter includes at least one of the following:

[0072] Activation function;

[0073] Number of floors;

[0074] Number of neurons;

[0075] Number of neurons per layer;

[0076] The connections between neurons;

[0077] The running time of the target algorithm.

[0078] Step 202: The first communication device interacts with at least one target communication device based on the target algorithm information.

[0079] Information interaction in this application embodiment may include receiving information and / or sending information.

[0080] In this embodiment of the application, the first communication device obtains target algorithm information, which includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm. The first communication device can determine the target algorithm used by the first communication device by interacting with at least one communication device based on the target algorithm information, such as the target neural network model used by the first communication device, thereby achieving the purpose of the neural network model used by the first communication device being a neural network model adapted to its resource capabilities.

[0081] Optionally, the first communication device interacts with at least one other communication device, including:

[0082] The first communication device sends first interaction information to at least one communication device based on the target algorithm information. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device on the resource information for running the target algorithm.

[0083] In one embodiment of this application, after the first communication device obtains the target algorithm information, it sends first interaction information to the second communication device based on the algorithm information. The second communication device further determines the neural network model used by the first communication device based on the first interaction information. In this way, through the interaction between the first communication device and the second communication device, the purpose of configuring a neural network model that is compatible with the neural network operation capability of the first communication device is achieved.

[0084] Optionally, the first communication device interacts with at least one other communication device, including:

[0085] The first communication device obtains at least one of a first indication information and a second indication information from at least one target communication device, wherein the second indication information is used to indicate the target algorithm used by the first communication device;

[0086] The first indication information includes at least one of the following:

[0087] Priority information for the target algorithm;

[0088] The time from configuration to activation of a neural network model;

[0089] The time required to switch between at least two neural network models;

[0090] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0091] Optionally, the first communication device acquires target algorithm information, including:

[0092] The target algorithm information is obtained based on at least one of the first indication information and the second indication information, wherein the second indication information is used to indicate the target algorithm used by the first communication device;

[0093] The first indication information includes at least one of the following:

[0094] Priority information for the target algorithm;

[0095] The time from configuration to activation of a neural network model;

[0096] The time required to switch between at least two neural network models;

[0097] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0098] Optionally, the first interactive information further includes at least one of the following:

[0099] The target algorithm's execution priority indication information, the execution priority of which can be pre-agreed or indicated by the first communication device or configured by other communication devices, for example, how to prioritize multiple different neural network models when resources are available;

[0100] The time from configuration to activation of a neural network model;

[0101] The time required to switch between at least two neural network models;

[0102] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0103] Optionally, the first processing resource information includes at least one of the following:

[0104] The activation function types supported by the first communication device include, but are not limited to, at least one of the following: sigmoid function, hyperbolic tangent function, rectified linear unit (ReLU), leaky-resolved linear unit (LReLU), ELU function, Gaussian error linear unit (GELU) function, and softmax function.

[0105] The types of operations supported by the first communication device;

[0106] The conversion relationship of available resources used between different operations supported by the first communication device;

[0107] Whether the first communication device supports different specific operations or super operations within the same time unit;

[0108] The combination relationship between different specific operations supported by the first communication device;

[0109] The first communication device supports the number of operations within a first target time period, where the first target time period includes at least one time unit; optionally, the unit of the time unit is at least one of a time unit, subcarrier interval, time slot, and frame. The aforementioned time unit can be milliseconds (ms) or seconds (s), etc. For example, reporting the number of operations supported within 1ms, 10ms, or a preset time set;

[0110] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time. For example, the number of operations supported in the first target time is N times the number of elements supported in the second target time, where N is calculated based on the relationship between the time lengths of the first and second target times. If N includes multiple values ​​(assuming the first target time includes time units corresponding to multiple time granularities), a value can be determined based on multiple values ​​according to predefined rules, such as selecting the minimum value, maximum value, or average value.

[0111] The time required for the first communication device to complete a specific operation for a specific target, where the specific target includes a target number (which may be agreed upon by a protocol or configured by the network), and the specific operation can be a single operation. When reporting, the time required for at least one operation to complete the specific target can be reported, or only the time required for at least one operation to complete the specific target can be reported, or the time required for different operations to complete the specific target can be reported separately. The specific operation can also be a combination of multiple operations. When reporting, the time required for at least one combination of operations to complete the specific target can be reported, or the time required for each combination of operations to complete the specific target can be reported separately. Furthermore, the time required for the first communication device to complete a specific operation for a specific target can also be determined according to a preset conversion rule, for example, based on the number of operations. Assuming the time required for the first communication device to complete a preset operation for the specific target is T0, and the time required to complete specific operation A... If the time required is M1 times T0, then the first communication device can report that the time required to complete a specific operation A is M1 times T0. For example, if the specific operation is a combination of operations, and the time required for the first communication device to complete a specific target combination of operations (including operation A1 and operation A2) is T1, the time required to complete specific operation A1 is M2 times T0, and the time required to complete specific operation A2 is M3 times T0, then T1 can be converted to T0 based on the smaller or larger value between M2 and M3. For example, if M2 is less than M3, then T1 can be converted to T0 equal to M2 times T0. For another example, if the specific operation includes multiple combinations of operations, assuming it includes operation combination 1 and operation combination 2, where the time required to complete operation combination 1 is M4 times T0, and the time required to complete operation combination 2 is M5 times T0, then the smaller or larger value between M4 and M5 can be selected to convert the time required to complete multiple operation combinations into a preset multiple of T0.

[0112] The first communication device supports supercomputing information, which includes supercomputing type and parameters corresponding to supercomputing type. For example, the supercomputing type is a specific combination of operations, such as convolution operation, and the corresponding parameters are kernel size and supercomputing capability. The supercomputing capability includes the time required to complete a specific supercomputing operation or the number of supercomputing operations completed per unit time.

[0113] The neural network model information supported by the first communication device includes the neural network type and the parameters corresponding to the neural network type. The neural network type can be Convolutional Neural Networks (CNN), transformer, MLP-Mixer, etc. The parameters corresponding to the neural network type include, but are not limited to, the number of layers supported, the maximum number of neurons in each layer, and the number of times a specific neural network inference is completed per unit time.

[0114] Model size, which indicates the scale of a neural network model, can be indicated by the number of model parameters. A larger number of model parameters indicates a larger scale of the neural network model.

[0115] The number of quantization bits in the target algorithm model, such as the number of quantization bits in the model parameters;

[0116] Number of model layers;

[0117] The maximum number of neurons in each layer of the model;

[0118] The running time of each neural network algorithm;

[0119] Running time of a combination of multiple neural network algorithms;

[0120] The target algorithm that the first communication device supports or does not support;

[0121] Whether the first communication device supports the target algorithm according to a predefined time;

[0122] The time required for the first communication device to run the target algorithm;

[0123] The expected time for the first communication device to run the target algorithm;

[0124] The target parameters of the target algorithm supported by the first communication device.

[0125] Wherein, the number of operations mentioned above may refer to the number of operations of the first objective operation, which includes at least one of the following:

[0126] Multiplication operation;

[0127] Addition operation;

[0128] Operations corresponding to nonlinear activation functions.

[0129] The operations corresponding to the aforementioned nonlinear activation functions can be categorized into one type of operation, and the first communication device reports the total number of operations corresponding to this type of operation to the second communication device; the operations corresponding to the aforementioned nonlinear activation functions can also be categorized into multiple types of operations, for example, the number of operations corresponding to each nonlinear activation function is reported separately; the operations corresponding to the aforementioned nonlinear activation functions can also be further divided into several subclasses, and the number of operations corresponding to each subclass is reported.

[0130] Specifically, the expected time for the first communication device to run the target algorithm includes the expected time for the first communication device to run the neural network model. Other communication devices besides the first communication device can configure their expected execution time according to each neural network model, and the first communication device indicates whether to support that expected time. Alternatively, other communication devices can configure multiple expected execution times for the neural network to the first communication device, and the first communication device indicates which expected time to support.

[0131] In addition, other communication devices are configured with the input types of neural networks and the preprocessing required for the input information.

[0132] The aforementioned expected time may or may not include the time required for preprocessing. In this embodiment, the expected time may be agreed upon in advance or indicated interactively whether it includes the time required for preprocessing.

[0133] Optionally, the number of operations supported by the first communication device within the first target time period includes at least one of the following:

[0134] The number of operations performed for each operation during the first target time period, for example, the number of operations for each operation of multiple nonlinear activation functions;

[0135] The terminal reports the number of operations performed simultaneously by at least two operations within a first target time period. This combination of at least two operations can be indicated by a first communication device, configured by a second communication device, or pre-agreed upon. The terminal can report the number of operations for each operation combination, or report the number of operations for multiple operation combinations simultaneously. The terminal can also report the number of operations for each operation in a combination individually, or add the number of operations for each operation in the combination before reporting, or convert the number of operations for at least two sets of operations in the combination into the number of operations corresponding to a preset operation before reporting. For example, an operation combination includes operation A and operation B. Operation A can complete N1 operations within the first target time period, and operation B can complete N2 operations within the first target time period. Assuming the preset operation can complete N0 operations within the first target time period, then according to the conversion ratio, both N1 and N2 are converted into a target multiple of N0, and then the reported number is based on the converted number. Of course, N1 can be converted into a certain multiple of N2, or N2 can be converted into a certain multiple of N1.

[0136] The total number of operations within the first target time period can be calculated by adding up the number of operations corresponding to each operation within the first target time period, or by converting the number of operations corresponding to each operation and the preset number of operations corresponding to preset operations based on conversion information, and then adding them together. The specific conversion method has been described above and will not be repeated here. The operations corresponding to this total number of operations can be any combination of operations or any combination of operations within a predefined set of operations.

[0137] Optionally, the unit of the time unit is a time unit, a subcarrier interval, a time slot, or a frame.

[0138] Optionally, the target algorithm recommendation information includes at least one of the following:

[0139] The first communication device provides recommendation information for the target parameters of the target algorithm;

[0140] The first communication device provides revision information for the target parameters of the target algorithm;

[0141] Information on whether the first communication device supports the target algorithm.

[0142] Optionally, the first processing resource information further includes at least one of the following:

[0143] The first communication device provides recommendation information for the target parameters of the target algorithm;

[0144] The first communication device supports the pre-agreed target algorithm.

[0145] Optionally, the target parameter includes at least one of the following:

[0146] Activation function;

[0147] Number of floors;

[0148] Number of neurons;

[0149] Number of neurons per layer;

[0150] The connections between neurons;

[0151] The running time of the target algorithm, such as the running time of a neural network model.

[0152] In this embodiment, other communication devices can configure multiple different options of the same neural network type to the first communication device (these options can also be described as the target parameters mentioned above). For example, configuring the same type of neural network model with different activation functions, or configuring the same type of neural network model with different processing power requirements, or configuring the same type of neural network model with different numbers of layers, or configuring the same type of neural network model with different numbers of neurons, or configuring the same type of neural network model with different neuron connection relationships. The first communication device indicates to the second communication device the activation functions, number of layers, number of neurons, number of neurons per layer, neuron connection relationships, etc., supported by the first communication device in the same type of neural network model. In addition, the above-mentioned first neural network model recommendation information also includes suggestions on the running time of the neural network model and whether the corresponding running time is supported.

[0153] Optionally, the permitted revision information for the above target parameters can be determined through agreement or information exchange. For example, the above layer number can be modified to C1 or C2.

[0154] In one specific embodiment of this application, a first communication device sends first processing resource information to a second communication device, the second communication device configures neural network model information based on the first processing resource information, the first communication device further indicates to the second communication device whether it supports the neural network model or provides corresponding recommendation information, the second communication device further configures the neural network model, and the first communication device can further provide instructions. In this way, through multiple rounds of interaction, the final neural network model used is determined.

[0155] Optionally, the first communication device interacts with at least one other communication device, including:

[0156] Based on the target algorithm information, the target algorithm used by the first communication device is determined;

[0157] Send the target algorithm used by the first communication device to at least one target communication device.

[0158] Here, after the first communication device obtains the target algorithm information, it can select the target algorithm to be used by the first communication device according to the configuration information and notify other communication devices so that other communication devices can refer to it when configuring the target algorithm for the first communication device in the future.

[0159] Optionally, the method in this application embodiment further includes:

[0160] When a pre-configured triggering condition is met, the first communication device sends at least one of the following: first processing resource information, target algorithm recommendation information, second interaction information, and second processing resource information. The second interaction information is used to indicate the shared capability of multiple target algorithms to the NPU, and the second processing resource information is used to indicate the resource information of the first communication device for running the target algorithm.

[0161] Optionally, the triggering condition includes at least one of the following:

[0162] The available resources of the first communication device are less than the preset resource amount;

[0163] The available resources of the first communication device cannot support the operation of the indicated target algorithm.

[0164] The first communication device cannot process the information in the manner specified in the first processing information or the second interactive information.

[0165] The available resources of the first communication device have changed, or the range of change of available resources exceeds a certain threshold;

[0166] The first communication device needs to run multiple target algorithms simultaneously.

[0167] Optionally, the first processing resource information or the second interaction information further includes at least one of the following:

[0168] Whether at least two target algorithms can run simultaneously; these at least two target algorithms can belong to the same algorithm type or different algorithm types;

[0169] A combination of target algorithms that can run simultaneously;

[0170] Combinations of target algorithms that cannot run simultaneously;

[0171] The running time of each target algorithm in the combination of target algorithms running simultaneously;

[0172] The first communication device runs at least two target algorithms in parallel, corresponding to their running times.

[0173] The priority of the target algorithm.

[0174] Optionally, the method in this application embodiment further includes:

[0175] The first communication device obtains update indication information from at least one target communication device, the update indication information being used to indicate an update to the target algorithm information;

[0176] The target algorithm information is updated according to the update instruction information.

[0177] For example, the neural network model information is updated according to the update instruction information mentioned above.

[0178] In this embodiment, the first communication device can interact with other communication devices to exchange various combinations of processing resource information, which can also be described as a combination of capability information. For example, one combination of processing resource information is suitable for high-performance scenarios, while another combination is suitable for low-power scenarios. The first communication device can determine the current combination of processing resource information based on its own state. The aforementioned combination of resource information may include at least one of the first processing resource information and the first interaction information.

[0179] In addition, the first processing resource information, target algorithm information, first interaction information, second processing resource information, second interaction information, first indication information, second indication information, or update indication information in the embodiments of this application can be sent through at least one of the following: Radio Resource Control (RRC) message, Media Access Control Unit (MAC CE) message, Uplink Control Information (UCI) message, and Non-Access Stratum (NAS) message.

[0180] In this embodiment, the first communication device can be a terminal or a network node. Other communication devices besides the first communication device can be core network nodes, such as Network Data Analysis Function (NWDAF) nodes or neural network processing nodes. These other communication devices can also be base stations or newly defined neural network processing nodes. Furthermore, these other communication devices can be a combination of multiple network nodes. For example, configuration information comes from the first network node; the first communication device reports first interaction information to the first or second network node based on the configuration information; the third network node performs a second round of configuration on the first communication device; and the first communication device indicates the final model support information to the third or fourth network node.

[0181] like Figure 3 As shown in the embodiments of this application, an information interaction method is also provided, including:

[0182] Step 301: The second communication device sends target algorithm information to the first communication device, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm.

[0183] The target algorithm may specifically include training algorithms, inference algorithms, etc.

[0184] Optionally, the target algorithm information includes at least one of the following:

[0185] The target algorithm requires the type of activation function;

[0186] The target algorithm requires a specific type of operation.

[0187] The conversion relationship of available resources used between different operations required by the target algorithm;

[0188] The target algorithm requires that it support different specific operations or super-operations within the same time unit.

[0189] The target algorithm requires a combination relationship between different specific operations;

[0190] The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit.

[0191] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0192] The target algorithm requires the time needed to complete a specific operation to achieve a specific objective.

[0193] The target algorithm requires supercomputing information;

[0194] The target algorithm requires information about the neural network model.

[0195] Model size;

[0196] The number of quantization bits in the target algorithm model;

[0197] Number of model layers;

[0198] The maximum number of neurons in each layer of the model;

[0199] The target algorithm requires the runtime of each neural network algorithm.

[0200] The target algorithm requires the runtime of a combination of multiple neural network algorithms;

[0201] The target algorithm requires a specific time.

[0202] The target algorithm requires the target parameters of the target algorithm;

[0203] At least one expected time to run the target algorithm;

[0204] The input type of the target algorithm;

[0205] The preprocessing information corresponding to the input information of the target algorithm;

[0206] Different target parameters corresponding to neural network models of the same target algorithm type;

[0207] A neural network model library. This neural network model library may include at least one neural network model.

[0208] Optionally, the target parameter includes at least one of the following:

[0209] Activation function;

[0210] Number of floors;

[0211] Number of neurons;

[0212] Number of neurons per layer;

[0213] The connections between neurons;

[0214] The running time of the target algorithm.

[0215] Step 302: The second communication device interacts with the target communication device based on the target algorithm information, wherein the target communication device includes at least one of the first communication device and the third communication device.

[0216] In this embodiment, the second communication device sends target algorithm information to the first communication device. The target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm. Subsequently, based on the target algorithm information, the communication devices can interact with each other to determine the target algorithm used by the first communication device, such as the target neural network model used by the first communication device. This achieves the goal of making the neural network model used by the first communication device compatible with its resource capabilities.

[0217] Optionally, the method in this application embodiment further includes:

[0218] The second communication device acquires first interaction information sent by the target communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the target algorithm.

[0219] Optionally, the method in this application embodiment further includes:

[0220] Send the first instruction message;

[0221] The first indication information includes at least one of the following:

[0222] Priority information for the target algorithm;

[0223] The time from configuration to activation of a neural network model;

[0224] The time required to switch between at least two neural network models;

[0225] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0226] Optionally, the method in this application embodiment further includes:

[0227] Send a second indication message, which is used to indicate the target algorithm used by the first communication device.

[0228] Optionally, the method in this application embodiment further includes:

[0229] Send update instruction information, which is used to instruct the configuration information to be updated.

[0230] Optionally, the method in this application embodiment further includes:

[0231] At least one of second interaction information and second processing resource information is obtained from the target communication device, wherein the second interaction information is used to indicate the shared capability of multiple target algorithms to the NPU, and the second processing resource information is used to indicate the resource information of the first communication device for running the target algorithm.

[0232] Optionally, the first processing resource information includes at least one of the following:

[0233] The activation function types supported by the first communication device;

[0234] The types of operations supported by the first communication device;

[0235] The conversion relationship of available resources used between different operations supported by the first communication device;

[0236] Whether the first communication device supports different specific operations or super operations within the same time unit;

[0237] The combination relationship between different specific operations supported by the first communication device;

[0238] The number of operations supported by the first communication device within a first target time period, wherein the first target time period includes at least one time unit;

[0239] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0240] The time required for the first communication device to complete a specific operation for a specific target;

[0241] The supercomputing information supported by the first communication device;

[0242] Information on the neural network model supported by the first communication device;

[0243] Model size;

[0244] The number of quantization bits in the target algorithm model;

[0245] Number of model layers;

[0246] The maximum number of neurons in each layer of the model;

[0247] The running time of each neural network algorithm;

[0248] Running time of a combination of multiple neural network algorithms;

[0249] The target algorithm that the first communication device supports or does not support;

[0250] Whether the first communication device supports the target algorithm according to a predefined time;

[0251] The time required for the first communication device to run the target algorithm;

[0252] The expected time for the first communication device to run the target algorithm;

[0253] The target parameters of the target algorithm supported by the first communication device.

[0254] Optionally, the target algorithm recommendation information includes at least one of the following:

[0255] The first communication device provides recommendation information for the target parameters of the target algorithm;

[0256] The first communication device provides revision information for the target parameters of the target algorithm;

[0257] Information on whether the first communication device supports the target algorithm.

[0258] Optionally, the first processing resource information or the second interaction information further includes at least one of the following:

[0259] Can at least two objective algorithms run simultaneously?

[0260] A combination of target algorithms that can run simultaneously;

[0261] Combinations of target algorithms that cannot run simultaneously;

[0262] The running time of each target algorithm in the combination of target algorithms running simultaneously;

[0263] The first communication device runs at least two target algorithms in parallel, corresponding to their running times.

[0264] The priority of the target algorithm.

[0265] Optionally, the target parameter includes at least one of the following:

[0266] Activation function;

[0267] Number of floors;

[0268] Number of neurons;

[0269] Number of neurons per layer;

[0270] The connections between neurons;

[0271] The running time of the target algorithm.

[0272] In this embodiment, the second communication device sends target algorithm information to the first communication device. The target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm. Subsequently, based on the target algorithm information, the communication devices can interact with each other to determine the target algorithm used by the first communication device, such as the target neural network model used by the first communication device. This achieves the goal of making the neural network model used by the first communication device compatible with its resource capabilities.

[0273] like Figure 4 As shown in the embodiments of this application, an information interaction method is also provided, including:

[0274] Step 401: The third communication device obtains the first interaction information sent by the target communication device. The first interaction information includes at least one of the first processing resource information and the target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the target algorithm. The target communication device includes at least one of the second communication device and the first communication device.

[0275] The first interactive information has been described in detail in the above method embodiments, and will not be repeated here.

[0276] Step 402: The third communication device interacts with the target communication device based on the first interaction information.

[0277] Optionally, the information exchanged includes at least one of the following:

[0278] Priority information for the target algorithm;

[0279] The time from configuration to activation of a neural network model;

[0280] The time required to switch between at least two neural network models;

[0281] Can the corresponding computations be performed during the switching between at least two neural network models?

[0282] Update instruction information, which is used to indicate that the configuration information of the target algorithm should be updated;

[0283] The target algorithm used by the first communication device.

[0284] In the method of this application embodiment, after the third communication device obtains the first interaction information, it feeds back to the first communication device or other communication devices the target algorithm used by the first communication device, the running priority indication information of the target algorithm, the time from configuration to activation of the neural network model, the time required to switch between at least two neural network models, etc., so that the first communication device can use the target algorithm adapted to its running resources.

[0285] In one specific embodiment of this application, the second communication device sends target algorithm information to the first communication device, the first communication device sends first interaction information to the second communication device, and the second communication device instructs the first communication device on the target neural network model used by the first communication device. Alternatively, in another embodiment of this application, the second communication device sends target algorithm information to the first communication device, the first communication device sends first interaction information to both the second and third communication devices, and the third communication device instructs the first communication device on the target neural network model used by the first communication device.

[0286] The information interaction method provided in this application can be executed by an information interaction device. This application uses an information interaction device to perform the information interaction method as an example to illustrate the information interaction device provided in this application.

[0287] like Figure 5 As shown in the figure, this application embodiment also provides an information interaction device 500, including:

[0288] The first transceiver module 501 is used to acquire target algorithm information, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm;

[0289] The first interaction module 502 is used to interact with at least one communication device based on the target algorithm information.

[0290] Optionally, the target algorithm information is obtained through a pre-agreed method, or the target algorithm information is configured by at least one of the communication devices.

[0291] Optionally, the first interaction module is configured to send first interaction information to at least one communication device based on the target algorithm information. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device on the resource information for running the target algorithm.

[0292] Optionally, the first interaction module is used for the first communication device to obtain at least one of a first indication information and a second indication information from at least one target communication device, wherein the second indication information is used to indicate the target algorithm used by the first communication device;

[0293] The first indication information includes at least one of the following:

[0294] Priority information for the target algorithm;

[0295] The time from configuration to activation of a neural network model;

[0296] The time required to switch between at least two neural network models;

[0297] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0298] Optionally, the first transceiver module is configured to obtain the target algorithm information based on at least one of the first indication information and the second indication information, wherein the second indication information is used to indicate the target algorithm used by the first communication device;

[0299] The first indication information includes at least one of the following:

[0300] Priority information for the target algorithm;

[0301] The time from configuration to activation of a neural network model;

[0302] The time required to switch between at least two neural network models;

[0303] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0304] Optionally, the first interactive information further includes at least one of the following:

[0305] Priority information for the target algorithm;

[0306] The time from configuration to activation of a neural network model;

[0307] The time required to switch between at least two neural network models;

[0308] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0309] Optionally, the first processing resource information includes at least one of the following:

[0310] The activation function types supported by the first communication device;

[0311] The types of operations supported by the first communication device;

[0312] The conversion relationship of available resources used between different operations supported by the first communication device;

[0313] Whether the first communication device supports different specific operations or super operations within the same time unit;

[0314] The combination relationship between different specific operations supported by the first communication device;

[0315] The number of operations supported by the first communication device within a first target time period, wherein the first target time period includes at least one time unit;

[0316] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0317] The time required for the first communication device to complete a specific operation for a specific target;

[0318] The supercomputing information supported by the first communication device;

[0319] Information on the neural network model supported by the first communication device;

[0320] Model size;

[0321] The number of quantization bits in the target algorithm model;

[0322] Number of model layers;

[0323] The maximum number of neurons in each layer of the model;

[0324] The running time of each neural network algorithm;

[0325] Running time of a combination of multiple neural network algorithms;

[0326] The target algorithm that the first communication device supports or does not support;

[0327] Whether the first communication device supports the target algorithm according to a predefined time;

[0328] The time required for the first communication device to run the target algorithm;

[0329] The expected time for the first communication device to run the target algorithm;

[0330] The target parameters of the target algorithm supported by the first communication device.

[0331] Optionally, the target algorithm information includes at least one of the following:

[0332] The target algorithm requires the type of activation function;

[0333] The target algorithm requires a specific type of operation.

[0334] The conversion relationship of available resources used between different operations required by the target algorithm;

[0335] The target algorithm requires that it support different specific operations or super-operations within the same time unit.

[0336] The target algorithm requires a combination relationship between different specific operations;

[0337] The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit.

[0338] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0339] The target algorithm requires the time needed to complete a specific operation to achieve a specific objective.

[0340] The target algorithm requires supercomputing information;

[0341] The target algorithm requires information about the neural network model.

[0342] Model size;

[0343] The number of quantization bits in the target algorithm model;

[0344] Number of model layers;

[0345] The maximum number of neurons in each layer of the model;

[0346] The target algorithm requires the runtime of each neural network algorithm.

[0347] The target algorithm requires the runtime of a combination of multiple neural network algorithms;

[0348] The target algorithm requires a specific time.

[0349] The target algorithm requires the target parameters of the target algorithm;

[0350] At least one expected time to run the target algorithm;

[0351] The input type of the target algorithm;

[0352] The preprocessing information corresponding to the input information of the target algorithm;

[0353] Different target parameters corresponding to neural network models of the same target algorithm type;

[0354] Neural network model library.

[0355] Optionally, the number of operations supported by the first communication device within the first target time period includes at least one of the following:

[0356] The number of operations performed for each type of operation during the first target time period;

[0357] The number of operations performed when at least two operations are executed simultaneously within the first target time period;

[0358] The total number of calculations within the first target time.

[0359] Optionally, the unit of the time unit is a time unit, a subcarrier interval, a time slot, or a frame.

[0360] Optionally, the target algorithm recommendation information includes at least one of the following:

[0361] The first communication device provides recommendation information for the target parameters of the target algorithm;

[0362] The first communication device provides revision information for the target parameters of the target algorithm;

[0363] Information on whether the first communication device supports the target algorithm.

[0364] Optionally, the first interaction module includes:

[0365] The first determining submodule is used to determine the target algorithm used by the first communication device based on the target algorithm information.

[0366] The first sending submodule is used to send the target algorithm used by the first communication device to at least one target communication device.

[0367] Optionally, the apparatus in this application embodiment further includes:

[0368] The third transceiver module is configured to, when a pre-configured triggering condition is met, send at least one of the following: first processing resource information, target algorithm recommendation information, second interaction information, and second processing resource information. The second interaction information is used to indicate the sharing capability of multiple target algorithms to the NPU, and the second processing resource information is used to indicate the resource information of the first communication device for running the target algorithm.

[0369] Optionally, the triggering condition includes at least one of the following:

[0370] The available resources of the first communication device are less than the preset resource amount;

[0371] The available resources of the first communication device cannot support the operation of the indicated target algorithm.

[0372] The first communication device cannot process the information in the manner specified in the first processing information or the second interactive information.

[0373] The available resources of the first communication device have changed, or the range of change of available resources exceeds a certain threshold;

[0374] The first communication device needs to run multiple target algorithms simultaneously.

[0375] Optionally, the first processing resource information or the second interaction information further includes at least one of the following:

[0376] Can at least two objective algorithms run simultaneously?

[0377] A combination of target algorithms that can run simultaneously;

[0378] Combinations of target algorithms that cannot run simultaneously;

[0379] The running time of each target algorithm in the combination of target algorithms running simultaneously;

[0380] The first communication device runs at least two target algorithms in parallel, corresponding to their running times.

[0381] The priority of the target algorithm.

[0382] Optionally, the apparatus in this application embodiment further includes:

[0383] The second acquisition module is used for the first communication device to acquire update indication information from at least one target communication device, wherein the update indication information is used to indicate that the target algorithm information is updated;

[0384] The update module is used to update the target algorithm information according to the update instruction information.

[0385] Optionally, the target parameter includes at least one of the following:

[0386] Activation function;

[0387] Number of floors;

[0388] Number of neurons;

[0389] Number of neurons per layer;

[0390] The connections between neurons;

[0391] The running time of the target algorithm.

[0392] The apparatus of this application embodiment acquires target algorithm information, which includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm. The first communication device can determine the target algorithm used by the first communication device by interacting with at least one communication device based on the target algorithm information, such as the target neural network model used by the first communication device, thereby achieving the purpose of the neural network model used by the first communication device being a neural network model adapted to its resource capabilities.

[0393] The information interaction device in this application embodiment can be an electronic device, such as an electronic device with an operating system, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the terminal can be, but is not limited to, the type of terminal 11 listed above, and other devices can be servers, network attached storage (NAS), etc., and this application embodiment does not specifically limit the type.

[0394] The information interaction device provided in this application embodiment can achieve... Figure 2 The various processes implemented in the method embodiments achieve the same technical effect, and will not be described again here to avoid repetition.

[0395] Optional, such as Figure 6As shown, this application embodiment also provides a communication device 600, including a processor 601 and a memory 602. The memory 602 stores a program or instructions that can run on the processor 601. For example, when the communication device 600 is a terminal, the program or instructions executed by the processor 601 implement the various steps of the information interaction method embodiment on the first communication device side described above, and can achieve the same technical effect. When the communication device 600 is a network-side device, the program or instructions executed by the processor 601 implement the various steps of the method embodiment on the second or third communication device side described above, and can achieve the same technical effect. To avoid repetition, further details are omitted here.

[0396] This application embodiment also provides a first communication device, which is a terminal, including a processor and a communication interface. The communication interface is used to send first processing resource information; the processor is used to acquire target algorithm information, the target algorithm including at least one of a neural network algorithm, an artificial intelligence algorithm, and a machine learning algorithm; and to interact with at least one communication device based on the target algorithm information. This embodiment corresponds to the above-described first communication device-side method embodiment. All implementation processes and methods of the above method embodiments can be applied to this embodiment and achieve the same technical effect. Specifically, Figure 7 A schematic diagram of the hardware structure of a terminal to implement an embodiment of this application.

[0397] The terminal 700 includes, but is not limited to, at least some of the following components: radio frequency unit 701, network module 702, audio output unit 703, input unit 704, sensor 705, display unit 706, user input unit 707, interface unit 708, memory 709, and processor 710.

[0398] Those skilled in the art will understand that the terminal 700 may also include a power supply (such as a battery) for supplying power to various components. The power supply may be logically connected to the processor 710 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system. Figure 7 The terminal structure shown does not constitute a limitation on the terminal. The terminal may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.

[0399] It should be understood that, in this embodiment, the input unit 704 may include a graphics processing unit (GPU) 7041 and a microphone 7042. The GPU 7041 processes image data of still images or videos obtained by an image capture device (such as a camera) in video capture mode or image capture mode. The display unit 706 may include a display panel 7061, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like. The user input unit 707 includes at least one of a touch panel 7071 and other input devices 7072. The touch panel 7071 is also called a touch screen. The touch panel 7071 may include two parts: a touch detection device and a touch controller. Other input devices 7072 may include, but are not limited to, a physical keyboard, function keys (such as volume control buttons, power buttons, etc.), a trackball, a mouse, and a joystick, which will not be described in detail here.

[0400] In this embodiment, after receiving downlink data from the network-side device, the radio frequency unit 701 can transmit it to the processor 710 for processing; in addition, the radio frequency unit 701 can send uplink data to the network-side device. Typically, the radio frequency unit 701 includes, but is not limited to, an antenna, amplifier, transceiver, coupler, low-noise amplifier, duplexer, etc.

[0401] The memory 709 can be used to store software programs or instructions, as well as various data. The memory 709 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, the memory 709 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM). The memory 709 in the embodiments of this application includes, but is not limited to, these and any other suitable types of memory.

[0402] Processor 710 may include one or more processing units; optionally, processor 710 integrates an application processor and a modem processor, wherein the application processor mainly handles operations involving the operating system, user interface, and applications, and the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into processor 710.

[0403] The radio frequency unit 701 is used to acquire target algorithm information, the target algorithm including at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm; and to interact with at least one communication device based on the target algorithm information.

[0404] Optionally, the target algorithm information is obtained through a pre-agreed method, or the target algorithm information is configured by at least one of the communication devices.

[0405] Optionally, the radio frequency unit 701 is used by the first communication device to send first interaction information to at least one communication device according to the target algorithm information. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device on the resource information for running the target algorithm.

[0406] Optionally, the radio frequency unit 701 is used for the first communication device to obtain at least one of a first indication information and a second indication information from at least one target communication device, wherein the second indication information is used to indicate the target algorithm used by the first communication device;

[0407] The first indication information includes at least one of the following:

[0408] Priority information for the target algorithm;

[0409] The time from configuration to activation of a neural network model;

[0410] The time required to switch between at least two neural network models;

[0411] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0412] Optionally, the radio frequency unit 701 is configured to obtain the target algorithm information according to at least one of the first indication information and the second indication information, wherein the second indication information is used to indicate the target algorithm used by the first communication device;

[0413] The first indication information includes at least one of the following:

[0414] Priority information for the target algorithm;

[0415] The time from configuration to activation of a neural network model;

[0416] The time required to switch between at least two neural network models;

[0417] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0418] Optionally, the first interactive information further includes at least one of the following:

[0419] Priority information for the target algorithm;

[0420] The time from configuration to activation of a neural network model;

[0421] The time required to switch between at least two neural network models;

[0422] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0423] Optionally, the first processing resource information includes at least one of the following:

[0424] The activation function types supported by the first communication device;

[0425] The types of operations supported by the first communication device;

[0426] The conversion relationship of available resources used between different operations supported by the first communication device;

[0427] Whether the first communication device supports different specific operations or super operations within the same time unit;

[0428] The combination relationship between different specific operations supported by the first communication device;

[0429] The number of operations supported by the first communication device within a first target time period, wherein the first target time period includes at least one time unit;

[0430] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0431] The time required for the first communication device to complete a specific operation for a specific target;

[0432] The supercomputing information supported by the first communication device;

[0433] Information on the neural network model supported by the first communication device;

[0434] Model size;

[0435] The number of quantization bits in the target algorithm model;

[0436] Number of model layers;

[0437] The maximum number of neurons in each layer of the model;

[0438] The running time of each neural network algorithm;

[0439] Running time of a combination of multiple neural network algorithms;

[0440] The target algorithm that the first communication device supports or does not support;

[0441] Whether the first communication device supports the target algorithm according to a predefined time;

[0442] The time required for the first communication device to run the target algorithm;

[0443] The expected time for the first communication device to run the target algorithm;

[0444] The target parameters of the target algorithm supported by the first communication device.

[0445] Optionally, the target algorithm information includes at least one of the following:

[0446] The target algorithm requires the type of activation function;

[0447] The target algorithm requires a specific type of operation.

[0448] The conversion relationship of available resources used between different operations required by the target algorithm;

[0449] The target algorithm requires that it support different specific operations or super-operations within the same time unit.

[0450] The target algorithm requires a combination relationship between different specific operations;

[0451] The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit.

[0452] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0453] The target algorithm requires the time needed to complete a specific operation to achieve a specific objective.

[0454] The target algorithm requires supercomputing information;

[0455] The target algorithm requires information about the neural network model.

[0456] Model size;

[0457] The number of quantization bits in the target algorithm model;

[0458] Number of model layers;

[0459] The maximum number of neurons in each layer of the model;

[0460] The target algorithm requires the runtime of each neural network algorithm.

[0461] The target algorithm requires the runtime of a combination of multiple neural network algorithms;

[0462] The target algorithm requires a specific time.

[0463] The target algorithm requires the target parameters of the target algorithm;

[0464] At least one expected time to run the target algorithm;

[0465] The input type of the target algorithm;

[0466] The preprocessing information corresponding to the input information of the target algorithm;

[0467] Different target parameters corresponding to neural network models of the same target algorithm type;

[0468] Neural network model library.

[0469] Optionally, the number of operations supported by the first communication device within the first target time period includes at least one of the following:

[0470] The number of operations performed for each type of operation during the first target time period;

[0471] The number of operations performed when at least two operations are executed simultaneously within the first target time period;

[0472] The total number of calculations within the first target time.

[0473] Optionally, the unit of the time unit is a time unit, a subcarrier interval, a time slot, or a frame.

[0474] Optionally, the target algorithm recommendation information includes at least one of the following:

[0475] The first communication device provides recommendation information for the target parameters of the target algorithm;

[0476] The first communication device provides revision information for the target parameters of the target algorithm;

[0477] Information on whether the first communication device supports the target algorithm.

[0478] Optionally, the radio frequency unit 701 is used to determine the target algorithm used by the first communication device based on the target algorithm information; and to send the target algorithm used by the first communication device to at least one target communication device.

[0479] Optionally, the radio frequency unit 701 is configured to send at least one of the following when a pre-configured trigger condition is met: first processing resource information, target algorithm recommendation information, second interaction information, and second processing resource information. The second interaction information is used to indicate the sharing capability of multiple target algorithms to the NPU, and the second processing resource information is used to indicate the resource information of the first communication device for running the target algorithm.

[0480] Optionally, the triggering condition includes at least one of the following:

[0481] The available resources of the first communication device are less than the preset resource amount;

[0482] The available resources of the first communication device cannot support the operation of the indicated target algorithm.

[0483] The first communication device cannot process the information in the manner specified in the first processing information or the second interactive information.

[0484] The available resources of the first communication device have changed, or the range of change of available resources exceeds a certain threshold;

[0485] The first communication device needs to run multiple target algorithms simultaneously.

[0486] Optionally, the first processing resource information or the second interaction information further includes at least one of the following:

[0487] Can at least two objective algorithms run simultaneously?

[0488] A combination of target algorithms that can run simultaneously;

[0489] Combinations of target algorithms that cannot run simultaneously;

[0490] The running time of each target algorithm in the combination of target algorithms running simultaneously;

[0491] The first communication device runs at least two target algorithms in parallel, corresponding to their running times.

[0492] The priority of the target algorithm.

[0493] Optionally, the radio frequency unit 701 is configured to obtain update indication information from at least one target communication device, the update indication information being used to indicate that the target algorithm information is updated; the processor 710 is configured to update the target algorithm information according to the update indication information.

[0494] Optionally, the target parameter includes at least one of the following:

[0495] Activation function;

[0496] Number of floors;

[0497] Number of neurons;

[0498] Number of neurons per layer;

[0499] The connections between neurons;

[0500] The running time of the target algorithm.

[0501] In this embodiment of the application, target algorithm information is obtained. The target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm. The first communication device can determine the target algorithm used by the first communication device by interacting with at least one communication device based on the target algorithm information, such as the target neural network model used by the first communication device. This achieves the purpose of making the neural network model used by the first communication device a neural network model that is compatible with its resource capabilities.

[0502] like Figure 8 As described above, this application embodiment also provides an information interaction device 800, including:

[0503] The second transceiver module 801 is used to send target algorithm information, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm;

[0504] The second interaction module 802 is used to interact with at least one communication device based on the target algorithm information.

[0505] Optionally, the apparatus in this application embodiment further includes:

[0506] The third acquisition module is used to acquire first interaction information sent by the target communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the target algorithm.

[0507] Optionally, the apparatus in this application embodiment further includes:

[0508] The fourth transceiver module is used to send the first instruction information;

[0509] The first indication information includes at least one of the following:

[0510] Priority information for the target algorithm;

[0511] The time from configuration to activation of a neural network model;

[0512] The time required to switch between at least two neural network models;

[0513] Whether the corresponding operations can be performed during the switching between at least two neural network models.

[0514] Optionally, the apparatus in this application embodiment further includes:

[0515] The fifth transceiver module is used to send second indication information, which is used to indicate the target algorithm used by the first communication device.

[0516] Optionally, the apparatus in this application embodiment further includes:

[0517] The sixth transceiver module is used to send update indication information, which is used to indicate that the configuration information should be updated.

[0518] Optionally, the apparatus in this application embodiment further includes:

[0519] The fourth acquisition module is used to acquire at least one of the second interaction information and the second processing resource information from the target communication device. The second interaction information is used to indicate the sharing capability of multiple target algorithms to the NPU, and the second processing resource information is used to indicate the resource information of the first communication device for running the target algorithm.

[0520] Optionally, the target algorithm information includes at least one of the following:

[0521] The target algorithm requires the type of activation function;

[0522] The target algorithm requires a specific type of operation.

[0523] The conversion relationship of available resources used between different operations required by the target algorithm;

[0524] The target algorithm requires that it support different specific operations or super-operations within the same time unit.

[0525] The target algorithm requires a combination relationship between different specific operations;

[0526] The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit.

[0527] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0528] The target algorithm requires the time needed to complete a specific operation to achieve a specific objective.

[0529] The target algorithm requires supercomputing information;

[0530] The target algorithm requires information about the neural network model.

[0531] Model size;

[0532] The number of quantization bits in the target algorithm model;

[0533] Number of model layers;

[0534] The maximum number of neurons in each layer of the model;

[0535] The target algorithm requires the runtime of each neural network algorithm.

[0536] The target algorithm requires the runtime of a combination of multiple neural network algorithms;

[0537] The target algorithm requires a specific time.

[0538] The target algorithm requires the target parameters of the target algorithm;

[0539] At least one expected time to run the target algorithm;

[0540] The input type of the target algorithm;

[0541] The preprocessing information corresponding to the input information of the target algorithm;

[0542] Different target parameters corresponding to neural network models of the same target algorithm type;

[0543] Neural network model library.

[0544] Optionally, the first processing resource information includes at least one of the following:

[0545] The activation function types supported by the first communication device;

[0546] The types of operations supported by the first communication device;

[0547] The conversion relationship of available resources used between different operations supported by the first communication device;

[0548] Whether the first communication device supports different specific operations or super operations within the same time unit;

[0549] The combination relationship between different specific operations supported by the first communication device;

[0550] The number of operations supported by the first communication device within a first target time period, wherein the first target time period includes at least one time unit;

[0551] The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time;

[0552] The time required for the first communication device to complete a specific operation for a specific target;

[0553] The supercomputing information supported by the first communication device;

[0554] Information on the neural network model supported by the first communication device;

[0555] Model size;

[0556] The number of quantization bits in the target algorithm model;

[0557] Number of model layers;

[0558] The maximum number of neurons in each layer of the model;

[0559] The running time of each neural network algorithm;

[0560] Running time of a combination of multiple neural network algorithms;

[0561] The target algorithm that the first communication device supports or does not support;

[0562] Whether the first communication device supports the target algorithm according to a predefined time;

[0563] The time required for the first communication device to run the target algorithm;

[0564] The expected time for the first communication device to run the target algorithm;

[0565] The target parameters of the target algorithm supported by the first communication device.

[0566] Optionally, the target algorithm recommendation information includes at least one of the following:

[0567] The first communication device provides recommendation information for the target parameters of the target algorithm;

[0568] The first communication device provides revision information for the target parameters of the target algorithm;

[0569] Information on whether the first communication device supports the target algorithm.

[0570] Optionally, the first processing resource information or the second interaction information further includes at least one of the following:

[0571] Can at least two objective algorithms run simultaneously?

[0572] A combination of target algorithms that can run simultaneously;

[0573] Combinations of target algorithms that cannot run simultaneously;

[0574] The running time of each target algorithm in the combination of target algorithms running simultaneously;

[0575] The first communication device runs at least two target algorithms in parallel, corresponding to their running times.

[0576] The priority of the target algorithm.

[0577] Optionally, the target parameter includes at least one of the following:

[0578] Activation function;

[0579] Number of floors;

[0580] Number of neurons;

[0581] Number of neurons per layer;

[0582] The connections between neurons;

[0583] The running time of the target algorithm.

[0584] The apparatus of this application embodiment sends target algorithm information to a first communication device. The target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm. Subsequently, based on the target algorithm information, the communication devices can interact with each other to determine the target algorithm used by the first communication device, such as the target neural network model used by the first communication device. This achieves the purpose of making the neural network model used by the first communication device a neural network model that is adapted to its resource capabilities.

[0585] like Figure 9 As described above, this application embodiment also provides an information interaction device 900, including:

[0586] The first acquisition module 901 is used to acquire first interaction information sent by the target communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the target algorithm. The target communication device includes at least one of a second communication device and a first communication device.

[0587] The third interaction module 902 is used to interact with the target communication device based on the first interaction information.

[0588] Optionally, the information exchanged with the target communication device includes at least one of the following:

[0589] Priority information for the target algorithm;

[0590] The time from configuration to activation of a neural network model;

[0591] The time required to switch between at least two neural network models;

[0592] Can the corresponding computations be performed during the switching between at least two neural network models?

[0593] Update instruction information, which is used to indicate that the configuration information of the target algorithm should be updated;

[0594] The target algorithm used by the first communication device.

[0595] The information interaction device in this application embodiment, after obtaining the first interaction information, feeds back to the first communication device or other communication devices the target algorithm used by the first communication device, the running priority indication information of the target algorithm, the time from configuration to activation of the neural network model, the time required to switch between at least two neural network models, etc., so that the first communication device can use the target algorithm adapted to its running resources.

[0596] This application also provides a second communication device, specifically a network-side device, including a processor and a communication interface. The communication interface is used to send target algorithm information to a first communication device, wherein the target algorithm includes at least one of a neural network algorithm, an artificial intelligence algorithm, and a machine learning algorithm. The second communication device interacts with a target communication device based on the target algorithm information, wherein the target communication device includes at least one of a first communication device and a third communication device. This network-side device embodiment corresponds to the above-described second communication device-side method embodiment. All implementation processes and methods of the above-described second communication device-side method embodiment can be applied to this network-side device embodiment and achieve the same technical effects.

[0597] This application embodiment also provides a third communication device. The second communication device is specifically a network-side device, including a processor and a communication interface. The communication interface is used to acquire first interactive information sent by a target communication device. The first interactive information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device on resource information for running the target algorithm. The target communication device includes at least one of a second communication device and a first communication device. The application interacts with the target communication device based on the first interactive information. This network-side device embodiment corresponds to the above-described third communication device-side method embodiment. All implementation processes and methods of the above-described third communication device-side method embodiment can be applied to this network-side device embodiment and achieve the same technical effects.

[0598] Specifically, embodiments of this application also provide a network-side device, which is specifically a second communication device or a third communication device. For example... Figure 10 As shown, the network-side device 1000 includes: an antenna 101, a radio frequency (RF) device 102, a baseband device 103, a processor 104, and a memory 105. The antenna 101 is connected to the RF device 102. In the uplink direction, the RF device 102 receives information through the antenna 101 and transmits the received information to the baseband device 103 for processing. In the downlink direction, the baseband device 103 processes the information to be transmitted and sends it to the RF device 102. The RF device 102 processes the received information and transmits it through the antenna 101.

[0599] The method executed by the second or third communication device in the above embodiments can be implemented in the baseband device 103, which includes a baseband processor.

[0600] The baseband device 103 may include, for example, at least one baseband board on which multiple chips are disposed, such as... Figure 10 As shown, one of the chips is, for example, a baseband processor, which is connected to the memory 105 via a bus interface to call the program in the memory 105 and execute the operation of the second or third communication device shown in the above method embodiment.

[0601] The network-side device may also include a network interface 106, such as a common public radio interface (CPRI).

[0602] Specifically, the network-side device 1000 of this embodiment further includes: instructions or programs stored in memory 105 and executable on processor 104, wherein processor 104 calls the instructions or programs in memory 105 to execute... Figure 8 or Figure 9The methods executed by each module shown achieve the same technical effect, and to avoid repetition, they will not be described in detail here.

[0603] Specifically, embodiments of this application also provide a network-side device, which is specifically a second communication device or a third communication device. For example... Figure 11 As shown, the network-side device 1100 includes a processor 1101, a network interface 1102, and a memory 1103. The network interface 1102 is, for example, a common public radio interface (CPRI).

[0604] Specifically, the network-side device 1100 of this embodiment further includes: instructions or programs stored in memory 1103 and executable on processor 1101, wherein processor 1101 calls the instructions or programs in memory 1103 to execute. Figure 8 or Figure 9 The methods executed by each module shown achieve the same technical effect, and to avoid repetition, they will not be described in detail here.

[0605] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described information interaction method embodiments and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0606] The processor is the processor in the terminal described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0607] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described information interaction method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0608] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0609] This application also provides a computer program / program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the above-described information interaction method embodiments, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0610] This application also provides an information interaction system, including: a terminal and a network-side device. The terminal can be used to execute the steps of the information interaction method executed by the first communication device as described above, and the network-side device can be used to execute the steps of the information interaction method executed by the second communication device as described above.

[0611] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0612] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0613] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. An information exchange method, characterized in that, include: The first communication device acquires target algorithm information, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm; The first communication device interacts with at least one target communication device based on the target algorithm information; The target algorithm information includes at least one of the following: The target algorithm requires the type of activation function; The target algorithm requires a specific type of operation. The conversion relationship of available resources used between different operations required by the target algorithm; The target algorithm requires that it support different specific operations or super-operations within the same time unit. The target algorithm requires a combination relationship between different specific operations; The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit. The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The target algorithm requires the time needed to complete a specific operation to achieve a specific objective. The target algorithm requires supercomputing information; The target algorithm requires information about the neural network model. The number of quantization bits in the target algorithm model; The target algorithm requires the runtime of each neural network algorithm. The target algorithm requires the runtime of a combination of multiple neural network algorithms; The target algorithm requires a specific time. The target algorithm requires the target parameters of the target algorithm; At least one expected time to run the target algorithm; Wherein, the first communication device interacts with at least one target communication device, including: The first communication device sends first interaction information to at least one communication device based on the target algorithm information. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device on the resource information for running the target algorithm. The first processing resource information includes at least one of the following: supercomputing information supported by the first communication device, the time required for the first communication device to run the target algorithm, and the expected time for the first communication device to run the target algorithm; the first processing resource information also includes at least one of the following: a combination of target algorithms that cannot run simultaneously, and the running time corresponding to the first communication device running at least two target algorithms in parallel. The target algorithm recommendation information includes the revision information of the target parameters of the target algorithm by the first communication device.

2. The method according to claim 1, characterized in that, The target algorithm information is obtained through a pre-agreed method, or the target algorithm information is configured by at least one of the communication devices.

3. The method according to claim 1, characterized in that, The first communication device interacts with at least one target communication device, including: The first communication device obtains at least one of a first indication information and a second indication information from at least one target communication device, wherein the second indication information is used to indicate the target algorithm used by the first communication device; The first indication information includes at least one of the following: Priority information for the target algorithm; The time from configuration to activation of a neural network model; The time required to switch between at least two neural network models; Whether the corresponding operations can be performed during the switching between at least two neural network models.

4. The method according to claim 1 or 2, characterized in that, The first communication device acquires target algorithm information, including: The target algorithm information is obtained based on at least one of the first indication information and the second indication information, wherein the second indication information is used to indicate the target algorithm used by the first communication device; The first indication information includes at least one of the following: Priority information for the target algorithm; The time from configuration to activation of a neural network model; The time required to switch between at least two neural network models; Whether the corresponding operations can be performed during the switching between at least two neural network models.

5. The method according to claim 1, characterized in that, The first interactive information also includes at least one of the following: Priority information for the target algorithm; The time from configuration to activation of a neural network model; The time required to switch between at least two neural network models; Whether the corresponding operations can be performed during the switching between at least two neural network models.

6. The method according to claim 1, characterized in that, The first processing resource information also includes at least one of the following: The activation function types supported by the first communication device; The types of operations supported by the first communication device; The conversion relationship of available resources used between different operations supported by the first communication device; Whether the first communication device supports different specific operations or super operations within the same time unit; The combination relationship between different specific operations supported by the first communication device; The number of operations supported by the first communication device within a first target time period, wherein the first target time period includes at least one time unit; The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The time required for the first communication device to complete a specific operation for a specific target; Information on the neural network model supported by the first communication device; Model size; The number of quantization bits in the target algorithm model; Number of model layers; The maximum number of neurons in each layer of the model; The running time of each neural network algorithm; Running time of a combination of multiple neural network algorithms; The target algorithm that the first communication device supports or does not support; Whether the first communication device supports the target algorithm according to a predefined time; The target parameters of the target algorithm supported by the first communication device.

7. The method according to claim 1, characterized in that, The target algorithm information also includes at least one of the following: Model size; Number of model layers; The maximum number of neurons in each layer of the model; The input type of the target algorithm; The preprocessing information corresponding to the input information of the target algorithm; Different target parameters corresponding to neural network models of the same target algorithm type; Neural network model library.

8. The method according to claim 6 or 7, characterized in that, The number of operations supported by the first communication device within the first target time period includes at least one of the following: The number of operations performed for each type of operation during the first target time period; The number of operations performed when at least two operations are executed simultaneously within the first target time period; The total number of calculations within the first target time.

9. The method according to claim 6 or 7, characterized in that, The unit of time unit is a time unit, subcarrier interval, time slot, or frame.

10. The method according to claim 1, characterized in that, The target algorithm recommendation information also includes at least one of the following: The first communication device provides revision information for the target parameters of the target algorithm; Information on whether the first communication device supports the target algorithm.

11. The method according to claim 1, characterized in that, The first communication device interacts with at least one target communication device, including: Based on the target algorithm information, the target algorithm used by the first communication device is determined; Send the target algorithm used by the first communication device to at least one target communication device.

12. The method according to claim 1, characterized in that, Also includes: When a pre-configured triggering condition is met, the first communication device sends at least one of the following: first processing resource information, target algorithm recommendation information, second interaction information, and second processing resource information. The second interaction information is used to indicate the shared capability of multiple target algorithms to the NPU, and the second processing resource information is used to indicate the resource information of the first communication device for running the target algorithm.

13. The method according to claim 12, characterized in that, The triggering condition includes at least one of the following: The available resources of the first communication device are less than the preset resource amount; The available resources of the first communication device cannot support the operation of the indicated target algorithm; The first communication device cannot process the information in the manner specified in the first processing information or the second interactive information. The available resources of the first communication device have changed, or the range of change of available resources exceeds a certain threshold; The first communication device needs to run multiple target algorithms simultaneously.

14. The method according to claim 12, characterized in that, The second interactive information also includes at least one of the following: Can at least two objective algorithms run simultaneously? A combination of target algorithms that can run simultaneously; Combinations of target algorithms that cannot run simultaneously; The running time of each target algorithm in the combination of target algorithms running simultaneously; The first communication device runs at least two target algorithms in parallel, corresponding to their running times. The priority of the target algorithm.

15. The method according to claim 1, characterized in that, Also includes: The first communication device obtains update indication information from at least one target communication device, the update indication information being used to indicate an update to the target algorithm information; The target algorithm information is updated according to the update instruction information.

16. The method according to claim 6 or 10, characterized in that, The target parameter includes at least one of the following: Activation function; Number of floors; Number of neurons; Number of neurons per layer; The connections between neurons; The running time of the target algorithm.

17. An information exchange method, characterized in that, include: The second communication device sends target algorithm information to the first communication device, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm; The second communication device interacts with the target communication device based on the target algorithm information, wherein the target communication device includes at least one of the first communication device and the third communication device; The target algorithm information includes at least one of the following: The target algorithm requires the type of activation function; The target algorithm requires a specific type of operation. The conversion relationship of available resources used between different operations required by the target algorithm; The target algorithm requires that it support different specific operations or super-operations within the same time unit. The target algorithm requires a combination relationship between different specific operations; The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit. The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The target algorithm requires the time needed to complete a specific operation to achieve a specific objective. The target algorithm requires supercomputing information; The target algorithm requires information about the neural network model. The number of quantization bits in the target algorithm model; The target algorithm requires the runtime of each neural network algorithm. The target algorithm requires the runtime of a combination of multiple neural network algorithms; The target algorithm requires a specific time. The target algorithm requires the target parameters of the target algorithm; At least one expected time to run the target algorithm; The method further includes: The second communication device acquires first interaction information sent by the target communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the resource information of the target algorithm. The first processing resource information includes at least one of the following: supercomputing information supported by the first communication device, the time required for the first communication device to run the target algorithm, and the expected time for the first communication device to run the target algorithm; the first processing resource information also includes at least one of the following: a combination of target algorithms that cannot run simultaneously, and the running time corresponding to the first communication device running at least two target algorithms in parallel. The target algorithm recommendation information includes the revision information of the target parameters of the target algorithm by the first communication device.

18. The method according to claim 17, characterized in that, Also includes: Send the first instruction message; The first indication information includes at least one of the following: Priority information for the target algorithm; The time from configuration to activation of a neural network model; The time required to switch between at least two neural network models; Whether the corresponding operations can be performed during the switching between at least two neural network models.

19. The method according to claim 17, characterized in that, Also includes: Send a second indication message, which is used to indicate the target algorithm used by the first communication device.

20. The method according to claim 17, characterized in that, Also includes: Send update indication information, which is used to indicate that the configuration information of the target algorithm should be updated.

21. The method according to claim 17, characterized in that, Also includes: At least one of second interaction information and second processing resource information is obtained from the target communication device, wherein the second interaction information is used to indicate the shared capability of multiple target algorithms to the NPU, and the second processing resource information is used to indicate the resource information of the first communication device for running the target algorithm.

22. The method according to claim 17, characterized in that, The target algorithm information also includes at least one of the following: Model size; Number of model layers; The maximum number of neurons in each layer of the model; The input type of the target algorithm; The preprocessing information corresponding to the input information of the target algorithm; Different target parameters corresponding to neural network models of the same target algorithm type; Neural network model library.

23. The method according to claim 17, characterized in that, The first processing resource information also includes at least one of the following: The activation function types supported by the first communication device; The types of operations supported by the first communication device; The conversion relationship of available resources used between different operations supported by the first communication device; Whether the first communication device supports different specific operations or super operations within the same time unit; The combination relationship between different specific operations supported by the first communication device; The number of operations supported by the first communication device within a first target time period, wherein the first target time period includes at least one time unit; The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The time required for the first communication device to complete a specific operation for a specific target; Information on the neural network model supported by the first communication device; Model size; The number of quantization bits in the target algorithm model; Number of model layers; The maximum number of neurons in each layer of the model; The running time of each neural network algorithm; Running time of a combination of multiple neural network algorithms; The target algorithm that the first communication device supports or does not support; Whether the first communication device supports the target algorithm according to a predefined time; The target parameters of the target algorithm supported by the first communication device.

24. The method according to claim 17, characterized in that, The target algorithm recommendation information also includes at least one of the following: The first communication device provides recommendation information for the target parameters of the target algorithm; Information on whether the first communication device supports the target algorithm.

25. The method according to claim 21, characterized in that, The second interactive information also includes at least one of the following: Can at least two objective algorithms run simultaneously? A combination of target algorithms that can run simultaneously; Combinations of target algorithms that cannot run simultaneously; The running time of each target algorithm in the combination of target algorithms running simultaneously; The first communication device runs at least two target algorithms in parallel, corresponding to their running times. The priority of the target algorithm.

26. The method according to any one of claims 22 to 24, characterized in that, The target parameter includes at least one of the following: Activation function; Number of floors; Number of neurons; Number of neurons per layer; The connections between neurons; The running time of the target algorithm.

27. An information exchange method, characterized in that, include: A third communication device acquires first interactive information sent by a target communication device. The first interactive information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the resource information of the target algorithm. The target communication device includes at least one of a second communication device and a first communication device. The first interactive information is sent based on the target algorithm information. The third communication device interacts with the target communication device based on the first interaction information; The target algorithm information includes at least one of the following: The target algorithm requires the type of activation function; The target algorithm requires a specific type of operation. The conversion relationship of available resources used between different operations required by the target algorithm; The target algorithm requires that it support different specific operations or super-operations within the same time unit. The target algorithm requires a combination relationship between different specific operations; The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit. The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The target algorithm requires the time needed to complete a specific operation to achieve a specific objective. The target algorithm requires supercomputing information; The target algorithm requires information about the neural network model. The number of quantization bits in the target algorithm model; The target algorithm requires the runtime of each neural network algorithm. The target algorithm requires the runtime of a combination of multiple neural network algorithms; The target algorithm requires a specific time. The target algorithm requires the target parameters of the target algorithm; At least one expected time to run the target algorithm; The first processing resource information includes at least one of the following: supercomputing information supported by the first communication device, the time required for the first communication device to run the target algorithm, and the expected time for the first communication device to run the target algorithm; the first processing resource information also includes at least one of the following: a combination of target algorithms that cannot run simultaneously, and the running time corresponding to the first communication device running at least two target algorithms in parallel. The target algorithm recommendation information includes the revision information of the target parameters of the target algorithm by the first communication device.

28. The method according to claim 27, characterized in that, The information exchanged between the third communication device and the target communication device includes at least one of the following: Priority information for the target algorithm; The time from configuration to activation of a neural network model; The time required to switch between at least two neural network models; Can the corresponding computations be performed during the switching between at least two neural network models? Update instruction information, which is used to indicate that the configuration information of the target algorithm should be updated; The target algorithm used by the first communication device.

29. An information interaction device, characterized in that, include: The first transceiver module is used to acquire target algorithm information, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm; The first interaction module is used to interact with at least one target communication device based on the target algorithm information; The target algorithm information includes at least one of the following: The target algorithm requires the type of activation function; The target algorithm requires a specific type of operation. The conversion relationship of available resources used between different operations required by the target algorithm; The target algorithm requires that it support different specific operations or super-operations within the same time unit. The target algorithm requires a combination relationship between different specific operations; The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit. The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The target algorithm requires the time needed to complete a specific operation to achieve a specific objective. The target algorithm requires supercomputing information; The target algorithm requires information about the neural network model. The number of quantization bits in the target algorithm model; The target algorithm requires the runtime of each neural network algorithm. The target algorithm requires the runtime of a combination of multiple neural network algorithms; The target algorithm requires a specific time. The target algorithm requires the target parameters of the target algorithm; At least one expected time to run the target algorithm; The first interaction module is used for: Based on the target algorithm information, first interaction information is sent to at least one communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device on the resource information for running the target algorithm. The first processing resource information includes at least one of the following: supercomputing information supported by the first communication device, the time required for the first communication device to run the target algorithm, and the expected time for the first communication device to run the target algorithm; the first processing resource information also includes at least one of the following: a combination of target algorithms that cannot run simultaneously, and the running time corresponding to the first communication device running at least two target algorithms in parallel. The target algorithm recommendation information includes the revision information of the target parameters of the target algorithm by the first communication device.

30. An information interaction device, characterized in that, include: The second transceiver module is used to send target algorithm information to the first communication device, wherein the target algorithm includes at least one of neural network algorithm, artificial intelligence algorithm and machine learning algorithm; The second interaction module is used to interact with the target communication device based on the target algorithm information, wherein the target communication device includes at least one of the first communication device and the third communication device. The target algorithm information includes at least one of the following: The target algorithm requires the type of activation function; The target algorithm requires a specific type of operation. The conversion relationship of available resources used between different operations required by the target algorithm; The target algorithm requires that it support different specific operations or super-operations within the same time unit. The target algorithm requires a combination relationship between different specific operations; The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit. The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The target algorithm requires the time needed to complete a specific operation to achieve a specific objective. The target algorithm requires supercomputing information; The target algorithm requires information about the neural network model. The number of quantization bits in the target algorithm model; The target algorithm requires the runtime of each neural network algorithm. The target algorithm requires the runtime of a combination of multiple neural network algorithms; The target algorithm requires a specific time. The target algorithm requires the target parameters of the target algorithm; At least one expected time to run the target algorithm; The device further includes: The third acquisition module is used to acquire first interaction information sent by the target communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the target algorithm. The first processing resource information includes at least one of the following: supercomputing information supported by the first communication device, the time required for the first communication device to run the target algorithm, and the expected time for the first communication device to run the target algorithm; the first processing resource information also includes at least one of the following: a combination of target algorithms that cannot run simultaneously, and the running time corresponding to the first communication device running at least two target algorithms in parallel. The target algorithm recommendation information includes the revision information of the target parameters of the target algorithm by the first communication device.

31. An information interaction device, characterized in that, include: A first acquisition module is configured to acquire first interaction information sent by a target communication device. The first interaction information includes at least one of first processing resource information and target algorithm recommendation information. The first processing resource information is used to instruct the first communication device to run the target algorithm. The target communication device includes at least one of a second communication device and a first communication device. The first interaction information is sent based on the target algorithm information. A third interaction module is configured to interact with the target communication device based on the first interaction information. The target algorithm information includes at least one of the following: The target algorithm requires the type of activation function; The target algorithm requires a specific type of operation. The conversion relationship of available resources used between different operations required by the target algorithm; The target algorithm requires that it support different specific operations or super-operations within the same time unit. The target algorithm requires a combination relationship between different specific operations; The target algorithm requires the number of operations to be supported within a first target time period, where the first target time period includes at least one time unit. The relationship between the number of operations supported in the first target time and the number of operations supported in the second target time, wherein the second target time includes at least one time unit, and the time unit included in the second target time is different from the time unit included in the first target time; The target algorithm requires the time needed to complete a specific operation to achieve a specific objective. The target algorithm requires supercomputing information; The target algorithm requires information about the neural network model. The number of quantization bits in the target algorithm model; The target algorithm requires the runtime of each neural network algorithm. The target algorithm requires the runtime of a combination of multiple neural network algorithms; The target algorithm requires a specific time. The target algorithm requires the target parameters of the target algorithm; At least one expected time to run the target algorithm; The first processing resource information includes at least one of the following: supercomputing information supported by the first communication device, the time required for the first communication device to run the target algorithm, and the expected time for the first communication device to run the target algorithm; the first processing resource information also includes at least one of the following: a combination of target algorithms that cannot run simultaneously, and the running time corresponding to the first communication device running at least two target algorithms in parallel. The target algorithm recommendation information includes the revision information of the target parameters of the target algorithm by the first communication device.

32. A communication device, characterized in that, The method includes a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions being executed by the processor to implement the steps of the information interaction method as described in any one of claims 1 to 16, or to implement the steps of the information interaction method as described in any one of claims 17 to 26, or to implement the steps of the information interaction method as described in any one of claims 27 to 28.

33. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the information interaction method as described in any one of claims 1 to 16, or implement the steps of the information interaction method as described in any one of claims 17 to 26, or implement the steps of the information interaction method as described in any one of claims 27 to 28.

Images