WO2024099175A1 - 一种算法管理方法和装置 - Google Patents
一种算法管理方法和装置 Download PDFInfo
- Publication number
- WO2024099175A1 WO2024099175A1 PCT/CN2023/128368 CN2023128368W WO2024099175A1 WO 2024099175 A1 WO2024099175 A1 WO 2024099175A1 CN 2023128368 W CN2023128368 W CN 2023128368W WO 2024099175 A1 WO2024099175 A1 WO 2024099175A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- information
- algorithm
- energy consumption
- monitoring information
- monitoring
- Prior art date
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 358
- 238000007726 management method Methods 0.000 title claims abstract description 68
- 238000005265 energy consumption Methods 0.000 claims abstract description 207
- 238000012544 monitoring process Methods 0.000 claims description 143
- 238000000034 method Methods 0.000 claims description 84
- 238000004891 communication Methods 0.000 claims description 56
- 238000010801 machine learning Methods 0.000 claims description 43
- 238000013473 artificial intelligence Methods 0.000 claims description 42
- 230000015654 memory Effects 0.000 claims description 18
- 230000008859 change Effects 0.000 claims description 9
- 230000004044 response Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 description 36
- 230000006835 compression Effects 0.000 description 20
- 238000007906 compression Methods 0.000 description 20
- 238000012545 processing Methods 0.000 description 17
- 238000004364 calculation method Methods 0.000 description 16
- 230000005540 biological transmission Effects 0.000 description 15
- 238000005516 engineering process Methods 0.000 description 11
- 101100233916 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) KAR5 gene Proteins 0.000 description 6
- 230000000737 periodic effect Effects 0.000 description 6
- 238000013507 mapping Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000000691 measurement method Methods 0.000 description 4
- 238000010295 mobile communication Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 101000827703 Homo sapiens Polyphosphoinositide phosphatase Proteins 0.000 description 3
- 102100023591 Polyphosphoinositide phosphatase Human genes 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 230000001960 triggered effect Effects 0.000 description 3
- 101001121408 Homo sapiens L-amino-acid oxidase Proteins 0.000 description 2
- 102100026388 L-amino-acid oxidase Human genes 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000011144 upstream manufacturing Methods 0.000 description 2
- 101000741965 Homo sapiens Inactive tyrosine-protein kinase PRAG1 Proteins 0.000 description 1
- 102100038659 Inactive tyrosine-protein kinase PRAG1 Human genes 0.000 description 1
- 101100012902 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) FIG2 gene Proteins 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000010267 cellular communication Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000037361 pathway Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/02—Power saving arrangements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/04—TPC
- H04W52/06—TPC algorithms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/04—TPC
- H04W52/18—TPC being performed according to specific parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/04—TPC
- H04W52/18—TPC being performed according to specific parameters
- H04W52/24—TPC being performed according to specific parameters using SIR [Signal to Interference Ratio] or other wireless path parameters
Definitions
- the embodiments of the present invention relate to the field of communication technology, and in particular to an algorithm management method and device.
- AI/ML performance monitoring is an important research direction, that is, monitoring the performance of AI/ML algorithms.
- AI/ML performance monitoring results may be used for algorithm management, such as lifecycle management (LCM).
- the present application provides an algorithm management method and device for performing algorithm management based on energy consumption.
- an algorithm management method applied to a first device comprising: obtaining energy consumption comparison information of a first algorithm and a second algorithm; and sending the first information to a second device based on the energy consumption comparison information.
- the first information is used by the second device to perform algorithm management on the first device.
- the above-mentioned first information may include energy consumption comparison information, and may also include one or more of the following: a use case indication; an algorithm indication; or a communication system configuration indication.
- the first information may be used by the second device to perform lifecycle management (LCM) of the algorithm on the first device.
- LCM lifecycle management
- the first device mentioned above may be a terminal device.
- the first algorithm and the second algorithm can be algorithms of the same type or different types.
- the first algorithm is an artificial intelligence/machine learning (AI/ML) algorithm
- the second algorithm is a traditional algorithm
- the first algorithm and the second algorithm are different artificial intelligence/machine learning (AI/ML) algorithms.
- the method for obtaining energy consumption comparison information specifically includes: obtaining first energy consumption information of the first algorithm and second energy consumption information of the second algorithm; and obtaining the energy consumption comparison information based on the first energy consumption information and the second energy consumption information.
- the energy consumption comparison information may be a quotient, a difference or a size comparison result of the algorithm energy consumption of different algorithms, or a result of using other calculation methods that can represent the energy consumption comparison of different algorithms.
- the first device may send the processed energy consumption comparison information as part of the first information to the second device, specifically: obtaining third energy consumption information according to the energy consumption comparison information and the power information and/or energy consumption configuration information of the first device; sending the first information including the third energy consumption information to the second device.
- the third energy consumption information may include various information such as energy consumption comparison information, power information of the first device, energy consumption configuration information of the first device, and may also include one or more energy consumption indicators calculated from the above information.
- triggering conditions for sending the first information to the second device may be different.
- the first device in response to determining that a preset sending condition is met, may send the first information to the second device.
- the preset sending condition may include periodic sending.
- Another possible implementation manner may be to send the first signal in response to receiving the first instruction sent by the second device. interest.
- the first instruction may include: a first instruction for indicating configuration of the first device; or a first instruction for indicating that the first device changes configuration; or a first instruction for querying the first information.
- the first instruction for indicating configuration of the first device may be a configuration message including configuration information.
- the first instruction for indicating that the first device changes configuration may be a configuration change message.
- the first instruction for querying the first information may be a query message.
- the first information may be sent in response to the first device connecting to the second device.
- the algorithm management method sends the first information to the second device according to the energy consumption comparison information, so that the second device can manage the algorithm of the first device by integrating the first information.
- an algorithm management method applied to a first device comprising: receiving second information sent by a second device; and managing the algorithm of the first device according to the second information.
- the second information includes monitoring information of the algorithm of the first device acquired by the second device.
- the monitoring information of the above algorithm may be performance monitoring information of the algorithm.
- the second information may include one or more of the following: a use case indication; an algorithm indication; or a communication system configuration indication.
- the second information may be used by the first device to perform lifecycle management (LCM) of the algorithm on the first device.
- LCM lifecycle management
- the first device mentioned above may be a terminal device.
- the first device may acquire the first energy consumption information of the first algorithm and the second energy consumption information of the second algorithm.
- the first device may manage an algorithm of the first device based on the first energy consumption information, the second energy consumption information, and the second information.
- the first algorithm and the second algorithm can be algorithms of the same type or different types.
- the first algorithm is an artificial intelligence/machine learning (AI/ML) algorithm
- the second algorithm is a traditional algorithm
- the first algorithm and the second algorithm are different artificial intelligence/machine learning (AI/ML) algorithms.
- the algorithm management method manages the algorithm of the first device by receiving the second information sent by the second device and integrating the first energy consumption information and the second energy consumption information.
- a method for managing an algorithm for a second device comprising: receiving first information sent by a first device; and managing the algorithm of the first device according to the first information.
- the first information includes energy consumption comparison information of the first algorithm and the second algorithm.
- the above-mentioned first information may include energy consumption comparison information, and may also include one or more of the following: a use case indication; an algorithm indication; or a communication system configuration indication.
- the second device can perform lifecycle management (LCM) of the algorithm on the above-mentioned first device.
- LCM lifecycle management
- the second device may be a network device.
- the network device may include an access network device and/or a core network device.
- the second device may obtain the first monitoring information of the first algorithm of the first device and/or the second monitoring information of the second algorithm of the first device.
- the first monitoring information and the second monitoring information are performance monitoring information.
- the second device may manage the algorithm of the first device according to the first information, the first monitoring information and/or the second monitoring information.
- the second device can perform lifecycle management (LCM) of the algorithm on the above-mentioned first device.
- LCM lifecycle management
- the first algorithm and the second algorithm can be algorithms of the same type or different types.
- the first algorithm is an artificial intelligence/machine learning (AI/ML) algorithm
- the second algorithm is a traditional algorithm
- the first algorithm and the second algorithm are different artificial intelligence/machine learning (AI/ML) algorithms.
- the algorithm management method manages the algorithm of the first device by receiving the first information sent by the first device and integrating the first information, the first monitoring information and the second monitoring information.
- an algorithm management method applied to a second device comprising: obtaining first monitoring information of a first algorithm of a first device, and/or second monitoring information of a second algorithm of the first device; and sending second information to the first device based on the first monitoring information and/or the second monitoring information.
- the second information is used by the first device to perform algorithm management on the first device.
- the second information may also include one or more of the following: a use case indication; an algorithm indication; or a communication system configuration indication.
- the second information may be used by the first device to perform lifecycle management (LCM) of the algorithm on the first device.
- LCM lifecycle management
- the first monitoring information and the second monitoring information may be performance monitoring information.
- the second device may be a network device.
- the network device may include an access network device and/or a core network device.
- the first algorithm and the second algorithm can be algorithms of the same type or different types.
- the first algorithm is an artificial intelligence/machine learning (AI/ML) algorithm
- the second algorithm is a traditional algorithm
- the first algorithm and the second algorithm are different artificial intelligence/machine learning (AI/ML) algorithms.
- the first monitoring information, and/or the second monitoring information are processed and sent to the first device as part of the second information, specifically including: obtaining third monitoring information based on the first monitoring information, and/or the second monitoring information; and sending the second information including the third monitoring information to the first device.
- the third monitoring information may be performance monitoring information.
- the algorithm management method enables the first device to comprehensively manage the algorithm of the first device by sending the second information.
- a device in a fifth aspect, is provided.
- the communication device may be the first device described in any one of the first and third aspects.
- the device has the functions of the first device.
- the device is, for example, a terminal device, including: a fixed device, a mobile device, a handheld device (such as a mobile phone), a wearable device, a vehicle-mounted device, or a wireless device built into the above device (for example, a communication module, a modem, or a chip system, etc.).
- the device includes a processing unit (sometimes also referred to as a processing module) and a transceiver unit (sometimes also referred to as a transceiver module).
- the transceiver unit can implement a sending function and a receiving function.
- a sending unit sometimes also referred to as a sending module
- a receiving unit sometimes also referred to as a receiving module
- the sending unit and the receiving unit can be the same functional module, which is called a transceiver unit, and the functional module can implement a sending function and a receiving function; or, the sending unit and the receiving unit can be different functional modules, and the transceiver unit is a general term for these functional modules.
- the transceiver unit (or, the receiving unit) is used to send first information (energy consumption information) to a second device, and the first information (energy consumption information) is used by the second device to perform algorithm management on the first device; the transceiver unit (or, the receiving unit) is used to receive instructions of the management algorithm from the second device.
- the device also includes a storage unit (sometimes also referred to as a storage module), and the processing unit is used to couple with the storage unit and execute the program or instructions in the storage unit, so that the device can perform the function of the first device described in any one of the first and third aspects above.
- a storage unit sometimes also referred to as a storage module
- the processing unit is used to couple with the storage unit and execute the program or instructions in the storage unit, so that the device can perform the function of the first device described in any one of the first and third aspects above.
- a device is provided.
- the communication device may be the second device described in any one of the second aspect and the fourth aspect.
- the device has the functions of the second device.
- the device is, for example, a network device.
- the device includes a processing unit (sometimes also referred to as a processing module) and a transceiver unit (sometimes also referred to as a transceiver module).
- the transceiver unit is capable of implementing a sending function and a receiving function. When the transceiver unit implements the sending function, it may be referred to as a sending unit (sometimes also referred to as a sending module).
- the transceiver unit When the transceiver unit implements the receiving function, it may be referred to as a receiving unit (sometimes also referred to as a receiving module).
- the sending unit and the receiving unit may be the same functional module, which is called a transceiver unit.
- the functional module is capable of implementing a sending function and a receiving function. Receiving function; or, the sending unit and the receiving unit may be different functional modules, and the sending and receiving unit is a general term for these functional modules.
- the transceiver unit (or, the receiving unit) is used to send second information (monitoring information) to the first device, and the second information (monitoring information) is used for the first device management algorithm; the transceiver unit (or, the receiving unit).
- the device also includes a storage unit (sometimes also referred to as a storage module), and the processing unit is used to couple with the storage unit and execute the program or instructions in the storage unit, so that the device can perform the function of the second device described in any one of the second and fourth aspects above.
- a storage unit sometimes also referred to as a storage module
- the processing unit is used to couple with the storage unit and execute the program or instructions in the storage unit, so that the device can perform the function of the second device described in any one of the second and fourth aspects above.
- the present invention provides a non-volatile computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by a first device including the first aspect or any possible implementation manner described above, the first device executes:
- the present invention provides a non-volatile computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by a first device including the second aspect or any possible implementation manner described above, the first device executes:
- Acquire first energy consumption information of a first algorithm and second energy consumption information of a second algorithm receive second information sent by the second device; manage the algorithm of the first device according to the first energy consumption information, the second energy consumption information, and the second information.
- the present invention provides a non-volatile computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by the first device described in the third aspect or any one of the above possible implementations, the second device executes:
- first monitoring information of a first algorithm of a first device and second monitoring information of a second algorithm of the first device receive first information sent by the first device; and manage the algorithm of the first device according to the first information, the first monitoring information, and the second monitoring information.
- the present invention provides a non-volatile computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by the first device described in the fourth aspect or any possible implementation manner, the second device executes:
- an algorithm management system comprising a first device and a second device.
- the first device can be the first device described in the first aspect or the second aspect or the fifth aspect or the seventh aspect or the eighth aspect or any of the above-mentioned implementable methods
- the second device can be the second device described in the third aspect or the fourth aspect or the sixth aspect or the ninth aspect or the tenth aspect or any of the above-mentioned implementable methods.
- FIG1 is a schematic flow chart of an algorithm management method provided by an embodiment of the present invention.
- FIG2 is a schematic flow chart of another algorithm management method provided by an embodiment of the present invention.
- FIG3 is a schematic diagram of a device provided by an embodiment of the present invention.
- FIG. 4 is a schematic diagram of another device provided by an embodiment of the present invention.
- the system architecture of the embodiment of the present application includes at least two devices, a first device and a second device, wherein the first device and the second device may have a communication function.
- the first device may be a terminal device
- the second device may be a network device.
- the terminal device is a device with wireless transceiver function, which can be a fixed device, a mobile device, a handheld device (such as a mobile phone), a wearable device, a vehicle-mounted device, or a wireless device built into the above device (such as a communication module, a modem, etc.).
- the terminal device is used to connect people, objects, machines, etc., and can be widely used in various scenarios, such as but not limited to the following scenarios: cellular communication, device-to-device communication (D2D), vehicle to everything (V2X), machine-to-machine/machine-type communications (M2M/MTC), Internet of Things (IoT), virtual reality (VR), augmented reality (AR), industrial control, self driving, remote medical, smart grid, smart furniture, smart office, smart wear, smart transportation, smart city, drone, robot and other scenarios.
- the terminal device may sometimes be referred to as user equipment (UE), terminal, access station, UE station, remote station, wireless communication equipment, or user device, etc.
- the above-mentioned network equipment may include access network equipment and/or core network equipment.
- the access network equipment is a device with wireless transceiver function, which is used to communicate with the terminal equipment.
- the access network equipment includes but is not limited to base stations (base transceiver station (BTS), Node B, eNodeB/eNB, or gNodeB/gNB), transmission reception points (TRP), base stations of subsequent evolution of the third generation partnership project (3GPP), access nodes in wireless fidelity (Wi-Fi) systems, wireless relay nodes, wireless backhaul nodes, etc.
- the base station may be: a macro base station, a micro base station, a pico base station, a small station, a relay station, etc.
- Multiple base stations may support networks with the same access technology, or networks with different access technologies.
- a base station may include one or more co-sited or non-co-sited transmission and reception points.
- the access network device may also be a wireless controller, a centralized unit (CU), and/or a distributed unit (DU) in a cloud radio access network (CRAN) scenario.
- the access network device may also be a server, etc.
- the access network device in the vehicle to everything (V2X) technology may be a road side unit (RSU).
- the base station may communicate with the terminal device, or communicate with the terminal device through a relay station.
- the terminal device may communicate with multiple base stations in different access technologies.
- the core network device is used to implement functions such as mobility management, data processing, session management, policy and billing.
- the core network equipment includes: access and mobility management function (AMF), session management function (SMF), policy control function (PCF) or user plane function (UPF), etc.
- AMF access and mobility management function
- SMF session management function
- PCF policy control function
- UPF user plane function
- the number of nouns means “singular noun or plural noun", that is, “one or more”.
- At least one means one or more
- plural means two or more.
- “And/or” describes the association relationship of associated objects, indicating that three relationships may exist.
- a and/or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural.
- the character "/” generally indicates that the previous and next associated objects are in an “or” relationship.
- A/B means: A or B.
- “At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items.
- At least one of a, b, or c means: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.
- the ordinal numbers such as "first” and “second” mentioned in the embodiments of the present application are used to distinguish multiple objects, and are not used to limit the size, content, order, timing, priority or importance of multiple objects.
- the first information and the second information can be the same information or different information, and this name does not indicate the difference in content, size, sender/receiver, priority or importance of the two information.
- the numbering of the steps in the various embodiments introduced in the present application is only to distinguish different steps, and is not used to limit the order of the steps. For example, S401 may occur before S402, or may occur after S402, or may occur at the same time as S402.
- the technical solution provided in the embodiments of the present application can be applied to the fourth generation mobile communication technology (the 4th generation, 4G) system, such as the long term evolution (long term evolution, LTE) system, or can be applied to the fifth generation mobile communication technology (the 5th generation, 5G) system, such as the new radio (new radio, NR) system, or can also be applied to the next generation mobile communication system or other similar communication systems, such as the sixth generation mobile communication technology (the 6th generation, 6G) system, etc., without specific limitation.
- the 4th generation, 4G such as the long term evolution (long term evolution, LTE) system
- the 5th generation, 5G such as the new radio (new radio, NR) system
- the next generation mobile communication system or other similar communication systems such as the sixth generation mobile communication technology (the 6th generation, 6G) system, etc., without specific limitation.
- AI/ML mode and traditional mode are expected to be used in wireless communication for a period of time.
- the traditional mode may include traditional algorithms, and the AI/ML mode may include one or more different AI/ML models.
- a model may also be referred to as an algorithm, and it is understandable that an AI/ML algorithm may be understood as an AI/ML model, or, with an AI/ML model as the core, supplemented by a traditional algorithm.
- traditional algorithms may include non-AI/ML algorithms, such as algorithms widely used in existing communication systems.
- the algorithms widely used in the existing communication systems may include algorithms based on mathematical models, and algorithms based on simple prediction, filtering, etc.
- the algorithms based on simple prediction, filtering, etc. may include algorithms based on Kalman filtering, sliding average, or autoregression, etc.
- CSI compression channel state information compression
- a traditional algorithm may include a codebook-based CSI compression algorithm.
- the energy consumption information may be power information, which is the energy consumption information per unit time.
- energy consumption information is used as an example in the embodiments of the present application.
- the second device in a wireless communication system can obtain monitoring information of the algorithm of the first device but does not have energy consumption information of the algorithm of the first device
- information transmission is required between the first device and the second device so that the first device or the second device can manage the algorithm of the first device by combining the monitoring information and energy consumption information.
- the energy consumption information may include one or more of the following: algorithm energy consumption, energy consumption comparison information.
- the energy consumption information may also include power information of the first device, and energy consumption configuration information of the first device.
- FIG1 exemplarily shows an algorithm management method implemented by a first device sending first information including energy consumption information to a second device, the method comprising:
- a first device obtains energy consumption comparison information.
- the energy consumption comparison information may include comparison information of the algorithm energy consumption of at least two algorithms executing the same task.
- the algorithm may include a traditional algorithm and an AI/ML algorithm. It should be noted that executing a task can be understood as executing K use cases, where K is a positive integer.
- the algorithm energy consumption may include the energy consumed by the algorithm to execute the task.
- use cases may include channel state information compression (CSI compression), positioning, beam management, etc.
- the calculation of a use case may be a CSI compression task (for example, outputting compressed CSI based on the input CSI before compression), a positioning calculation (for example, outputting position information or other positioning auxiliary information based on the input information required for positioning), or a beam management calculation (for example, outputting Kb beam information with the strongest expected signal based on the input of multiple beam information, where Kb is a positive integer).
- the first algorithm may be an AI/ML algorithm, that is, the AI/ML algorithm is used to perform the CSI compression task
- the second algorithm may be a traditional algorithm, that is, a codebook-based CSI compression task is performed
- the first algorithm and the second algorithm may be different AI/ML algorithms for performing the CSI compression task.
- Algorithm energy consumption can also include the energy consumption of auxiliary tasks of the task, the energy consumption of upstream and downstream tasks of the task, and other energy consumption caused by executing the task.
- the energy consumption of the above auxiliary tasks may include but is not limited to the energy consumption caused by one or more of the following tasks: preprocessing of CSI data, performance monitoring of one or more AI/ML algorithms, and performance monitoring of traditional algorithms.
- the energy consumption of upstream and downstream tasks of the task may include but is not limited to the energy consumption caused by one or more of the following tasks: acquisition of CSI before compression, and sending of CSI after compression.
- the comparison information may be a quotient, a difference or a size comparison result of the algorithm energy consumption of different algorithms, or a result of using other calculation methods that can represent the energy consumption comparison of different algorithms.
- the method for obtaining the algorithm energy consumption and/or energy consumption comparison information is not limited here, and includes a method for directly obtaining energy consumption comparison information and a method for indirectly obtaining energy consumption comparison information.
- the method for directly obtaining energy consumption comparison information may refer to a method for directly obtaining energy consumption comparison information of multiple algorithms, such as a configuration acquisition method, an estimation method, and an actual measurement method.
- the indirect method of obtaining energy consumption comparison information may refer to obtaining energy consumption comparison information by calculation after obtaining the algorithm energy consumption information of each algorithm respectively. Different methods of obtaining algorithm energy consumption information may be used for different algorithms and different use cases.
- the method of obtaining algorithm energy consumption information may include a mixture of one or more of the following methods: configuration acquisition method, estimation method, and actual measurement method.
- the above configuration acquisition method may include: the first device acquiring from factory configuration information of the first device, the first device acquiring from other devices, for example, acquiring from a second device, or acquiring from a server.
- the above estimation method can be calculated using an approximate formula.
- the background power ratio can be understood as the ratio of background power to peak power.
- the algorithm operation number N can be obtained by the first device through the server, the second device, or the algorithm is received and evaluated by itself.
- the server can be the server of the first device, the server of the second device or a third-party server.
- the processor energy efficiency E and the background power ratio U_P of the algorithm processor can be obtained through the first device, from the second device, or from other devices.
- the method for obtaining the computing resource utilization U_C includes obtaining through the second device, obtaining by executing the algorithm on the first device, or obtaining through other devices.
- the above-mentioned actual measurement method obtains the corresponding algorithm energy consumption or energy consumption comparison information by actually executing the algorithm.
- the above configuration acquisition method, estimation method, and actual measurement method are exemplary methods.
- the energy consumption comparison information method can be directly obtained or the energy consumption comparison information can be obtained indirectly.
- the energy consumption information may include the energy consumption comparison information.
- the energy consumption information may also include power information of the first device and energy consumption configuration information of the first device.
- the energy consumption configuration information may include performance priority, endurance priority, or balanced mode.
- the energy consumption information may include energy consumption comparison information, power information of the first device, energy consumption configuration information of the first device, and other information, and may also include one or more energy consumption indicators calculated from the above information.
- the embodiment of the present application does not limit the method of calculating one or more energy consumption indicators from various information in the energy consumption information.
- an energy consumption indicator is obtained by calculating energy consumption comparison information, power information, and energy consumption configuration information.
- This indicator is called the equivalent energy consumption comparison information of the two algorithms.
- the embodiment of the present application does not limit the method for obtaining equivalent energy consumption comparison information.
- a possible comprehensive idea is: the smaller the power or the more the energy consumption configuration information focuses on battery life, the comparison value of the energy consumption of the two algorithms is magnified, guiding the priority selection of low energy consumption mode.
- Two possible calculation methods are as follows:
- the power ratio value range can be (0%, 100%), when the energy consumption configuration information is performance priority, the configuration value is A1, when the energy consumption configuration information is balanced mode, the configuration value is A2, when the energy consumption configuration information is endurance priority, the configuration value is A3, where A1, A2 and A3 are positive numbers, and A1A2A3.
- the second possible calculation method is: According to the formula The energy consumption comparison value is obtained. When ⁇ >0, the energy consumption of the first algorithm is greater than that of the second algorithm. Conversely, when ⁇ 0, the energy consumption of the first algorithm is less than that of the second algorithm.
- the equivalent energy consumption comparison information may be the above calculation result, or may be the result of the above calculation result after post-processing such as saturation truncation, quantization, codebook mapping, etc.
- the first device sends first information (energy consumption information) to the second device, and the second device receives the first information (energy consumption information).
- the first information includes the energy consumption information and/or algorithm request information obtained according to the energy consumption information.
- the presentation method of the energy consumption information may include one or more of the following: a numerical value, a segmentation level of the energy consumption information.
- the numerical value may be a calculation result or a result of a calculation result after post-processing such as saturation truncation, quantization, and codebook mapping.
- the presentation method of the segmentation level of the energy consumption information may include gear information.
- the segmentation method of the gear information may be agreed upon in advance by the first device and the second device.
- the above-mentioned algorithm request information may include a request for replacing the algorithm obtained by the first device according to the energy consumption information.
- the algorithms may be sorted according to the size of the algorithm energy consumption to obtain the energy consumption ranking of the algorithms.
- a power threshold of the first device is set. When the power of the first device is greater than the power threshold, it is considered that the power is sufficient.
- the first device may send an algorithm request information to the second device according to the position of the current algorithm in the energy consumption ranking of the algorithm, and apply to switch to an algorithm with better performance.
- the current algorithm may be the algorithm currently being used by the first device in the use case.
- the first information may also include one or more of the following information: a use case indication, an algorithm indication, and a communication system configuration indication.
- the use case indication may include use case information
- the algorithm indication may include an algorithm identifier of the algorithm.
- the algorithm identifier may also be referred to as an algorithm ID.
- the algorithm indication may adopt either an explicit indication method such as an algorithm identifier or an implicit indication method, which is not limited in the embodiments of the present application.
- the communication system configuration may be a network configuration or a device configuration.
- the network configuration may include a frequency band
- the device configuration may include the number of antennas of the first device.
- the transmission mode of the transmission may include periodic transmission and non-periodic transmission.
- the periodic sending may be configured by the second device at a periodicity.
- the non-periodic transmission may include event-triggered transmission.
- the triggering conditions (events) for event triggering transmission may include connection between the first device and the second device, configuration modification of the first device, reception of a configuration change message sent by the second device, reception of a configuration message sent by the second device, power less than a threshold, power change or power change rate greater than a threshold, reception of a query message sent by the second device, etc.
- the first device sends or requests permission to send the first information.
- the first device can send the first information as the capability information of the first device to the second device in response to the first device being connected to the second device.
- the first device and the second device can agree in advance on the content and method of transmission.
- the transmission content may include the meaning represented by each field of the first information.
- the first device can send the first information to the second device in response to receiving the query message. That is, the query message can be used to instruct the first device to send the first information of the target algorithm and the current algorithm.
- the target algorithm may include the algorithm used by the first device in the target configuration of the second device.
- the first device can send a first message to the second device in response to the first device receiving the configuration message.
- the configuration message sent by the second device to the first device can serve as a basis for changing the configuration of the first device.
- the first device can send the first information to the first device in response to receiving the configuration change message.
- the configuration change message may include configuration information or may not include configuration information.
- the first device can determine the first information of the target algorithm and the current algorithm.
- the above configuration information may include one or more of the following: a sending mode, and configuration information of the sending mode.
- the configuration information of the sending mode may include the sending period;
- the configuration information of the sending mode may include the configuration of the triggering event, and/or the threshold in the triggering event.
- the configuration information may also include one or more of the following: use case indication, algorithm indication, and communication system configuration indication.
- use case indication is the same as that of S103.
- part or all of the content of the above configuration information may also be determined in advance by the first device and the second device.
- S104 The second device obtains monitoring information.
- the monitoring information may include monitoring information of one or more algorithms, or monitoring comparison information between algorithms. Taking the first algorithm and the second algorithm of the first device as an example, the monitoring information may include the first monitoring information of the first algorithm of the first device and the second monitoring information of the second algorithm, or the monitoring comparison information obtained by comparing the first monitoring information and the second monitoring information.
- the method of obtaining monitoring information of different algorithms may be different, and the embodiments of the present application do not limit the method of obtaining monitoring information.
- the first device is a terminal device
- the second device is a network device
- the monitoring information is performance monitoring information as an example.
- the terminal device sends compressed CSI to the network device.
- the terminal device can send uncompressed or high-fidelity compressed CSI to the network device, and the network device decompresses the compressed CSI, compares the decompressed output CSI with the uncompressed or high-fidelity compressed CSI, and obtains the algorithm performance monitoring information of the current algorithm compressed and decompressed on the CSI compression use case.
- the comparison method can include calculating the mean square error (MSE), and the algorithm performance can be expressed by the mean square error (MSE).
- the performance monitoring information may include performance monitoring information of one or more algorithms, or performance monitoring comparison information between algorithms, and may also include one or more performance monitoring indicators calculated from the above information.
- the above performance indicators may include one or more of the following: algorithm performance, communication system performance.
- the algorithm performance may include but is not limited to one or more of the following: accuracy, recall rate, error, etc.
- GCS generalized cosine similarity
- NMSE normalized mean square error
- the communication system performance may include but is not limited to one or more of the following: throughput, bit error rate (BER), block error rate (BLER), ACK (acknowledgment) ratio/NACK (not acknowledgment) ratio, retransmission rate, etc.
- the representation of the performance monitoring information may indicate a setting or a prior agreement.
- the representation of the performance monitoring information may include performance results, statistical results, and performance comparison results.
- the performance results and/or statistical results include but are not limited to mean, variance, percentile, histogram, and level.
- the performance comparison result may include the performance monitoring comparison results of two algorithms.
- the monitoring information may further include one or more of the following information: a use case indication, an algorithm indication, and a communication system configuration indication.
- the description of the use case indication, the algorithm indication, and the communication system configuration indication is the same as that of S103.
- the second device may obtain the use case indication from the first device; when the first information does not include the use case indication, the second device obtains the use case indication from itself.
- S104 may be executed before S101, or after S101, or simultaneously with S101.
- S104 may be executed before S102, or after S102, or simultaneously with S102.
- S104 may be executed before S103, or after S103, or simultaneously with S103.
- S104 may also not be executed, which is not limited in the embodiments of the present application.
- the second device manages an algorithm of the first device according to the first information (energy consumption information) and monitoring information.
- the first information (energy consumption information) and monitoring information may include one or more of the following information: a use case indication, an algorithm indication, and a communication system configuration indication.
- a use case indication an algorithm indication
- a communication system configuration indication The description of the use case indication, the algorithm indication, and the communication system configuration indication is the same as that of S103.
- the monitoring information obtained by the second device may be performance monitoring information, and the monitoring comparison information in the monitoring information may be performance monitoring comparison information.
- the second device can obtain energy consumption information based on the first information. Specifically, when the energy consumption information includes energy consumption comparison information, power information of the first device, energy consumption configuration information of the first device and other information, the second device can manage the algorithm of the first device based on the various information or based on the processed information; when the energy consumption information includes one or more energy consumption indicators, the second device can manage the algorithm of the first device based on the energy consumption indicators.
- the method of processing various information includes but is not limited to obtaining one or more indicators by calculating various information, such as using the calculation method in S102. It can be understood that for different algorithms, different use cases and the like, there may be different schemes for the second device to manage the algorithm of the first device based on the first information and monitoring information, which is not limited in this application.
- the second device may perform lifecycle management (LCM) of the algorithm on the first device based on the performance monitoring comparison information and the energy consumption comparison information in the energy consumption information.
- the energy consumption comparison information affects the performance comparison threshold used for the lifecycle management (LCM) of the algorithm
- the performance monitoring comparison information affects the energy consumption comparison threshold used for the lifecycle management (LCM) of the algorithm.
- the mapping relationship between the energy consumption comparison threshold and the performance comparison threshold may be (Te-1, Tp-1), (Te-2, Tp-2), (Te-3, Tp-3)...(Te-L, Tp-L), where Te is the energy consumption comparison threshold, Tp is the performance comparison threshold, and L is a positive integer.
- Te-i where i is a positive integer of [1, L]
- the performance comparison threshold is Tp-i.
- a performance comparison threshold is mapped.
- the second device can manage the algorithm on the first device and fallback or switch the second algorithm to the first algorithm.
- the second device may manage the algorithm of the first device only according to the first information (energy consumption information), which is not described herein.
- S101 , S102 , S103 and S105 in FIG. 1 may be performed separately.
- FIG 1 the example of a first device sending first information (energy consumption information) to a second device is used to illustrate that the second device manages the algorithm based on the energy consumption information and the monitoring information. It can be understood that, as shown in Figure 2, the second device can also send monitoring information to the first device, and the first device manages the algorithm based on the monitoring information and the energy consumption or energy consumption information.
- a first device obtains algorithm energy consumption or energy consumption information.
- the algorithm energy consumption may include the energy consumed by the algorithm to perform K use case calculations, where K is a positive integer.
- the energy consumption information may include energy consumption comparison information obtained by comparing the algorithm energy consumption of at least two algorithms performing the same task.
- the description of the algorithm energy consumption is the same as S101, and the description of the energy consumption information is the same as S102.
- the energy consumption information may further include power information of the first device and energy consumption configuration information of the first device.
- the method for acquiring the algorithm energy consumption and energy consumption comparison information is the same as S101.
- the energy consumption information can be expressed in the same way as S102.
- S201 may be executed before S202, or after S202, or simultaneously with S202.
- S201 may be executed before S203, or after S203, or simultaneously with S203.
- S201 may also not be executed, which is not limited in the embodiments of the present application.
- S202 The second device obtains monitoring information.
- the description of the second device obtaining monitoring information is the same as S104.
- S203 The second device sends second information (monitoring information).
- the second information may include monitoring information.
- the format of the monitoring information may be agreed upon by the second device and the first device.
- the monitoring information may include monitoring information of one or more algorithms, or monitoring comparison information between algorithms, and other information, and may also include one or more monitoring indicators calculated from the above information.
- the monitoring information may be performance monitoring information.
- the performance monitoring information may include performance monitoring information of one or more algorithms, or performance monitoring comparison information between algorithms.
- the description of the performance monitoring information is the same as S104.
- the second information may further include one or more of the following: a use case indication, an algorithm indication, and a communication system configuration indication.
- a use case indication an algorithm indication
- a communication system configuration indication The description of the use case indication, the algorithm indication, and the communication system configuration indication is the same as that of S103.
- the first device receives the second information, and manages the algorithm of the first device according to the monitoring information and the algorithm energy consumption or energy consumption information.
- the first device may obtain monitoring information based on the second information.
- the monitoring information may be performance monitoring information.
- the first device may perform lifecycle management (LCM) of the algorithm for the first device based on the energy consumption information and the performance monitoring information.
- LCM lifecycle management
- the energy consumption information affects the performance comparison threshold for life cycle management (LCM) of the algorithm, or the performance monitoring comparison information affects the energy consumption comparison threshold for life cycle management (LCM) of the algorithm.
- the energy consumption information affecting the performance comparison threshold for life cycle management (LCM) of the algorithm and the performance monitoring comparison information being the quotient of the first monitoring information and the second monitoring information as an example, a possible management solution is the same as S105.
- a performance comparison threshold can be obtained according to the energy consumption information mapping.
- the performance monitoring comparison information is less than the performance comparison threshold, it is considered that the first algorithm does not have a performance advantage over the second algorithm.
- the first device can request the second device to switch or fall back to the second algorithm with lower energy consumption, or the first device determines to switch or fall back to the second algorithm with lower energy consumption.
- the first algorithm is the AI/ML algorithm
- the second algorithm is the traditional algorithm as an example, when the first algorithm does not have a performance advantage over the second algorithm, the first device can request to fall back to the traditional algorithm and perform codebook-based CSI compression.
- the first device may manage the algorithm of the first device only according to the second information, which is not repeated here.
- FIG3 shows a schematic diagram of a communication structure provided in an embodiment of the present application.
- the communication device 300 includes at least one processor 301.
- the processor 301 can be used for internal processing of the device to implement certain control processing functions.
- the processor 301 includes instructions.
- the processor 301 can store data.
- different processors can be independent devices, can be located in different physical locations, and can be located on different integrated circuits.
- different processors can be integrated into one or more processors, for example, integrated on one or more integrated circuits.
- the communication device 300 includes one or more memories 303 for storing instructions.
- data may also be stored in the memory 303.
- the processor and the memory may be provided separately or integrated together.
- the communication device 300 includes a communication line 302 and at least one communication interface 304. Since the memory 303, the communication line 302 and the communication interface 304 are all optional, they are all indicated by dotted lines in FIG. 3 .
- the communication device 300 may further include a transceiver and/or an antenna.
- the transceiver may be used to send information to other devices or receive information from other devices.
- the transceiver may be referred to as a transceiver, a transceiver circuit, an input/output interface, etc., and is used to implement the transceiver function of the communication device 300 through an antenna.
- the transceiver includes a transmitter and a receiver.
- the transmitter may be used to generate a radio frequency signal from a baseband signal
- the receiver may be used to convert the radio frequency signal into a baseband signal.
- Processor 301 may include a general-purpose central processing unit (CPU), a microprocessor, an application specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present application.
- CPU central processing unit
- ASIC application specific integrated circuit
- the communication link 302 may include a pathway for transmitting information between the above-mentioned components.
- the communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), wired access networks, etc.
- RAN radio access network
- WLAN wireless local area networks
- wired access networks etc.
- the memory 303 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.
- the memory 303 may exist independently and be connected to the processor 301 through the communication line 302. Alternatively, the memory 303 may also be integrated with the processor 301.
- the memory 303 is used to store computer-executable instructions for executing the solution of the present application, and the execution is controlled by the processor 301.
- the processor 301 is used to execute the computer-executable instructions stored in the memory 303, thereby implementing the steps performed by the first device described in the embodiment shown in FIG. 1 or FIG. 2, or implementing the steps performed by the second device described in the embodiment shown in FIG. 1 or FIG. 2.
- the computer-executable instructions in the embodiments of the present application may also be referred to as application code, which is not specifically limited in the embodiments of the present application.
- the processor 301 may include one or more CPUs, such as CPU0 and CPU1 in FIG. 3 .
- the communication device 300 may include multiple processors, such as the processor 301 and the processor 305 in FIG3. Each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor.
- the processor here may refer to one or more devices, circuits, and/or processing cores for processing data (such as computer program instructions).
- the chip When the device shown in FIG. 3 is a chip, such as a chip of the first device or a chip of the second device, the chip includes a processor 301 (may also include a processor 305), a communication line 302 and a communication interface 304, and optionally, may include a memory 303.
- the communication interface 304 may be an input interface, a pin or a circuit, etc.
- the memory 303 may be a register, a cache, etc.
- the processor 301 and the processor 305 may be a general-purpose CPU, a microprocessor, an ASIC, or one or more integrated circuits for controlling the execution of the program of the algorithm management method of any of the above embodiments.
- the embodiment of the present application can divide the functional modules of the device according to the above method example.
- each functional module can be divided corresponding to each function, or two or more functions can be integrated into one processing module.
- the above integrated module can be implemented in the form of hardware or in the form of software functional modules.
- the division of modules in the embodiment of the present application is schematic, which is only a logical function division, and there may be other division methods in actual implementation.
- Figure 4 shows a schematic diagram of a device, and the device 400 can be the first device or the second device involved in the above-mentioned various method embodiments, or a chip in the first device or a chip in the second device.
- the device 400 includes a sending unit 401, a processing unit 402 and a receiving unit 403.
- the device 400 can be used to implement the steps performed by the first device or the second device in the algorithm management method of the embodiment of the present application.
- the relevant features can refer to the embodiments shown in Figure 1 or Figure 2 above, and will not be repeated here.
- the functions/implementation processes of the sending unit 401, the receiving unit 403, and the processing unit 402 in FIG4 may be implemented by the processor 301 in FIG3 calling the computer execution instructions stored in the memory 303.
- the functions/implementation processes of the processing unit 402 in FIG4 may be implemented by the processor 301 in FIG3 calling the computer execution instructions stored in the memory 303
- the functions/implementation processes of the sending unit 401 and the receiving unit 403 in FIG4 may be implemented by the communication interface 304 in FIG3.
- the functions/implementation processes of the sending unit 401 and the receiving unit 403 can also be implemented through pins or circuits.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Software Systems (AREA)
- Mathematical Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- General Engineering & Computer Science (AREA)
- Computational Mathematics (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Operations Research (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
本申请涉及一种算法管理方法和装置。第一设备获取第一算法和第二算法的能耗对比信息,根据所述能耗对比信息,向第二设备发送所述第一信息;其中,所述第一信息用于第二设备对所述第一设备进行算法管理。
Description
本申请要求于2022年12月29日提交国家知识产权局、申请号为202211711157.6、申请名称为“一种算法管理方法和装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本申请要求于2022年11月10日提交国家知识产权局、申请号为202211406336.9、申请名称为“一种设备监控管理方法和装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本发明实施例涉及通信技术领域,尤其涉及一种算法管理方法和装置。
在无线通信***中引入人工智能(AI)/机器学习(ML)技术是近些年的研究热点。其中,AI/ML性能监控是重要的研究方向,即监控AI/ML算法的性能表现。AI/ML性能监控结果可能会用于算法的管理,例如生命周期管理(lifecycle management,LCM)。
发明内容
本申请提供一种算法管理方法和装置,用于根据能耗进行算法管理。
第一方面,提供一种应用于第一设备的算法管理方法,包括:获取第一算法和第二算法的能耗对比信息;根据上述能耗对比信息,向第二设备发送上述第一信息。
其中,上述第一信息用于第二设备对上述第一设备进行算法管理。
上述第一信息可以包括能耗对比信息,还可以包括以下一项或多项:用例指示;算法指示;或,通信***配置指示。
上述第一信息可以用于上述第二设备对上述第一设备进行算法的生命周期管理(lifecycle management,LCM)。
上述第一设备可以为终端设备。
第一算法和第二算法可以为同类型或不同类型的算法。
一种可能的实施方式,上述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,上述第二算法为传统算法。
另一种可能的实施方式,上述第一算法和上述第二算法为不同的人工智能/机器学习(AI/ML)算法。
在一种可选的实施方式中,能耗对比信息的获取方法,具体包括:获取上述第一算法的第一能耗信息和上述第二算法的第二能耗信息;根据上述第一能耗信息和上述第二能耗信息,获得上述能耗对比信息。
其中,能耗对比信息可以是不同算法的算法能耗之商,之差或大小对比结果,或采用其它能够表示不同算法能耗对比的计算方法的结果。
在一种可选的实施方式中,上述第一设备可以将能耗对比信息经处理后作为第一信息的一部分向第二设备发送,具体为:根据能耗对比信息,和上述第一设备的电量信息和/或能耗配置信息,获得第三能耗信息;向上述第二设备发送包括上述第三能耗信息的上述第一信息。其中,上述第三能耗信息可以包括能耗对比信息,第一设备的电量信息,第一设备的能耗配置信息等各项信息,也可以包括由上述各项信息计算得到的一个或多个能耗指标。
需要说明的是,向第二设备发送第一信息的触发条件可以不同。
在一种可能的实施方式,可以响应于确定满足预设发送条件,上述第一设备向上述第二设备发送上述第一信息。其中,预设发送条件可以包括周期性发送。
另一种可能的实施方式,可以响应于接收到上述第二设备发送的第一指令,发送上述第一信
息。
其中,上述第一指令可以包括:用于指示上述第一设备配置的第一指令;或,用于指示所述第一设备更改配置的第一指令;或,用于查询上述第一信息的第一指令。其中,用于指示第一设备配置的第一指令可以是包括配置信息的配置消息。用于指示所述第一设备更改配置的第一指令可以是更改配置消息。用于查询上述第一信息的第一指令可以为查询消息。
另一种可能的实施方式,可以响应于第一设备连接第二设备,发送上述第一信息。
上述算法管理方法,通过根据能耗对比信息向第二设备发送第一信息,使第二设备能够综合第一信息管理第一设备的算法。
第二方面,提供一种应用于第一设备的算法管理方法,包括:接收第二设备发送的第二信息;根据上述第二信息,管理上述第一设备的算法。
其中,上述第二信息包括上述第二设备获取的上述第一设备的算法的监控信息。
上述算法的监控信息可以为算法的性能监控信息。
上述第二信息可以包括以下一项或多项:用例指示;算法指示;或,通信***配置指示。
上述第二信息可以用于上述第一设备对上述第一设备进行算法的生命周期管理(lifecycle management,LCM)。
上述第一设备可以为终端设备。
在一种可选的实施方式中,上述第一设备可以获取第一算法的所述第一能耗信息和第二算法的所述第二能耗信息。
在一种可选的实施方式中,上述第一设备可以根据上述第一能耗信息,上述第二能耗信息,上述第二信息,管理上述第一设备的算法。
第一算法和第二算法可以为同类型或不同类型的算法。
一种可能的实施方式,上述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,上述第二算法为传统算法。
另一种可能的实施方式,上述第一算法和上述第二算法为不同的人工智能/机器学习(AI/ML)算法。
上述算法管理方法,通过接收第二设备发送第二信息,综合第一能耗信息,第二能耗信息,第二信息管理上述第一设备的算法。
第三方面,提供一种用于第二设备的算法管理方法,包括:接收第一设备发送的第一信息;根据上述第一信息,管理上述第一设备的算法。
其中,上述第一信息包括上述第一算法和上述第二算法的能耗对比信息。
上述第一信息可以包括能耗对比信息,还可以包括以下一项或多项:用例指示;算法指示;或,通信***配置指示。
根据上述第一信息,第二设备可以对上述第一设备进行算法的生命周期管理(lifecycle management,LCM)。
上述第二设备可以是网络设备。上述网络设备,可以包括接入网设备,和/或核心网设备。
在一种可选的实施方式中,第二设备可以获取上述第一设备的上述第一算法的第一监控信息,和/或,上述第一设备的上述第二算法的第二监控信息。
上述第一监控信息,上述第二监控信息为性能监控信息。
在一种可选的实施方式中,第二设备可以根据上述第一信息,上述第一监控信息和/或上述第二监控信息,管理上述第一设备的算法。
根据上述第一信息,上述第一监控信息,上述第二监控信息,第二设备可以对上述第一设备进行算法的生命周期管理(lifecycle management,LCM)。
第一算法和第二算法可以为同类型或不同类型的算法。
一种可能的实施方式,上述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,上述第二算法为传统算法。
另一种可能的实施方式,上述第一算法和上述第二算法为不同的人工智能/机器学习(AI/ML)算法。
上述算法管理方法,通过接收第一设备发送的第一信息,综合第一信息,第一监控信息,第二监控信息管理第一设备的算法。
第四方面,提供一种应用于第二设备的算法管理方法,包括:获取第一设备的第一算法的第一监控信息,和/或上述第一设备的第二算法的第二监控信息;根据上述第一监控信息,和/或上述第二监控信息,向上述第一设备发送第二信息。
其中,上述第二信息用于上述第一设备对上述第一设备进行算法管理。
第二信息还可以包括以下一项或多项:用例指示;算法指示;或,通信***配置指示。上述第二信息可以用于上述第一设备对上述第一设备进行算法的生命周期管理(LCM)。
上述第一监控信息,上述第二监控信息可以为性能监控信息。
上述第二设备可以是网络设备。上述网络设备,可以包括接入网设备,和/或核心网设备。
第一算法和第二算法可以为同类型或不同类型的算法。
一种可能的实施方式,上述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,上述第二算法为传统算法。
另一种可能的实施方式,上述第一算法和上述第二算法为不同的人工智能/机器学习(AI/ML)算法。
在一种可选的实施方式中,将第一监控信息,和/或上述第二监控信息经处理后作为第二信息的一部分向上述第一设备发送,具体为包括:根据上述第一监控信息,和/或上述第二监控信息,获得第三监控信息;向上述第一设备发送包括上述第三监控信息的上述第二信息。
其中,上述第三监控信息可以为性能监控信息。
上述算法管理方法,通过发送第二信息,使第一设备可以综合管理第一设备的算法。
第五方面,提供一种装置。该通信装置可以为上述第一方面和第三方面中的任一方面所述的第一设备。所述装置具备上述第一设备的功能。所述装置例如为终端设备,包括:固定设备,移动设备、手持设备(例如手机)、穿戴设备、车载设备,或内置于上述设备中的无线装置(例如,通信模块,调制解调器,或芯片***等)。
一种可选的实现方式中,所述装置包括处理单元(有时也称为处理模块)和收发单元(有时也称为收发模块)。收发单元能够实现发送功能和接收功能,在收发单元实现发送功能时,可称为发送单元(有时也称为发送模块),在收发单元实现接收功能时,可称为接收单元(有时也称为接收模块)。发送单元和接收单元可以是同一个功能模块,该功能模块称为收发单元,该功能模块能实现发送功能和接收功能;或者,发送单元和接收单元可以是不同的功能模块,收发单元是对这些功能模块的统称。
在一种可选的实施方式中,所述收发单元(或,所述接收单元),用于向第二设备发送第一信息(能耗信息),所述第一信息(能耗信息)用于第二设备对所述第一设备进行算法管理;所述收发单元(或,所述接收单元),用于接收来自第二设备的管理算法的指令。
在一种可选的实施方式中,所述装置还包括存储单元(有时也称为存储模块),所述处理单元用于与所述存储单元耦合,并执行所述存储单元中的程序或指令,使能所述装置执行上述第一方面和第三方面中的任一方面所述的第一设备的功能。
第六方面,提供一种装置。所述通信装置可以为上述第二方面和第四方面中的任一方面所述的第二设备。所述装置具备上述第二设备的功能。所述装置例如为网络设备。一种可选的实现方式中,所述装置包括处理单元(有时也称为处理模块)和收发单元(有时也称为收发模块)。收发单元能够实现发送功能和接收功能,在收发单元实现发送功能时,可称为发送单元(有时也称为发送模块),在收发单元实现接收功能时,可称为接收单元(有时也称为接收模块)。发送单元和接收单元可以是同一个功能模块,该功能模块称为收发单元,该功能模块能实现发送功能和
接收功能;或者,发送单元和接收单元可以是不同的功能模块,收发单元是对这些功能模块的统称。
在一种可选的实施方式中,所述收发单元(或,所述接收单元),用于向第一设备发送第二信息(监控信息),所述第二信息(监控信息)用于第一设备管理算法;所述收发单元(或,所述接收单元)。
在一种可选的实施方式中,所述装置还包括存储单元(有时也称为存储模块),所述处理单元用于与所述存储单元耦合,并执行所述存储单元中的程序或指令,使能所述装置执行上述第二方面和第四方面中的任一方面所述的第二设备的功能。
第七方面,本发明提供一种存储一个或多个程序的非易失性计算机可读存储介质,该一个或多个程序包括指令,指令当被包括第一方面或上述任一种可能的实现方式所述的第一设备执行时,使得第一设备执行:
获取第一算法和第二算法的能耗对比信息;根据上述能耗对比信息,向第二设备发送上述第一信息。
第八方面,本发明提供一种存储一个或多个程序的非易失性计算机可读存储介质,该一个或多个程序包括指令,指令当被包括第二方面或上述任一种可能的实现方式所述的第一设备执行时,使得第一设备执行:
获取第一算法的第一能耗信息和第二算法的第二能耗信息;接收所述第二设备发送的第二信息;根据所述第一能耗信息,所述第二能耗信息,所述第二信息,管理所述第一设备的算法。
第九方面,本发明提供一种存储一个或多个程序的非易失性计算机可读存储介质,该一个或多个程序包括指令,指令当被包括第三方面或上述任一种可能的实现方式所述的第一设备执行时,使得第二设备执行:
获取第一设备的第一算法的第一监控信息和上述第一设备的第二算法的第二监控信息;接收第一设备发送的第一信息;根据上述第一信息,上述第一监控信息,上述第二监控信息,管理上述第一设备的算法。
第十方面,本发明提供一种存储一个或多个程序的非易失性计算机可读存储介质,该一个或多个程序包括指令,指令当被包括第四方面或上述任一种可能的实现方式所述的第一设备执行时,使得第二设备执行:
获取第一设备的第一算法的第一监控信息和所述第一设备的第二算法的第二监控信息;
根据所述第一监控信息和所述第二监控信息,向所述第一设备发送第二信息。
第十一方面,提供一种算法管理***,包括第一设备和第二设备。
其中,第一设备可以是如第一方面或第二方面或第五方面或第七方面或第八方面或上述任一可实现方式所述的第一设备、第二设备可以是如第三方面或第四方面或第六方面或第九方面或第十方面或上述任一可实现方式所述的第二设备。
图1为本发明实施例提供的一种算法管理方法流程示意图;
图2为本发明实施例提供的另一种算法管理方法流程示意图;
图3为本发明实施例提供的一种装置的示意图;
图4为本发明实施例提供的另一种装置的示意图。
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚地描述。显然,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员所获得的所有其他实施例,都属于本发明保护的范围。
本申请实施例的***架构至少包括第一设备和第二设备两个设备,所述第一设备和第二设备可以具有通信功能。示例性的,第一设备可以是终端设备,第二设备可以是网络设备。
上述终端设备是一种具有无线收发功能的设备,可以是固定设备,移动设备、手持设备(例如手机)、穿戴设备、车载设备,或内置于上述设备中的无线装置(例如,通信模块,调制解调
器,或芯片***等)。所述终端设备用于连接人,物,机器等,可广泛用于各种场景,例如包括但不限于以下场景:蜂窝通信、设备到设备通信(device-to-device,D2D)、车到一切(vehicle to everything,V2X)、机器到机器/机器类通信(machine-to-machine/machine-type communications,M2M/MTC)、物联网(internet of things,IoT)、虚拟现实(virtual reality,VR)、增强现实(augmented reality,AR)、工业控制(industrial control)、无人驾驶(self driving)、远程医疗(remote medical)、智能电网(smart grid)、智能家具、智能办公、智能穿戴、智能交通,智慧城市(smart city)、无人机、机器人等场景的终端设备。所述终端设备有时可称为用户设备(user equipment,UE)、终端、接入站、UE站、远方站、无线通信设备、或用户装置等等。
上述网络设备,可以包括接入网设备,和/或核心网设备。所述接入网设备为具有无线收发功能的设备,用于与所述终端设备进行通信。所述接入网设备包括但不限于基站(基站收发信站点(base transceiver station,BTS),Node B,eNodeB/eNB,或gNodeB/gNB)、收发点(transmission reception point,TRP),第三代合作伙伴计划(3rd generation partnership project,3GPP)后续演进的基站,无线保真(wireless fidelity,Wi-Fi)***中的接入节点,无线中继节点,无线回传节点等。所述基站可以是:宏基站,微基站,微微基站,小站,中继站等。多个基站可以支持同一种接入技术的网络,也可以支持不同接入技术的网络。基站可以包含一个或多个共站或非共站的传输接收点。所述接入网设备还可以是云无线接入网络(cloud radio access network,CRAN)场景下的无线控制器、集中单元(centralized unit,CU),和/或分布单元(distributed unit,DU)。所述接入网设备还可以是服务器等。例如,车到一切(vehicle to everything,V2X)技术中的接入网设备可以为路侧单元(road side unit,RSU)。基站可以与终端设备进行通信,也可以通过中继站与终端设备进行通信。终端设备可以与不同接入技术中的多个基站进行通信。所述核心网设备用于实现移动管理,数据处理,会话管理,策略和计费等功能。不同接入技术的***中实现核心网功能的设备名称可以不同,本申请实施例并不对此进行限定。以5G***为例,所述核心网设备包括:访问和移动管理功能(access and mobility management function,AMF)、会话管理功能(session management function,SMF)、策略控制功能(policy control function,PCF)或用户面功能(user plane function,UPF)等。
本申请实施例中,对于名词的数目,除非特别说明,表示“单数名词或复数名词”,即"一个或多个”。“至少一个”是指一个或者多个,“多个”是指两个或两个以上。“和/或”,描述关联对象的关联关系,表示可以存在三种关系,例如,A和/或B,可以表示:单独存在A,同时存在A和B,单独存在B的情况,其中A,B可以是单数或者复数。字符“/”一般表示前后关联对象是一种“或”的关系。例如,A/B,表示:A或B。“以下至少一项(个)”或其类似表达,是指的这些项中的任意组合,包括单项(个)或复数项(个)的任意组合。例如,a,b,或c中的至少一项(个),表示:a,b,c,a和b,a和c,b和c,或a和b和c,其中a,b,c可以是单个,也可以是多个。
本申请实施例提及“第一”、“第二”等序数词是用于对多个对象进行区分,不用于限定多个对象的大小、内容、顺序、时序、优先级或者重要程度等。例如,第一信息和第二信息,可以是同一个信息,也可以是不同的信息,且,这种名称也并不是表示这两个信息的内容、大小、发送端/接收端、优先级或者重要程度等的不同。另外,本申请所介绍的各个实施例中对于步骤的编号,只是为了区分不同的步骤,并不用于限定步骤之间的先后顺序。例如,S401可以发生在S402之前,或者可能发生在S402之后,或者也可能与S402同时发生。
本申请实施例提供的技术方案可以应用于***移动通信技术(the 4th generation,4G)***中,例如长期演进(long term evolution,LTE)***,或可以应用于第五代移动通信技术(the 5th generation,5G)***中,例如新空口(new radio,NR)***,或者还可以应用于下一代移动通信***或其他类似的通信***,例如第六代移动通信技术(the 6th generation,6G)***等,具体的不做限制。
如下介绍本申请实施例涉及的技术特征。
无线通信***引入AI/ML技术后,AI/ML模式和传统模式预期在一段时间内可以用于无线通
信***。传统模式可以包括传统算法,AI/ML模式可以包括一个或多个不同的AI/ML模型。模型也可以称为算法,可以理解的是,AI/ML算法可以理解为AI/ML模型,或,以AI/ML模型为核心,辅以传统算法。其中,传统算法可以包括非AI/ML算法,例如已有通信***中广泛使用的算法。所述已有通信***中广泛使用的算法可以包括基于数学模型的算法,基于简单的预测、滤波等算法。其中,所述基于简单的预测、滤波等算法可以包括基于卡尔曼滤波,滑动平均,或,自回归的算法等。示例性的,对于信道状态信息压缩(CSI compression)用例,传统算法可以包括基于码本的CSI压缩算法。
在进行算法管理(例如,生命周期管理)时,需要综合考虑算法的监控信息和能耗信息。替代性的,所述能耗信息可以为功率信息,功率信息是单位时间内的能耗信息。为描述方便,本申请实施例中以能耗信息为例进行说明。
例如,无线通信***中的第二设备能够获取第一设备算法的监控信息,但不掌握第一设备算法的能耗信息时,需要第一设备和第二设备之间进行信息传输,使得第一设备或第二设备可以综合监控信息和能耗信息对第一设备算法进行管理。
其中,能耗信息可以包括以下的一项或多项:算法能耗,能耗对比信息。可选的,能耗信息还可以包括第一设备的电量信息,第一设备的能耗配置信息。
图1示例性地示出了一种通过第一设备向第二设备发送包括能耗信息的第一信息实现的算法管理方法,所述方法包括:
S101、第一设备获取能耗对比信息。
所述能耗对比信息可以包括至少两个算法执行同一个任务的算法能耗的对比信息。其中,算法可以包括传统算法,AI/ML算法。需要说明的是,执行一个任务可以理解为执行K次用例,其中K为正整数。所述算法能耗,可以包括该算法执行任务所消耗的能量。
其中,用例可以包括信道状态信息压缩(CSI compression),定位(Positioning),波束管理(Beam Management)等。一次用例的计算可以是一次CSI压缩任务(例如,基于输入压缩前的CSI输出压缩后的CSI),一次定位计算(例如,基于输入定位所需信息,输出位置信息或其它定位辅助信息),或一次波束管理计算(例如,基于输入多个波束信息,输出预期信号最强的Kb个波束信息,Kb为正整数)。
以能耗对比信息为第一设备的第一算法和第二算法的算法能耗的对比信息,用例为信道状态信息压缩(CSI compression)为例:第一算法可以为AI/ML算法,即使用AI/ML算法进行CSI压缩任务,第二算法可以为传统算法,即进行基于码本的CSI压缩任务;或,第一算法和第二算法可以为使用不同的AI/ML算法进行CSI压缩任务。
算法能耗还可以包括任务的辅助工作的能耗,任务的上下游工作的能耗等执行任务造成的其他工作能耗。以用例为信道状态信息压缩(CSI compression)为例,上述辅助工作的能耗可包括但不限于以下一项或多项工作造成的能耗:CSI数据的预处理,一个或多个AI/ML算法的性能监控,传统算法的性能监控。任务的上下游工作的能耗可包括但不限于以下一项或多项工作造成的能耗:压缩前CSI的获取,压缩后CSI的发送。
所述对比信息,可以是不同算法的算法能耗之商,之差或大小对比结果,或采用其它能够表示不同算法能耗对比的计算方法的结果。
需要说明的是,所述算法能耗和/或能耗对比信息的获取方法,此处不做限制,包括直接获取能耗对比信息方法,间接获得能耗对比信息方法。
其中,所述直接获得能耗对比信息方法,可以指直接获取多个算法的能耗对比信息的方法,例如:配置获取法,估算法,实测法。
其中,所述间接获得能耗对比信息方法,可以指在分别获取各个算法的算法能耗信息后,通过计算得到能耗对比信息。对于不同的算法,不同的用例可以采用不同的算法能耗信息获取方法,算法能耗信息获取方法可以包括以下一个或多个方法的混合:配置获取法,估算法,实测法。
上述配置获取法可以包括:由第一设备从第一设备的出厂配置信息获取,由第一设备从其他设备获取,例如,从第二设备获取,或从服务器获取。
上述估算法可使用近似公式进行计算。对于近似公式不做限制,此处提供一种可能的算法能
耗计算公式J=N/E*[(U_P*(1-U_C))/U_C+1],其中J为能耗,N为算法的运算数(例如乘和/或加运算数),E为执行算法的处理器能量效率,U_P为执行算法的处理器的本底功率比例,U_C为算法在处理器中执行时的计算资源利用率。其中,本底功率比例可理解为本底功率与峰值功率的比值。
其中,所述算法运算数N可由第一设备通过服务器获取,第二设备获取,或,接收到所述算法后自行评估获取。其中,服务器可为第一设备的服务器,第二设备的服务器或第三方服务器。所述处理器能量效率E,算法处理器的本底功率比例U_P可通过第一设备获取,从第二设备获取,或,从其他设备获取。所述计算资源利用率U_C的获取方法包括通过第二设备获取,通过第一设备执行算法获取,或,通过其他设备获取。
上述实测法,通过实际执行算法获得相应的算法能耗或能耗对比信息。
可以理解的是,上述配置获取法,估算法,实测法,是示例性方法。通过调整上述方法中配置的数据,公式参数等方式,可以直接获得能耗对比信息方法或间接获得能耗对比信息。
S102、根据能耗对比信息,获得能耗信息。
示例性的,所述能耗信息可以包括所述能耗对比信息。可选的,所述能耗信息还可以包括第一设备的电量信息,第一设备的能耗配置信息。示例性的,能耗配置信息可包括性能优先,续航优先,或,均衡模式。
可以理解的是,所述能耗信息可以包括能耗对比信息,第一设备的电量信息,第一设备的能耗配置信息等各项信息,也可以包括由上述各项信息计算得到的一个或多个能耗指标。本申请实施例对能耗信息中各项信息计算得到一个或多个能耗指标的方式不作限定。
示例性的,本申请实施例提供一种可能的能耗指标计算方法:通过对能耗对比信息,电量信息,能耗配置信息进行计算得到一个能耗指标,该指标称为两种算法的等效能耗对比信息。本申请实施例对获得等效能耗对比信息的方法不作限定,可能的综合思路为:电量越小或能耗配置信息越关注续航,则放大两算法能耗的对比值,引导优先选择低能耗模式。两种可能的计算方法如下:
第一种可能的计算方法为:根据公式α=((log第一算法能耗/第二算法能耗))/((电量"比例"×配置值))计算得到以数值形式呈现的等效能耗对比信息。其中,电量比例取值范围可以为(0%,100%),当能耗配置信息为性能优先时配置值为A1,当能耗配置信息为均衡模式时配置值为A2,当能耗配置信息为续航优先时配置值为A3,其中A1,A2和A3为正数,且A1A2A3。
第二种可能的计算方法为:根据公式得到能耗对比值,当β>0时第一算法能耗比第二算法能耗大,反之,当β<0时则第一算法能耗比第二算法能耗小。设定电量阈值,当电量小于电量阈值时,能耗对比值的绝对值增加B1,当能耗配置信息被配置为续航优先,能耗对比值的绝对值增加B2,当能耗配置信息被配置为性能优先,能耗对比值的绝对值减少B3,其中,B1,B2和B3为正数。当β=0时,第一算法能耗与第二算法能耗相等,此时,电量和能耗配置信息不影响能耗对比值β。
需要说明的是,等效能耗对比信息可以是,上述计算结果,也可以是上述计算结果经饱和截断、量化、码本映射等后处理后的结果。
S103、第一设备向第二设备发送第一信息(能耗信息),第二设备接收第一信息(能耗信息)。
所述第一信息包括所述能耗信息,和/或,根据所述能耗信息得到的算法请求信息。
上述能耗信息的呈现方式可包括以下的一种或多种:数值,能耗信息分段等级。其中,数值可以为计算结果或计算结果经饱和截断、量化、码本映射等后处理后的结果。能耗信息分段等级的呈现方式可包括档位信息。其中,档位信息的分段方法可以由第一设备和第二设备事先约定。
上述算法请求信息可以包括第一设备根据能耗信息获得的更换算法的请求。示例性的,可将算法按照算法能耗的大小进行排序,得到算法的能耗排序。设定第一设备电量阈值,当第一设备电量大于电量阈值时,认为电量充裕。当电量充裕,能耗配置信息被配置为性能优先时,第一设备可根据当前算法在算法的能耗排序中的位置,向第二设备发送算法请求信息,申请切换至性能更优的算法。其中,当前算法可以为第一设备在用例上正在使用的算法。
可以理解的是,一般情况下,性能更优的算法为算法的能耗排序中能耗更高的算法。
可选的,第一信息还可包括以下一项或多项信息:用例指示,算法指示,通信***配置指示。其中,用例指示可以包括用例信息,算法指示可以包括算法的算法标识。其中,算法标识也可以称为算法ID。示例性的,算法为传统算法时,也可以省略算法标识,通过不包括算法标识来指示采用的是传统算法。也就是说,算法指示既可以采用算法标识等明示的指示方式,也可以采用隐示的指示方式,本申请实施例不作限定。其中,通信***配置可以是网络配置,设备配置。网络配置可以包括频段,设备配置可以包括第一设备天线数。
所述发送的发送模式可包括周期性发送,非周期性发送。
所述周期性发送,可以通过第二设备配置发送的周期。
所述非周期性发送,可以包括事件触发发送。
所述事件触发发送的触发条件(事件)可以包括第一设备和第二设备连接,第一设备的配置修改,接收到第二设备发送的更改配置消息,接收到第二设备发送的配置消息,电量小于阈值,电量变化量或电量变化率大于阈值,接收到第二设备发送的查询消息等。当满足事件触发条件时,第一设备发送或请求得到许可后发送第一信息。
示例性的,假设事件触发发送的触发条件为第一设备连接第二设备,第一设备可以响应于第一设备连接第二设备,将第一信息作为第一设备的能力信息发送第二设备。第一设备和第二设备可以对发送内容和发送方式进行事先约定。其中,发送内容可以包括第一信息各字段所表示的含义。
示例性的,假设事件触发发送的触发条件为接收到第二设备发送的查询消息,第一设备可以响应于接收到查询消息,向第二设备发送第一信息。也就是说,查询消息可以用于指示第一设备发送目标算法和当前算法的第一信息。其中,目标算法可以包括第二设备目标配置第一设备使用的算法。
示例性的,假设事件触发发送的触发条件为接收到第二设备发送的配置消息,第一设备可以响应于第一设备接收到配置消息,向第二设备发送第一消息。其中,第二设备向第一设备发送的配置消息可以作为第一设备配置更改的依据。
示例性的,假设事件触发发送的触发条件为接收到第二设备发送的更改配置消息,第一设备可以响应于接收到更改配置消息,向第一设备发送第一信息。其中,所述更改配置消息既可以包括配置信息,也可以不包括配置信息。示例性的,若更改配置消息不包括配置信息,可以由第一设备确定目标算法和当前算法的第一信息。
上述配置信息可以包括以下一项或多项:发送模式,发送模式的配置信息。例如,发送模式为周期性发送时,发送模式的配置信息可以包括发送周期;发送模式为事件触发发送时,发送模式的配置信息可以包括触发事件的配置,和/或触发事件中的阈值。
上述配置信息还可以包括以下一项或多项:用例指示,算法指示,通信***配置指示。用例指示,算法指示,通信***配置指示的描述同S103。
需要说明的是,上述配置信息的部分或全部内容也可以是由第一设备和第二设备事先确定的。
S104、第二设备获得监控信息。
所述监控信息可以包括一个或多个算法的监控信息,或算法之间的监控对比信息。以第一设备的第一算法和第二算法为例,监控信息可以包括第一设备的第一算法的第一监控信息和第二算法的第二监控信息,或,根据第一监控信息和第二监控信息的对比获得的监控对比信息。对于不同用例,不同算法的监控信息的获取方式可能不同,本申请实施例对监控信息的获取方式不作限定。
以用例为CSI压缩,第一设备为终端设备,第二设备为网络设备,监控信息为性能监控信息为例。可以理解的是,在正常工作情况下,终端设备将压缩的CSI发送给网络设备。为获取性能监控信息,终端设备可以将未压缩或高保真压缩的CSI发送给网络设备,网络设备将压缩的CSI进行解压缩,将解压缩输出的CSI和未压缩或高保真压缩的CSI进行对比,获得当前算法在CSI压缩用例上压缩和解压缩的算法性能监控信息。其中,对比方法可以包括计算均方误差(mean-square error,MSE),算法性能可以用均方误差(MSE)进行表示。
可以理解的是,性能监控信息可以包括一个或多个算法的性能监控信息,或算法之间的性能监控对比信息等各项信息,也可以包括由上述各项信息计算得到的一个或多个性能监控指标。
上述性能指标可包括以下一项或多项:算法性能,通信***性能。其中,算法性能可包括但不限于以下一项或多项:精度,召回率,误差等。例如,面向CSI压缩的通用余弦相似度(generalized cosine similarity,GCS)或归一化均方差(normalized mean square error,NMSE),面向定位的位置精度。其中,通信***性能可以包括但不限于以下一项或多项:吞吐率、误码率(bit error rate,BER)、误块率(block error rate,BLER)、ACK(acknowledgment)比例/NACK(not acknowledgment)比例、重传率等。
可以理解的是,性能监控信息的表示方式可以指示设定或事先约定。性能监控信息的表示方式可以包括性能结果,统计结果,性能对比结果。其中,性能结果和/或统计结果,包括但不限于均值,方差,百分位,直方图,等级。其中,性能对比结果可以包括两种算法的性能监控对比结果。
可选的,监控信息还可包括以下一个或多个信息:用例指示,算法指示,通信***配置指示。用例指示,算法指示,通信***配置指示的描述同S103。当第一信息中包括用例指示,第二设备可以从所述第一设备中获得用例指示;当第一信息中未包括用例指示,从自身获得用例指示。
需要说明的是,S104可以在S101之前执行,或者在S101之后执行,或者与S101同时执行。S104可以在S102之前执行,或者在S102之后执行,或者与S102同时执行。另外,S104可以在S103之前执行,或者在S103之后执行,或者与S103同时执行。S104还可以不执行,本申请实施例不作限定。
S105、第二设备根据所述第一信息(能耗信息)和监控信息,管理第一设备的算法。
可选的,第一信息(能耗信息)和监控信息可以包括以下一个或多个信息:用例指示,算法指示,通信***配置指示。用例指示,算法指示,通信***配置指示的描述同S103。
第二设备获得的监控信息可以是性能监控信息,监控信息中的监控对比信息可以是性能监控对比信息。
第二设备可以根据第一信息获得能耗信息。其中,当能耗信息包括能耗对比信息,第一设备的电量信息,第一设备的能耗配置信息等各项信息,第二设备可以根据各项信息或根据处理后的各项信息管理第一设备的算法;当能耗信息包括一个或多个能耗指标,第二设备可以根据能耗指标管理第一设备算法。其中,处理各项信息的方式包括但不限于通过对各项信息计算得到一个或多个指标,例如采用S102中的计算方法。可以理解的是,对于不同的算法,不同的用例等情况,第二设备根据第一信息和监控信息管理第一设备的算法,可能存在不同的方案,本申请不作限定。
示例性的,第二设备可以根据性能监控对比信息和能耗信息中的能耗对比信息,对第一设备进行算法的生命周期管理(LCM)。一种可能的方式中,能耗对比信息影响用于算法的生命周期管理(LCM)的性能对比阈值,或,性能监控对比信息影响用于算法的生命周期管理(LCM)的能耗对比阈值。以能耗对比信息影响用于算法的生命周期管理(LCM)的性能对比阈值,性能监控对比信息为第一监控信息和第二监控信息的商为例,提供一种可能的管理方案:
设定性能对比阈值和能耗对比阈值,建立能耗对比阈值与性能对比阈值之间的映射关系,能耗对比阈值Te越大,则性能对比阈值Tp越大。例如,能耗对比阈值与性能对比阈值之间的映射关系可为(Te-1,Tp-1),(Te-2,Tp-2),(Te-3,Tp-3)…(Te-L,Tp-L),其中Te为能耗对比阈值,Tp为性能对比阈值,L为正整数。对于任一能耗对比阈值Te-i,其中i为[1,L]的正整数,当满足Te-i≤能耗对比信息≤Te-(i+1)时,性能对比阈值为Tp-i。
根据能耗对比信息,映射得到性能对比阈值,当性能监控对比信息大于性能对比阈值时,第二设备可以管理第一设备上的算法,将第二算法回落(fallback)或切换(switch)到第一算法。
可选的,当S104不执行时,第二设备可以仅根据所述第一信息(能耗信息),管理第一设备的算法,在此不再赘述。
可以理解的是,上述判断条件及第二设备对第一设备的管理方法仅是示例,第二设备还可以使用其它判断条件和/或管理方法,本实施例不作限定。
在部分实施例中,图1中的S101,S102,S103和S105可以单独执行。
图1中以第一设备向第二设备发送第一信息(能耗信息)为例进行说明,第二设备根据能耗信息和监控信息管理算法,可以理解的是,如图2所示,还可由第二设备向第一设备发送监控信息,第一设备根据监控信息,和,算法能耗或能耗信息管理算法。
S201、第一设备获得算法能耗或能耗信息。
可以理解的是,所述算法能耗,可以包括算法执行K次用例计算所消耗的能量,其中K为正整数。所述能耗信息,可以包括由至少两个算法执行同一个任务的算法能耗的对比信息得到的能耗对比信息。算法能耗的描述同S101,能耗信息的描述同S102。
可选的,能耗信息还可以包括第一设备的电量信息,第一设备的能耗配置信息。
其中,算法能耗,能耗对比信息的获取方法同S101。能耗信息的表示方式可以同S102。
需要说明的是,S201可以在S202之前执行,或者在S202之后执行,或者与S202同时执行。另外,S201可以在S203之前执行,或者在S203之后执行,或者与S203同时执行。S201还可以不执行,本申请实施例不作限定。
S202、第二设备获得监控信息。
第二设备获得监控信息的描述同S104。
S203、第二设备发送第二信息(监控信息)。
所述第二信息可包括监控信息。监控信息的格式可由第二设备与第一设备约定。监控信息可以包括一个或多个算法的监控信息,或算法之间的监控对比信息等各项信息,也可以包括由上述各项信息计算得到的一个或多个监控指标。
示例性的,监控信息可为性能监控信息。性能监控信息可包括一个或多个算法的性能监控信息,或算法之间的性能监控对比信息。性能监控信息的描述同S104。
可选的,第二信息还可以包括以下一项或多项:用例指示,算法指示,通信***配置指示。用例指示,算法指示,通信***配置指示的描述同S103。
S204、第一设备接收第二信息,根据所述监控信息和,算法能耗或能耗信息,管理第一设备的算法。
第一设备可以根据第二信息,获得监控信息。监控信息可以是性能监控信息。示例性的,第一设备可以根据能耗信息和性能监控信息,对第一设备进行算法的生命周期管理(LCM)。
一种可能的方式中,能耗信息影响用于算法的生命周期管理(LCM)的性能对比阈值,或,性能监控对比信息影响用于算法的生命周期管理(LCM)的能耗对比阈值。以能耗信息影响用于算法的生命周期管理(LCM)的性能对比阈值,性能监控对比信息为第一监控信息和第二监控信息的商为例,一种可能的管理方案同S105。
可以理解的是,在该管理方案中,可以根据能耗信息映射得到性能对比阈值,当性能监控对比信息小于性能对比阈值时,认为第一算法相较于第二算法不具有性能优势。若第一设备电量小于设定的电量阈值时,第一设备可以向第二设备请求切换或回退到能耗较低的第二算法,或,由第一设备确定切换或回退到能耗较低的第二算法。以用例为CSI,第一算法为AI/ML算法,第二算法为传统算法为例,当第一算法相较于第二算法不具有性能优势,第一设备可以请求回退至传统算法,进行基于码本的CSI压缩。
可选的,当S201不执行时,第一设备可以仅根据所述第二信息,管理第一设备的算法,在此不再赘述。
图3示出了本申请实施例提供的一种通信的结构示意图。
该通信装置300包括至少一个处理器301。处理器301可以用于装置的内部处理,实现一定的控制处理功能。可选地,处理器301包括指令。可选地,处理器301可以存储数据。可选地,不同的处理器可以是独立的器件,可以位于不同物理位置,可以位于不同的集成电路上。可选地,不同的处理器可以集成在一个或多个处理器中,例如,集成在一个或多个集成电路上。
可选地,通信装置300包括一个或多个存储器303,用以存储指令。可选地,所述存储器303中还可以存储有数据。所述处理器和存储器可以单独设置,也可以集成在一起。
可选地,通信装置300包括通信线路302,以及至少一个通信接口304。其中,因为存储器303、通信线路302以及通信接口304均为可选项,因此在图3中均以虚线表示。
可选地,通信装置300还可以包括收发器和/或天线。其中,收发器可以用于向其他装置发送信息或从其他装置接收信息。所述收发器可以称为收发机、收发电路、输入输出接口等,用于通过天线实现通信装置300的收发功能。可选地,收发器包括发射机(transmitter)和接收机(receiver)。示例性地,发射机可以用于将基带信号生成射频(radio frequency)信号,接收机可以用于将射频信号转换为基带信号。
处理器301可以包括一个通用中央处理器(central processing unit,CPU),微处理器,特定应用集成电路(application specific integrated circuit,ASIC),或一个或多个用于控制本申请方案程序执行的集成电路。
通信线路302可包括一通路,在上述组件之间传送信息。
通信接口304,使用任何收发器一类的装置,用于与其他设备或通信网络通信,如以太网,无线接入网(radio access network,RAN),无线局域网(wireless local area networks,WLAN),有线接入网等。
存储器303可以是只读存储器(read-only memory,ROM)或可存储静态信息和指令的其他类型的静态存储设备,随机存取存储器(random access memory,RAM)或者可存储信息和指令的其他类型的动态存储设备,也可以是电可擦可编程只读存储器(electrically erasable programmable read-only memory,EEPROM)、只读光盘(compact disc read-only memory,CD-ROM)或其他光盘存储、光碟存储(包括压缩光碟、激光碟、光碟、数字通用光碟、蓝光光碟等)、磁盘存储介质或者其他磁存储设备、或者能够用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算机存取的任何其他介质,但不限于此。存储器303可以是独立存在,通过通信线路302与处理器301相连接。或者,存储器303也可以和处理器301集成在一起。
其中,存储器303用于存储执行本申请方案的计算机执行指令,并由处理器301来控制执行。处理器301用于执行存储器303中存储的计算机执行指令,从而实现图1或图2所示的实施例中所述的第一设备所执行的步骤,或,实现图1或图2所示的实施例所述的第二设备所执行的步骤。
可选的,本申请实施例中的计算机执行指令也可以称之为应用程序代码,本申请实施例对此不作具体限定。
在具体实现中,作为一种实施例,处理器301可以包括一个或多个CPU,例如图3中的CPU0和CPU1。
在具体实现中,作为一种实施例,通信装置300可以包括多个处理器,例如图3中的处理器301和处理器305。这些处理器中的每一个可以是一个单核(single-CPU)处理器,也可以是一个多核(multi-CPU)处理器。这里的处理器可以指一个或多个设备、电路、和/或用于处理数据(例如计算机程序指令)的处理核。
当图3所示的装置为芯片时,例如是第一设备的芯片,或第二设备的芯片,则该芯片包括处理器301(还可以包括处理器305)、通信线路302和通信接口304,可选的,该可包括存储器303。具体地,通信接口304可以是输入接口、管脚或电路等。存储器303可以是寄存器、缓存等。处理器301和处理器305可以是一个通用的CPU,微处理器,ASIC,或一个或多个用于控制上述任一实施例的算法管理方法的程序执行的集成电路。
本申请实施例可以根据上述方法示例对装置进行功能模块的划分,例如,可以对应各个功能划分各个功能模块,也可以将两个或两个以上的功能集成在一个处理模块中。上述集成的模块既可以采用硬件的形式实现,也可以采用软件功能模块的形式实现。需要说明的是,本申请实施例中对模块的划分是示意性的,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式。比如,在采用对应各个功能划分各个功能模块的情况下,图4示出了一种装置示意图,该装置400可以是上述各个方法实施例中所涉及的第一设备或第二设备,或者为第一设备中的芯片或第二设备中的芯片。该装置400包括发送单元401、处理单元402和接收单元403。
应理解,该装置400可以用于实现本申请实施例的算法管理方法中由第一设备或第二设备执行的步骤,相关特征可以参照上文图1或图2所示实施例,此处不再赘述。
可选的,图4中的发送单元401、接收单元403以及处理单元402的功能/实现过程可以通过图3中的处理器301调用存储器303中存储的计算机执行指令来实现。或者,图4中的处理单元402的功能/实现过程可以通过图3中的处理器301调用存储器303中存储的计算机执行指令来实现,图4中的发送单元401和接收单元403的功能/实现过程可以通过图3中的通信接口304来实现。
可选的,当该装置400是芯片或电路时,则发送单元401和接收单元403的功能/实现过程还可以通过管脚或电路等来实现。
Claims (32)
- 一种算法管理方法,应用于第一设备,包括:获取第一算法和第二算法的能耗对比信息;根据所述能耗对比信息,向第二设备发送所述第一信息;其中,所述第一信息用于所述第二设备对所述第一设备进行算法管理。
- 根据权利要求1所述的方法,其特征在于:所述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,所述第二算法为传统算法;或所述第一算法和所述第二算法为不同的人工智能/机器学习(AI/ML)算法。
- 根据权利要求1-2任一项所述的方法,其特征在于,所述获取第一算法和第二算法的能耗对比信息,包括:获取所述第一算法的第一能耗信息和所述第二算法的第二能耗信息;根据所述第一能耗信息和所述第二能耗信息,获得所述能耗对比信息。
- 根据权利要求1-3任一项所述的方法,其特征在于,所述根据所述能耗对比信息,向第二设备发送所述第一信息,具体包括:根据能耗对比信息,和所述第一设备的电量信息和/或能耗配置信息,获得第三能耗信息;向所述第二设备发送包括所述第三能耗信息的所述第一信息。
- 根据权利要求1-4任一项所述的方法,其特征在于,所述第一信息,还包括以下一项或多项:用例指示;算法指示;或,通信***配置指示。
- 根据权利要求1-5任一项所述的方法,其特征在于,所述向第二设备发送所述第一信息,具体包括:响应于确定满足预设发送条件,所述第一设备向所述第二设备发送所述第一信息;或响应于接收到所述第二设备发送的第一指令,发送所述第一信息。
- 根据权利要求6所述的方法,其特征在于,所述第一指令,包括:用于指示所述第一设备更改配置的第一指令;或,用于指示所述第一设备配置的第一指令;或,用于查询所述第一信息的第一指令。
- 根据权利要求1-7任一项所述的方法,其特征在于,所述第一信息用于所述第二设备对所述第一设备进行算法管理,具体包括:所述第一信息用于所述第二设备对所述第一设备进行算法的生命周期管理(lifecycle management,LCM)。
- 一种算法管理方法,应用于第一设备,其特征在于,所述包括:接收第二设备发送的第二信息;其中,所述第二信息包括所述第二设备获取的第一设备的算法的监控信息;根据所述第二信息,管理所述第一设备的算法。
- 根据权利要求9所述的方法,其特征在于,所述方法还包括:获取第一算法的所述第一能耗信息和第二算法的所述第二能耗信息。
- 根据权利要求10所述的方法,其特征在于,所述根据所述第二信息,管理所述第一设备的算法,包括:根据所述第一能耗信息,第二能耗信息,所述第二信息,管理所述第一设备的算法。
- 根据权利要求9-11任一项所述的方法,其特征在于:所述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,所述第二算法为传统算法;或所述第一算法和所述第二算法为不同的人工智能/机器学习(AI/ML)算法。
- 根据权利要求9-12任一项所述的方法,其特征在于,所述算法的监控信息为算法的性能监控信息。
- 根据权利要求9-13任一项所述的方法,其特征在于,所述第二信息,还包括以下一项或 多项:用例指示;算法指示;或,通信***配置指示。
- 根据权利要求9-14任一项所述的方法,其特征在于,所述第二信息用于所述第一设备对所述第一设备进行算法管理,具体包括:所述第二信息用于所述第一设备对所述第一设备进行算法的生命周期管理(lifecycle management,LCM)。
- 一种算法管理方法,应用于第二设备,其特征在于,所述方法包括:接收第一设备发送的第一信息;其中,所述第一信息包括第一算法和第二算法的能耗对比信息;根据所述第一信息,管理所述第一设备的算法。
- 根据权利要求16所述的方法,其特征在于,所述方法还包括:获取所述第一设备的所述第一算法的第一监控信息,和/或,所述第一设备的所述第二算法的第二监控信息。
- 根据权利要求17所述的方法,其特征在于,根据所述第一信息,管理所述第一设备的算法,包括:根据所述第一信息,第一监控信息和/或第二监控信息,管理所述第一设备的算法。
- 根据权利要求16-18任一项所述的方法,其特征在于:所述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,所述第二算法为传统算法;或所述第一算法和所述第二算法为不同的人工智能/机器学习(AI/ML)算法。
- 根据权利要求16-19任一项所述的方法,其特征在于,所述第一信息,还包括以下一项或多项:用例指示;算法指示;或,通信***配置指示。
- 根据权利要求16-20任一项所述的方法,其特征在于,所述第一监控信息,所述第二监控信息为性能监控信息。
- 根据权利要求16-21任一项所述的方法,其特征在于,所述根据所述第一信息,所述第一监控信息,所述第二监控信息,管理所述第一设备的算法,包括:根据所述第一信息,所述第一监控信息,所述第二监控信息,对所述第一设备进行算法的生命周期管理(lifecycle management,LCM)。
- 根据权利要求16-22任一项所述的方法,其特征在于,所述第二设备为网络设备。
- 一种算法管理方法,应用于第二设备,其特征在于,所述方法包括:获取第一设备的第一算法的第一监控信息,和/或所述第一设备的第二算法的第二监控信息;根据所述第一监控信息,和/或所述第二监控信息,向所述第一设备发送第二信息;其中,所述第二信息用于所述第一设备对所述第一设备进行算法管理。
- 根据权利要求24所述的方法,其特征在于,根据所述第一监控信息和所述第二监控信息,向所述第一设备发送第二信息,具体包括:根据所述第一监控信息,和/或所述第二监控信息,获得第三监控信息;向所述第一设备发送包括所述第三监控信息的所述第二信息。
- 根据权利要求24-25任一项所述的方法,其特征在于:所述第一算法为人工智能/机器学习(Artificial Intelligence/Machine Learning,AI/ML)算法,所述第二算法为传统算法;或所述第一算法和所述第二算法为不同的人工智能/机器学习(AI/ML)算法。
- 根据权利要求24-26任一项所述的方法,其特征在于,所述第二信息,还包括以下一项或多项:用例指示;算法指示;或,通信***配置指示。
- 根据权利要求24-27任一项所述的方法,其特征在于,所述第一监控信息,所述第二监控信息为性能监控信息。
- 根据权利要求24-28任一项所述的方法,其特征在于,所述第三监控信息为性能监控信息。
- 根据权利要求24-29任一项所述的方法,其特征在于,所述第二信息用于所述第一设备 对所述第一设备进行算法管理,包括:所述第二信息用于所述第一设备对所述第一设备进行算法的生命周期管理(LCM)。
- 根据权利要求24-30任一项所述的方法,其特征在于,所述第二设备包括网络设备。
- 一种装置,其特征在于,包括:存储器,用于存储指令;处理器,用于调用所述存储器中的指令,执行如权利要求1-8任一所述的方法,或执行如权利要求9-15任一所述的方法,或执行如权利要求16-23任一所述的方法,或执行如权利要求24-31任一所述的方法。
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211406336.9 | 2022-11-10 | ||
CN202211406336 | 2022-11-10 | ||
CN202211711157.6 | 2022-12-29 | ||
CN202211711157.6A CN118019093A (zh) | 2022-11-10 | 2022-12-29 | 一种算法管理方法和装置 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2024099175A1 true WO2024099175A1 (zh) | 2024-05-16 |
Family
ID=90957379
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2023/128368 WO2024099175A1 (zh) | 2022-11-10 | 2023-10-31 | 一种算法管理方法和装置 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN118019093A (zh) |
WO (1) | WO2024099175A1 (zh) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006018629A (ja) * | 2004-07-02 | 2006-01-19 | Raito System Partner:Kk | 電力消費量の削減高管理システム並びに電力消費量の削減高分配方法及びそのためのプログラム |
CN110856235A (zh) * | 2018-08-20 | 2020-02-28 | 华为技术有限公司 | 信息发送、接收方法与通信设备 |
WO2020087292A1 (zh) * | 2018-10-30 | 2020-05-07 | Oppo广东移动通信有限公司 | 无线通信方法、终端设备和网络设备 |
WO2021142637A1 (zh) * | 2020-01-14 | 2021-07-22 | Oppo广东移动通信有限公司 | 人工智能操作处理方法、装置、***、终端及网络设备 |
CN113329482A (zh) * | 2021-05-21 | 2021-08-31 | Oppo广东移动通信有限公司 | 一种功耗控制方法、装置、设备及计算机存储介质 |
-
2022
- 2022-12-29 CN CN202211711157.6A patent/CN118019093A/zh active Pending
-
2023
- 2023-10-31 WO PCT/CN2023/128368 patent/WO2024099175A1/zh unknown
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006018629A (ja) * | 2004-07-02 | 2006-01-19 | Raito System Partner:Kk | 電力消費量の削減高管理システム並びに電力消費量の削減高分配方法及びそのためのプログラム |
CN110856235A (zh) * | 2018-08-20 | 2020-02-28 | 华为技术有限公司 | 信息发送、接收方法与通信设备 |
WO2020087292A1 (zh) * | 2018-10-30 | 2020-05-07 | Oppo广东移动通信有限公司 | 无线通信方法、终端设备和网络设备 |
WO2021142637A1 (zh) * | 2020-01-14 | 2021-07-22 | Oppo广东移动通信有限公司 | 人工智能操作处理方法、装置、***、终端及网络设备 |
CN113329482A (zh) * | 2021-05-21 | 2021-08-31 | Oppo广东移动通信有限公司 | 一种功耗控制方法、装置、设备及计算机存储介质 |
Also Published As
Publication number | Publication date |
---|---|
CN118019093A (zh) | 2024-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11451452B2 (en) | Model update method and apparatus, and system | |
WO2022148490A1 (zh) | Csi的上报方法及装置 | |
CN112106395A (zh) | 用户设备能力上报方法、设备及计算机可读存储介质 | |
WO2022121804A1 (zh) | 半异步联邦学习的方法和通信装置 | |
CN112868265B (zh) | 网络资源的管理方法、管理装置、电子设备及存储介质 | |
CN114071560B (zh) | 网络优化方法、装置及存储介质 | |
RU2711514C1 (ru) | Администрирование управления потоком разбитого носителя 5g | |
CN114885359A (zh) | 时延性能评估方法、接入网设备及存储介质 | |
WO2021218797A1 (zh) | 信息处理方法、装置、设备及计算机可读存储介质 | |
CN112333826B (zh) | 一种业务接纳方法及装置 | |
WO2024099175A1 (zh) | 一种算法管理方法和装置 | |
WO2023103591A1 (zh) | 一种保障射频单元运行可靠性的方法及通信装置 | |
WO2020134610A1 (zh) | 数据调度方法及装置 | |
WO2022205985A1 (zh) | Csi报告优先级确定方法及装置 | |
CN116489785B (zh) | 感知节点确定方法、装置、计算机设备和存储介质 | |
US20240089742A1 (en) | Data transmission method and related apparatus | |
CN116489703B (zh) | 感知节点确定方法、感知节点控制方法和相关设备 | |
CN111083734B (zh) | 一种数据包的压缩方法及装置 | |
US20220286810A1 (en) | Relative positioning method, apparatus, and system | |
WO2024011581A1 (zh) | 一种通信方法及装置 | |
WO2024012326A1 (zh) | 一种通信方法、装置及*** | |
WO2023208082A1 (zh) | 一种通信方法、装置、芯片及模组设备 | |
WO2024125509A1 (zh) | 模型更新方法、装置、设备及存储介质 | |
US20240049020A1 (en) | Data acquisition in a communication network | |
WO2024067143A1 (zh) | 一种信息传输方法、装置和*** |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 23887865 Country of ref document: EP Kind code of ref document: A1 |