WO2024067067A1 - 下行波束预测方法、设备、装置及存储介质 - Google Patents

下行波束预测方法、设备、装置及存储介质 Download PDF

Info

Publication number
WO2024067067A1
WO2024067067A1 PCT/CN2023/118271 CN2023118271W WO2024067067A1 WO 2024067067 A1 WO2024067067 A1 WO 2024067067A1 CN 2023118271 W CN2023118271 W CN 2023118271W WO 2024067067 A1 WO2024067067 A1 WO 2024067067A1
Authority
WO
WIPO (PCT)
Prior art keywords
configuration
pattern
information
prediction
measurement set
Prior art date
Application number
PCT/CN2023/118271
Other languages
English (en)
French (fr)
Inventor
王达
高秋彬
黄秋萍
Original Assignee
大唐移动通信设备有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 大唐移动通信设备有限公司 filed Critical 大唐移动通信设备有限公司
Publication of WO2024067067A1 publication Critical patent/WO2024067067A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W72/00Local resource management
    • H04W72/04Wireless resource allocation
    • H04W72/044Wireless resource allocation based on the type of the allocated resource
    • H04W72/046Wireless resource allocation based on the type of the allocated resource the resource being in the space domain, e.g. beams
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L5/00Arrangements affording multiple use of the transmission path
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L5/00Arrangements affording multiple use of the transmission path
    • H04L5/003Arrangements for allocating sub-channels of the transmission path
    • H04L5/0048Allocation of pilot signals, i.e. of signals known to the receiver
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W72/00Local resource management
    • H04W72/04Wireless resource allocation
    • H04W72/044Wireless resource allocation based on the type of the allocated resource
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W72/00Local resource management
    • H04W72/20Control channels or signalling for resource management
    • H04W72/23Control channels or signalling for resource management in the downlink direction of a wireless link, i.e. towards a terminal

Definitions

  • the present disclosure relates to the field of wireless communication technology, and in particular to a downlink beam prediction method, device, apparatus and storage medium.
  • the base station needs to cyclically send Tx beams in different directions, and the terminal (User Equipment, UE) uses Rx beams to receive Tx beams and measures the Channel State Information Reference Signal (CSI-RS) or Synchronization Signal Block (SSB) signals sent on all Tx beams to select the beam with the best receiving performance (such as Reference Signal Received Power (RSRP)).
  • CSI-RS Channel State Information Reference Signal
  • SSB Synchronization Signal Block
  • the base station will then use the downlink beam to transmit information to the terminal.
  • RSRP Reference Signal Received Power
  • the CSI-RS or SSB signal needs to be sent on all Tx beams, which consumes a lot of resources.
  • the terminal needs to measure the CSI-RS or SSB signals sent on all Tx beams respectively, so the terminal implementation is relatively complex and the measurement overhead is relatively large.
  • the embodiments of the present disclosure provide a downlink beam prediction method, device, apparatus and storage medium.
  • the present disclosure provides a downlink beam prediction method, which is applied to a first communication Equipment, including:
  • the first information includes one or more of a data set, a reference signal RS configuration, and beam description information;
  • one or more target downlink beams are predicted, and the target downlink beams are used for information transmission between the first communication device and the second communication device.
  • the receiving the first information includes:
  • the data set includes one or more of the following:
  • One or more data set samples wherein the data set samples include multiple beam identifiers, RS measurement results of the beams corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to a measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the method further includes:
  • an embodiment of the present disclosure further provides a downlink beam prediction method, which is applied to a second communication device, including:
  • the first information including one or more of a data set, an RS configuration, and a beam description information; the first information is used to train or update an artificial intelligence or machine learning AI/ML model, the AI/ML model is used to predict one or more target downlink beams, the target downlink The beam is used for information transmission between the second communication device and the first communication device;
  • the first information is sent to the first communication device.
  • the data set includes one or more of the following:
  • One or more data set samples wherein the data set samples include multiple beam identifiers, a reference signal measurement result of a beam corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to a measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the method further comprises:
  • the second information includes one or more of the following:
  • an embodiment of the present disclosure further provides a downlink beam prediction method, which is applied to a terminal, including:
  • one or more target downlink beams are predicted, and the target downlink beams are used for information transmission between the network device and the terminal.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to a prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate a first RS configuration belonging to a measurement set.
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the method further includes:
  • an embodiment of the present disclosure further provides a downlink beam prediction method, which is applied to a network device, including:
  • RS configuration is used to train or update an artificial intelligence or machine learning AI/ML model, where the AI/ML model is used to predict one or more target downlink beams, where the target downlink beams are used for information transmission between the network device and the terminal;
  • the RS configuration is sent to the terminal.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to a prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the method further comprises:
  • the second information includes one or more of the following:
  • the present disclosure also provides a first communication device, including a memory, a transceiver Machine, processor;
  • a memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations:
  • the first information includes one or more of a data set, a reference signal RS configuration, and beam description information;
  • one or more target downlink beams are predicted, and the target downlink beams are used for information transmission between the first communication device and the second communication device.
  • the receiving the first information includes:
  • the data set includes one or more of the following:
  • One or more data set samples wherein the data set samples include multiple beam identifiers, RS measurement results of the beams corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to a measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the operation further includes:
  • an embodiment of the present disclosure further provides a second communication device, including a memory, a transceiver, and a processor;
  • a memory for storing a computer program; a transceiver for receiving a computer program under the control of the processor;
  • a processor is used to read the computer program in the memory and perform the following operations:
  • the first information includes one or more of a data set, an RS configuration, and a beam description information; the first information is used to train or update an artificial intelligence or machine learning AI/ML model, where the AI/ML model is used to predict one or more target downlink beams, and the target downlink beam is used for information transmission between the second communication device and the first communication device;
  • the first information is sent to the first communication device.
  • the data set includes one or more of the following:
  • One or more data set samples wherein the data set samples include multiple beam identifiers, a reference signal measurement result of a beam corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to a measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the operation further includes:
  • the second information includes one or more of the following:
  • an embodiment of the present disclosure further provides a terminal, including a memory, a transceiver, and a processor;
  • a memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations:
  • one or more target downlink beams are predicted, and the target downlink beams are used for information transmission between the network device and the terminal.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to a prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the operation further includes:
  • an embodiment of the present disclosure further provides a network device, including a memory, a transceiver, and a processor;
  • a memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations:
  • RS configuration is used to train or update an artificial intelligence or machine learning AI/ML model, where the AI/ML model is used to predict one or more target downlink beams, where the target downlink beams are used for information transmission between the network device and the terminal;
  • the RS configuration is sent to the terminal.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to a prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the operation further includes:
  • the second information includes one or more of the following:
  • an embodiment of the present disclosure further provides a downlink beam prediction device, which is applied to a first communication device, including:
  • a first receiving unit configured to receive first information, where the first information includes one or more of a data set, a reference signal RS configuration, and a beam description information;
  • a first model unit configured to train or update an artificial intelligence or machine learning AI/ML model for downlink beam prediction based on the first information
  • the first prediction unit is used to predict one or more target downlink beams based on the trained or updated AI/ML model, and the target downlink beams are used for information transmission between the first communication device and the second communication device.
  • the receiving the first information includes:
  • the data set includes one or more of the following:
  • One or more data set samples wherein the data set samples include multiple beam identifiers, RS measurement results of the beams corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to a measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the device further includes:
  • the first sending unit is configured to send second information to the second communication device, where the second information includes one or more of the following:
  • an embodiment of the present disclosure further provides a downlink beam prediction device, which is applied to a second communication device, including:
  • a first determination unit is used to determine first information, where the first information includes one or more of a data set, an RS configuration, and a beam description information; the first information is used to train or update an artificial intelligence or machine learning AI/ML model, where the AI/ML model is used to predict one or more target downlink beams, and the target downlink beam is used for information transmission between the second communication device and the first communication device;
  • the second sending unit is configured to send the first information to the first communication device.
  • the data set includes one or more of the following:
  • One or more data set samples wherein the data set samples include multiple beam identifiers, a reference signal measurement result of a beam corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to a measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the device further comprises:
  • a second receiving unit configured to receive second information sent by the first communication device
  • the second information includes one or more of the following:
  • the number of input beams and the number of output beams of the AI/ML model used to predict the target downlink beam Quantity information are the number of input beams and the number of output beams of the AI/ML model used to predict the target downlink beam Quantity information.
  • an embodiment of the present disclosure further provides a downlink beam prediction device, applied to a terminal, including:
  • a third receiving unit configured to receive a reference signal RS configuration sent by a network device
  • a second model unit configured to train or update an artificial intelligence or machine learning AI/ML model for downlink beam prediction based on the RS configuration
  • the second prediction unit is used to predict one or more target downlink beams based on the trained or updated AI/ML model, and the target downlink beams are used for information transmission between the network device and the terminal.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to a prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate a beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the device further includes:
  • the third sending unit is configured to send second information to the network device, where the second information includes one or more of the following:
  • an embodiment of the present disclosure further provides a downlink beam prediction device, which is applied to a network device, including:
  • a second determination unit configured to determine a reference signal RS configuration, wherein the RS configuration is used to train or update an artificial intelligence or machine learning AI/ML model, wherein the AI/ML model is used to predict one or more target downlink beams, and the target downlink beam is used for information transmission between the network device and the terminal;
  • the fourth sending unit is used to send the RS configuration to the terminal.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to a prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern the third pattern being used to indicate an RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • a fifth pattern is used to indicate the wavelet satisfied by the RS configuration belonging to the measurement set. Bundle number arrangement rule.
  • the device further comprises:
  • a fourth receiving unit configured to receive second information sent by the terminal
  • the second information includes one or more of the following:
  • the embodiment of the present disclosure also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is used to enable a computer to execute the downlink beam prediction method described in the first aspect as described above, or execute the downlink beam prediction method described in the second aspect as described above, or execute the downlink beam prediction method described in the third aspect as described above, or execute the downlink beam prediction method described in the fourth aspect as described above.
  • the embodiment of the present disclosure further provides a communication device, in which a computer program is stored, and the computer program is used to enable the communication device to execute the downlink beam prediction method described in the first aspect as described above, or execute the downlink beam prediction method described in the second aspect as described above, or execute the downlink beam prediction method described in the third aspect as described above, or execute the downlink beam prediction method described in the fourth aspect as described above.
  • the embodiment of the present disclosure also provides a processor-readable storage medium, wherein the processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the downlink beam prediction method described in the first aspect as described above, or execute the downlink beam prediction method described in the second aspect as described above, or execute the downlink beam prediction method described in the third aspect as described above, or execute the downlink beam prediction method described in the fourth aspect as described above.
  • the present disclosure also provides a chip product, wherein the chip product stores There is a computer program, which is used to enable the chip product to execute the downlink beam prediction method described in the first aspect as described above, or execute the downlink beam prediction method described in the second aspect as described above, or execute the downlink beam prediction method described in the third aspect as described above, or execute the downlink beam prediction method described in the fourth aspect as described above.
  • the downlink beam prediction method, device, apparatus and storage medium provided by the embodiments of the present disclosure, by sending one or more of a data set, RS configuration, and beam description information to a first communication device, the first communication device can train or update the AI/ML model used for downlink beam prediction based on this information, and predict one or more target downlink beams based on the trained or updated AI/ML model, thereby effectively saving RS transmission resources, UE measurement overhead and reducing UE measurement delay.
  • FIG1 is a schematic diagram of a flow chart of a downlink beam prediction method provided by an embodiment of the present disclosure
  • FIG2 is a second flow chart of a downlink beam prediction method provided by an embodiment of the present disclosure.
  • FIG3 is a third flow chart of a downlink beam prediction method provided by an embodiment of the present disclosure.
  • FIG4 is a fourth flow chart of a downlink beam prediction method provided in an embodiment of the present disclosure.
  • FIG5 is a schematic diagram of the structure of a first communication device provided in an embodiment of the present disclosure.
  • FIG6 is a schematic diagram of the structure of a second communication device provided in an embodiment of the present disclosure.
  • FIG7 is a schematic diagram of the structure of a terminal provided in an embodiment of the present disclosure.
  • FIG8 is a schematic diagram of the structure of a network device provided in an embodiment of the present disclosure.
  • FIG9 is a schematic diagram of a structure of a downlink beam prediction device provided in an embodiment of the present disclosure.
  • FIG10 is a second schematic diagram of the structure of a downlink beam prediction device provided in an embodiment of the present disclosure.
  • FIG11 is a third structural diagram of a downlink beam prediction device provided in an embodiment of the present disclosure.
  • FIG. 12 is a fourth schematic diagram of the structure of the downlink beam prediction device provided in an embodiment of the present disclosure.
  • the term "and/or” describes the association relationship of associated objects, indicating that three relationships may exist.
  • a and/or B may represent three situations: A exists alone, A and B exist at the same time, and B exists alone.
  • the character "/" generally indicates that the associated objects before and after are in an "or” relationship.
  • plurality in the embodiments of the present disclosure refers to two or more than two, and other quantifiers are similar thereto.
  • One method is to measure the RSRP of all downlink beam pairs (DL beam pairs) (a beam pair consisting of a Tx beam of the base station and an Rx beam of the UE), and the beam pair corresponding to the largest RSRP is the optimal beam pair.
  • the base station uses the Tx beam in the beam pair to send information to the UE, and the UE uses the Rx beam in the beam pair to receive information.
  • the UE fixes or selects the best DL Rx beam, receives and measures the received power RSRP of all DL Tx beams sent by the base station, and the Tx beam corresponding to the largest RSRP is the optimal Tx beam, and informs the base station.
  • the base station subsequently uses this Tx beam to send information to the UE.
  • Another method is that the base station fixes or selects the best DL Tx beam, and the UE uses all DL Rx beams to receive and measure the received power RSRP of the DL Tx beam sent by the base station.
  • the Rx beam corresponding to the maximum RSRP is the optimal Rx beam.
  • the base station subsequently uses the Tx beam to send information to the UE, the UE uses the optimal Rx beam to receive.
  • the UE needs to use all Rx beams to receive the CSI-RS/SSB of each Tx beam sent by the base station for measurement.
  • the UE fixes or selects the best Rx beam and measures all Tx beams sent by the base station. It needs to measure 32 Tx beams to calculate the best Tx beam. The same problem also exists for the DL Rx beam measurement method.
  • the present disclosure proposes to use artificial intelligence (AI) or machine learning (AI) technology to predict the downlink beam.
  • AI artificial intelligence
  • AI machine learning
  • the present disclosure can also be used in other scenarios, such as measuring other wide beams that are easy to measure for sending SSB, or beams in other frequency bands, so as to predict the narrow beam that sends CSI-RS, or predict the best beam in the high-frequency beam; for example, using the previously measured beam measurement results, predict the beam (pair) with the best receiving performance in the Tx beam or beam pair sent by the base station in the future; for example, in DL Rx beam prediction, the base station fixes or selects the best Tx beam, continuously sends the reference signal (RS), and the UE uses different DL Rx beams to receive and measure RS, and predicts the best Rx beam through the measurement results.
  • This can save RS transmission resources, UE measurement overhead, and reduce UE measurement delay.
  • FIG. 1 is a flow chart of a downlink beam prediction method provided in an embodiment of the present disclosure. As shown in FIG. 1 , the method is applied to a first communication device, including:
  • Step 100 Receive first information, where the first information includes one or more of a data set, a reference signal RS configuration, and beam description information.
  • Step 101 Based on the first information, train or update the artificial intelligence or machine learning AI/ML model for downlink beam prediction.
  • Step 102 Based on the trained or updated AI/ML model, predict one or more target A downlink beam, a target downlink beam is used for information transmission between a first communication device and a second communication device.
  • the first communication device may be a terminal or a network device (eg, a base station), and correspondingly, the second communication device may be a network device or a terminal that performs information transmission with the first communication device.
  • a network device eg, a base station
  • the second communication device may be a network device or a terminal that performs information transmission with the first communication device.
  • the second communication device may be a network device.
  • the terminal may receive the first information (sent by the network device or a third-party device), and based on the received first information, the terminal may train or update the AI/ML model for downlink beam prediction, thereby predicting one or more target downlink beams, and the predicted target downlink beams are used for information transmission between the terminal and the network device.
  • the second communication device may be a terminal.
  • the network device may receive the first information (sent by the terminal or a third-party device), and based on the received first information, the network device may train or update the AI/ML model for downlink beam prediction, thereby predicting one or more target downlink beams, and the predicted target downlink beams are used for information transmission between the terminal and the network device.
  • the second communication device can send a first information to the first communication device, and the first information may include one or more of a data set, a reference signal RS configuration, and a beam description information.
  • the first communication device After the first communication device receives the first information, it can train or update the AI/ML model for downlink beam prediction according to the first information.
  • updating the model includes fine-tuning the model, that is, after the first communication device receives the first information, it can fine-tune the AI/ML model for downlink beam prediction according to the first information.
  • the functions of the AI/ML model for downlink beam prediction may include: spatial domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, and so on.
  • Spatial domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the optimal beam among a large number of beams;
  • frequency domain prediction refers to measuring the beam on frequency 1 and predicting the optimal beam among another beam on frequency 2;
  • time domain prediction refers to measuring the beam at the present moment and predicting the optimal beam at the future moment.
  • the RS measurement result can be any measurement result that can represent the downlink beam reception performance, such as RSRP, and the specific form is not limited.
  • the first communication device can predict one or more target downlink beams based on the RS measurement results through the trained or updated model.
  • the predicted target downlink beam can be a Tx beam, or an Rx beam, or a beam pair, which is used for information transmission between the first communication device and the second communication device.
  • the first information received by the first communication device may be predefined by the second communication device or a third-party device.
  • the third-party device when the third-party device predefines the first information, in one implementation, can send the first information to the first communication device and the second communication device; or, the third-party device can send the first information to the second communication device, and the second communication device sends the first information to the first communication device after receiving the first information.
  • the first information may be transmitted via a control channel or a data channel of the Uu interface, or may be transmitted via other interfaces or other channels, and the specific circumstances are not limited.
  • the downlink beam prediction method provided by the embodiment of the present disclosure sends one or more of a data set, an RS configuration, and a beam description information to a first communication device.
  • the first communication device can train or update the AI/ML model used for downlink beam prediction based on this information, and predict one or more target downlink beams based on the trained or updated AI/ML model, thereby effectively saving RS transmission resources, UE measurement overhead, and reducing UE measurement delay.
  • receiving the first information includes:
  • the first information sent by the second communication device is received.
  • the first information received by the terminal may be sent by the network device to the terminal.
  • the network device predefines and sends the first information to the terminal.
  • the network device predefines an RS configuration and sends the RS configuration to the terminal, and the terminal trains or updates the AI/ML model for downlink beam prediction according to the RS configuration sent by the network device.
  • the first information received by the network device may be sent by the terminal.
  • the first information sent by the terminal may include a data set and/or beam description information, but does not include RS configuration.
  • the terminal predefines the first information and sends the first information to the network device.
  • the terminal predefines a data set and sends the data set to the network device.
  • the network equipment trains or updates the AI/ML model used for downlink beam prediction based on the data set sent by the terminal.
  • the dataset contains one or more of the following:
  • One or more data set samples each data set sample including multiple beam identifiers, RS measurement results of the beam corresponding to each beam identifier, and one or more beam identifiers used as prediction labels.
  • the data set may be one or more data sets, and the data set identifiers represent different data sets, and different data sets correspond to different data set identifiers.
  • the data set sample includes multiple beam IDs, the RS measurement results of the beams corresponding to these beam IDs, and one or more beam IDs as prediction tags. It is understandable that in the process of pre-defining the data set by the terminal or network device or third-party device, each downlink beam corresponds to a beam ID.
  • a downlink beam can be a Tx beam, an Rx beam, or a beam pair.
  • the first communication device may use the RS measurement results of the downlink beams corresponding to multiple beam IDs in the data set samples as the input of the model; and use one or more beam IDs in the data set samples as prediction labels as the output labels of the model.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the RS configuration may be one or more, and the RS configuration identifier represents different RS configurations, and different RS configurations correspond to different RS configuration identifiers.
  • the RS configuration may include the type of RS (e.g., CSI-RS, SSB, Phase-tracking Reference Signal (PT-RS), Cell Reference Signal (CRS), Demodulation Reference Signal (DRS), etc. Reference Signal, DMRS), RS transmission resources and RS identification (RS ID), etc.
  • RS ID is RS indicator (RS Indication), for example, the indicator of CSI-RS is CRI, and the indicator of SSB is SSBRI.
  • the measurement set refers to the RS configuration or beam in the set, which is used to obtain RS measurement results as input to the model;
  • the prediction set refers to the RS configuration or beam in the set, which is used to determine one or more downlink beams as prediction labels.
  • the first communication device measures the RS sent on the beam in the measurement set, obtains the RS measurement result, and uses the RS measurement result as the input of the model; the first communication device measures the RS sent on the beam in the prediction set, calculates one or more best beams using a non-AI/ML method, and uses the one or more best beams as the output labels of the model.
  • the first communication device Since the first communication device has different behaviors after receiving the first RS configuration belonging to the measurement set and the second RS configuration belonging to the prediction set, it is necessary to distinguish which set different RS configurations belong to.
  • the prediction set When the measurement set is a subset of the prediction set, only the prediction set may be RS configured, and the RS configuration may include a second RS configuration belonging to the prediction set and a measurement set pattern, and the measurement set pattern is used to indicate the first RS configuration belonging to the measurement set in the prediction set.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set.
  • the first pattern may indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the first pattern may be a string or a string equal to the number of RS resource configurations.
  • the second pattern may indicate an RS identifier corresponding to the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the second pattern is: ⁇ 2, 4, 6, 8 ⁇ , which may indicate that the 2nd, 4th, 6th, and 8th RSs in the prediction set are the measurement set.
  • the third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the third pattern may indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the third pattern may be configured as: RSs with even numbers in the prediction set are the measurement set.
  • Different rules may be represented by a character string, for example, 2 bits may represent 4 rules.
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set.
  • the fourth pattern can indicate the beam number satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the fourth pattern is: ⁇ 2, 4, 6, 8 ⁇ , which may indicate that the 2nd, 4th, 6th, and 8th downlink beams in the prediction set are the measurement set.
  • the fifth pattern is used to indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the fifth pattern can indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the fifth pattern is configured as follows: downlink beams with even numbers in the prediction set are used as the measurement set.
  • Different rules can be represented by character strings. For example, 2 bits can represent 4 rules.
  • a measurement set pattern indication method may be used during RS configuration, thereby saving bit overhead for configuration.
  • the beam description information includes one or more of the following:
  • the first communication device may also train or update the AI/ML model for downlink beam prediction based on the beam description information
  • the beam description information may include one or more of the beam ID, RS ID, antenna configuration information of the network device, beam angle information, and beam width information.
  • the beam description information may be used as auxiliary information together with the data set and/or RS configuration to train or update the AI/ML model for downlink beam prediction, as well as subsequent prediction processes.
  • a beam description information includes description information of one or more downlink beams.
  • a downlink beam can be a Tx beam, or an Rx beam, or a beam pair.
  • the multiple downlink beams can be described separately in a list manner.
  • each downlink beam corresponds to a beam ID.
  • RS ID is an indicator of the RS sent on the downlink beam, for example, the indicator of CSI-RS is CRI, and the indicator of SSB is SSBRI.
  • the antenna configuration information of the network device of the downlink beam, the angle information of the downlink beam, and the width information of the downlink beam can also be used to describe the downlink beam.
  • the method further includes:
  • the second information is sent to the second communication device.
  • the first communication device may send second information to the second communication device, where the second information is used to inform the second communication device of the results of the one or more target downlink beams predicted by the first communication device.
  • the second communication device After the second communication device receives the second information sent by the first communication device, it can learn the one or more target downlink beams predicted by the first communication device based on the first information and the second information. Then, the first communication device and the second communication device can transmit information on the target downlink beam (i.e., best beam) to improve the transmission performance of the system.
  • the target downlink beam i.e., best beam
  • the second information includes one or more of the following:
  • the second information may include the beam ID corresponding to the target downlink beam, so that the second communication device can learn the target downlink beam inferred by the first communication device through the AI/ML model based on the beam ID corresponding to the target downlink beam.
  • the mapping relationship between the beam ID and the physical beam is known to the second communication device, that is, the beam ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the first communication device using the first information can be recognized by the second communication device and correspond to the actual physical target downlink beam, so that the first communication device and the second communication device can subsequently communicate using the target downlink beam.
  • the first communication device only infers one target downlink beam through the AI/ML model, and the first communication device can send the beam ID corresponding to the target downlink beam to the second communication device.
  • the first communication device infers multiple target downlink beams through an AI/ML model, and the first communication device can select a beam ID corresponding to one of the target downlink beams and send it to the second communication device; or, the first communication device can send beam IDs corresponding to multiple target downlink beams to the second communication device, and the second communication device selects one of them.
  • the second information may include the RS ID corresponding to the target downlink beam, so that the second communication device can learn the target downlink beam inferred by the first communication device through the AI/ML model based on the RS ID corresponding to the target downlink beam.
  • the mapping relationship between the RS ID and the physical beam is known to the second communication device, that is, The RS ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the first communication device using the first information can be recognized by the second communication device and correspond to the actual physical target downlink beam, so that the first communication device and the second communication device can subsequently communicate using the target downlink beam.
  • the first communication device only infers one target downlink beam through the AI/ML model, and the first communication device can send the RS ID corresponding to the target downlink beam to the second communication device.
  • the first communication device infers multiple target downlink beams through an AI/ML model, and the first communication device can select an RS ID corresponding to one of the target downlink beams and send it to the second communication device; or, the first communication device can send RS IDs corresponding to multiple target downlink beams to the second communication device, and the second communication device selects one of them.
  • each model can correspond to a model identifier (model ID).
  • model ID the first communication device not only needs to send the beam ID or RS ID corresponding to the predicted target downlink beam, but also needs to send the model ID corresponding to the predicted target downlink beam.
  • the second communication device can know which AI/ML model the first communication device uses to infer the target downlink beam based on the model identifier.
  • the second information may include an identifier of the data set used to predict the target downlink beam, so that the second communication device can know, based on the identifier of the data set used to predict the target downlink beam, the AI/ML model used by the first communication device to infer the target downlink beam.
  • the first communication device may also send the identifier of the data set, representing the above-mentioned model ID.
  • the second communication device can know the mapping relationship between the beam ID output by the first communication device using the model and the physical beam through the identifier of the data set.
  • the second information may include an identifier of the RS configuration for predicting the target downlink beam, so that the second communication device can know, based on the identifier of the RS configuration for predicting the target downlink beam, through which AI/ML model the first communication device inferred the target downlink beam.
  • the first communication device may also send an RS configuration identifier representing the above-mentioned model ID.
  • the second communication device can know the mapping relationship between the RS ID output by the first communication device using the model and the physical beam through the RS configuration identifier.
  • the second information may include an identifier of the beam description information used to predict the target downlink beam, so that the second communication device can know which AI/ML model the first communication device uses to infer the target downlink beam based on the identifier of the beam description information used to predict the target downlink beam.
  • the first communication device may also send an identifier of the beam description information, representing the above-mentioned model ID.
  • the second communication device can know the mapping relationship between the beam ID output by the model used by the first communication device or the RS ID and the physical beam through the identifier configured by the RS.
  • the second information may include the input beam number and output beam number information of the AI/ML model used to predict the target downlink beam, so that the second communication device can know which AI/ML model the first communication device uses to infer the target downlink beam based on the input beam number and output beam number information used to predict the target downlink beam.
  • the number of input beams can be the number of input Tx beams of the model, or the number of input Rx beams of the model, or the number of input beam pairs of the model, etc.
  • the number of output beams can be the number of output Tx beams of the model, or the number of output Rx beams of the model, or the number of output beam pairs of the model, etc.
  • the first communication device trains or updates an AI/ML model for each ⁇ input beam number, output beam number ⁇ pair, then the ⁇ input beam number, output beam number ⁇ pair
  • the information can represent the model. That is, the first communication device sends a pair of ⁇ input beam number, output beam number ⁇ information that corresponds to the model ID.
  • the second communication device can know the mapping relationship between the output beam ID or RS ID of the model used by the first communication device and the physical beam through the pair of ⁇ input beam number, output beam number ⁇ information.
  • the first information may be associated with one or more of an application scenario, a network device configuration, and a model function.
  • different data sets, different RS configurations, and different beam description information can be targeted at different application scenarios, or base station configurations, or AI functions, respectively.
  • Application scenarios include: urban, rural, indoor, outdoor, highway, high-speed rail, urban macro cell (Uma), urban micro cell (Umi), etc.;
  • base station configuration includes: base station antenna configuration, beam configuration, reference signal configuration, etc.;
  • AI functions mainly include air domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc., among which, air domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the best beam among a large number of beams; frequency domain prediction refers to measuring the beam on frequency 1, and predicting the best beam among another beam on frequency 2; time domain prediction refers to measuring the beam at the present moment, and predicting the best beam at the future moment.
  • an AI/ML model for downlink beam prediction used in different scenarios can be obtained.
  • FIG. 2 is a second flow chart of a downlink beam prediction method provided by an embodiment of the present disclosure. As shown in FIG. 2 , the method is applied to a second communication device, including:
  • Step 200 determine the first information, where the first information includes one or more of a data set, an RS configuration, and a beam description information; the first information is used to train or update an artificial intelligence or machine learning AI/ML model, and the AI/ML model is used to predict one or more target downlink beams, and the target downlink beam is used for information transmission between the second communication device and the first communication device.
  • the first information includes one or more of a data set, an RS configuration, and a beam description information
  • the first information is used to train or update an artificial intelligence or machine learning AI/ML model, and the AI/ML model is used to predict one or more target downlink beams, and the target downlink beam is used for information transmission between the second communication device and the first communication device.
  • Step 201 Send first information to a first communication device.
  • the first communication device may be a terminal or a network device (eg, a base station), and correspondingly, the second communication device may be a network device or a terminal that performs information transmission with the first communication device.
  • a network device eg, a base station
  • the second communication device may be a network device or a terminal that performs information transmission with the first communication device.
  • the second communication device may be a network device.
  • the terminal can receive first information (sent by a network device or a third-party device). Based on the received first information, the terminal can train or update the AI/ML model used for downlink beam prediction, thereby predicting one or more target downlink beams. The predicted target downlink beam is used for information transmission between the terminal and the network device.
  • the second communication device is a terminal.
  • the network device can receive the first information (sent by the terminal or a third-party device), and based on the received first information, the network device can train or update the AI/ML model for downlink beam prediction, thereby predicting one or more target downlink beams, and the predicted target downlink beams are used for information transmission between the terminal and the network device.
  • the second communication device can send first information to the first communication device, and the first information may include one or more of a data set, a reference signal RS configuration, and a beam description information.
  • the first information determined by the second communication device may be pre-defined by the second communication device itself; or may be pre-defined by a third-party device and sent to the second communication device.
  • the second communication device After the second communication device determines the first information, it can send the first information to the first communication device. After the first communication device receives the first information, it can train or update the AI/ML model for downlink beam prediction according to the first information. Wherein, updating the model includes fine-tuning the model, that is, after the first communication device receives the first information, it can fine-tune the AI/ML model for downlink beam prediction according to the first information.
  • the functions of the AI/ML model for downlink beam prediction may include: spatial domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • Spatial domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the optimal beam among a large number of beams; frequency domain prediction refers to measuring the beam on frequency 1 and predicting the optimal beam among another beam on frequency 2; time domain prediction refers to measuring the beam at the present moment and predicting the optimal beam at the future moment.
  • the RS measurement result can be any measurement result that can represent the downlink beam reception performance, such as RSRP, and the specific form is not limited.
  • the first communication device can communicate with the RS based on the RS measurement result.
  • the trained or updated model predicts one or more target downlink beams, and the predicted target downlink beam may be a Tx beam, or an Rx beam, or a beam pair, which is used for information transmission between the first communication device and the second communication device.
  • the first information may be transmitted via a control channel or a data channel of the Uu interface, or may be transmitted via other interfaces or other channels, and the specific circumstances are not limited.
  • the downlink beam prediction method provided by the embodiment of the present disclosure sends one or more of a data set, an RS configuration, and a beam description information to a first communication device.
  • the first communication device can train or update the AI/ML model used for downlink beam prediction based on this information, and predict one or more target downlink beams based on the trained or updated AI/ML model, thereby effectively saving RS transmission resources, UE measurement overhead, and reducing UE measurement delay.
  • the dataset contains one or more of the following:
  • Each data set sample including multiple beam identifiers, a reference signal measurement result of a beam corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the data set may be one or more data sets, and the data set identifiers represent different data sets, and different data sets correspond to different data set identifiers.
  • the data set sample includes multiple beam IDs, the RS measurement results of the beams corresponding to these beam IDs, and one or more beam IDs as predicted labels. It is understandable that in the process of pre-defining the data set by the terminal or network device or third-party device, each downlink beam corresponds to a beam ID.
  • a downlink beam can be a Tx beam, an Rx beam, or a beam pair.
  • the first communication device may use the RS measurement results of the downlink beams corresponding to multiple beam IDs in the data set samples as the input of the model; and use one or more beam IDs in the data set samples as prediction labels as the output labels of the model.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the RS configuration may be one or more, and the RS configuration identifier represents different RS configurations, and different RS configurations correspond to different RS configuration identifiers.
  • the RS configuration may include the type of RS (e.g., CSI-RS, SSB, PT-RS, CRS, DMRS, etc.), RS transmission resources, and RS ID, etc.
  • the RS ID is an RS indicator, for example, the indicator of CSI-RS is CRI, and the indicator of SSB is SSBRI.
  • the measurement set refers to the RS configuration or beam in the set, which is used to obtain RS measurement results as input to the model;
  • the prediction set refers to the RS configuration or beam in the set, which is used to determine one or more downlink beams as prediction labels.
  • the first communication device measures the RS sent on the beam in the measurement set, obtains the RS measurement result, and uses the RS measurement result as the input of the model; the first communication device measures the RS sent on the beam in the prediction set, calculates one or more best beams using a non-AI/ML method, and uses the one or more best beams as the output labels of the model.
  • the first communication device Since the first communication device has different behaviors after receiving the first RS configuration belonging to the measurement set and the second RS configuration belonging to the prediction set, it is necessary to distinguish which set different RS configurations belong to.
  • the prediction set When the measurement set is a subset of the prediction set, only the prediction set may be RS configured, and the RS configuration may include a second RS configuration belonging to the prediction set and a measurement set pattern, and the measurement set pattern is used to indicate the first RS configuration belonging to the measurement set in the prediction set.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set.
  • the first pattern may be used to indicate whether each second RS configuration belonging to the prediction set is Belongs to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the second pattern may indicate an RS identifier corresponding to the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the second pattern is: ⁇ 2, 4, 6, 8 ⁇ , which may indicate that the 2nd, 4th, 6th, and 8th RSs in the prediction set are the measurement set.
  • the third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the third pattern may indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the third pattern may be configured as: RSs with even numbers in the prediction set are the measurement set.
  • Different rules may be represented by a character string, for example, 2 bits may represent 4 rules.
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set.
  • the fourth pattern can indicate the beam number satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the fourth pattern is: ⁇ 2, 4, 6, 8 ⁇ , which can indicate that the 2nd, 4th, 6th, and 8th downlink beams in the prediction set are the measurement set.
  • the fifth pattern is used to indicate the wavelet satisfied by the RS configuration belonging to the measurement set. Bundle number arrangement rule.
  • the fifth pattern can indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the fifth pattern is configured as follows: downlink beams with even numbers in the prediction set are used as the measurement set.
  • Different rules can be represented by a character string, for example, 2 bits can represent 4 rules.
  • a measurement set pattern indication method may be used during RS configuration, thereby saving bit overhead for configuration.
  • the beam description information includes one or more of the following:
  • the first communication device may also train or update the AI/ML model for downlink beam prediction based on the beam description information
  • the beam description information may include one or more of the beam ID, RS ID, antenna configuration information of the network device, beam angle information, and beam width information.
  • the beam description information may be used as auxiliary information together with the data set and/or RS configuration to train or update the AI/ML model for downlink beam prediction, as well as subsequent prediction processes.
  • a beam description information includes description information of one or more downlink beams.
  • a downlink beam can be a Tx beam, or an Rx beam, or a beam pair.
  • the multiple downlink beams can be described separately in a list manner.
  • each downlink beam corresponds to a beam ID.
  • RS ID is an indicator of the RS sent on the downlink beam, for example, the indicator of CSI-RS is CRI, and the indicator of SSB is SSBRI.
  • the antenna configuration information of the network device of the downlink beam, the angle information of the downlink beam and the width information of the downlink beam can also be used to describe the downlink beam.
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • different data sets, different RS configurations, and different beam description information can target different application scenarios, or base station configurations, or AI functions, respectively.
  • Application scenarios include: urban, rural, indoor, outdoor, highway, high-speed rail, Uma, Umi, etc.;
  • base station configuration includes: base station antenna configuration, beam configuration, reference signal configuration, etc.;
  • AI functions mainly include air domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc., among which, air domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the best beam among a large number of beams;
  • frequency domain prediction refers to measuring the beam on frequency 1, and predicting the best beam among another beam on frequency 2;
  • time domain prediction refers to measuring the beam at the present moment and predicting the best beam at the future moment.
  • an AI/ML model for downlink beam prediction used in different scenarios can be obtained.
  • the method further comprises:
  • a target downlink beam for information transmission with the first communication device is determined.
  • the first communication device may send second information to the second communication device, where the second information is used to inform the second communication device of the results of the one or more target downlink beams predicted by the first communication device.
  • the second communication device After the second communication device receives the second information sent by the first communication device, it can learn the one or more target downlink beams predicted by the first communication device based on the first information and the second information. Then, the first communication device and the second communication device can transmit information on the target downlink beam (i.e., best beam) to improve the transmission performance of the system.
  • the target downlink beam i.e., best beam
  • the second information includes one or more of the following:
  • the second information may include the beam ID corresponding to the target downlink beam, so that the second communication device can know the beam ID corresponding to the target downlink beam that the first communication device has inferred through the AI/ML model. Measured target downlink beam.
  • the first communication device only infers one target downlink beam through the AI/ML model, and the first communication device can send the beam ID corresponding to the target downlink beam to the second communication device.
  • the mapping relationship between the beam ID and the physical beam is known to the second communication device, that is, the beam ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the first communication device using the first information can be recognized by the second communication device and correspond to the actual physical target downlink beam, so that the first communication device and the second communication device can subsequently communicate using the target downlink beam.
  • the first communication device infers multiple target downlink beams through an AI/ML model, and the first communication device can select a beam ID corresponding to one of the target downlink beams and send it to the second communication device; or, the first communication device can send beam IDs corresponding to multiple target downlink beams to the second communication device, and the second communication device selects one of them.
  • the second information may include the RS ID corresponding to the target downlink beam, so that the second communication device can learn the target downlink beam inferred by the first communication device through the AI/ML model based on the RS ID corresponding to the target downlink beam.
  • the mapping relationship between the RS ID and the physical beam is known to the second communication device. That is, the RS ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the first communication device using the first information can be recognized by the second communication device and correspond to the actual physical target downlink beam, so that the first communication device and the second communication device can subsequently communicate using the target downlink beam.
  • the first communication device only infers one target downlink beam through the AI/ML model, and the first communication device can send the RS ID corresponding to the target downlink beam to the second communication device.
  • the first communication device infers multiple target downlink waves through an AI/ML model.
  • the first communication device may select an RS ID corresponding to one of the target downlink beams and send it to the second communication device; or, the first communication device may send RS IDs corresponding to multiple target downlink beams to the second communication device, and the second communication device may select one of them.
  • each model can correspond to a model ID.
  • the first communication device not only needs to send the beam ID or RS ID corresponding to the predicted target downlink beam, but also needs to send the model ID corresponding to the predicted target downlink beam.
  • the second communication device can know which AI/ML model the first communication device uses to infer the target downlink beam based on the model identifier.
  • the second information may include an identifier of the data set used to predict the target downlink beam, so that the second communication device can know, based on the identifier of the data set used to predict the target downlink beam, the AI/ML model used by the first communication device to infer the target downlink beam.
  • the first communication device may also send the identifier of the data set, representing the above-mentioned model ID.
  • the second communication device can know the mapping relationship between the beam ID output by the first communication device using the model and the physical beam through the identifier of the data set.
  • the second information may include an identifier of the RS configuration for predicting the target downlink beam, so that the second communication device can know, based on the identifier of the RS configuration for predicting the target downlink beam, the AI/ML model through which the first communication device inferred the target downlink beam.
  • the first communication device may also send an RS configuration identifier representing the above-mentioned model ID.
  • the second communication device can know the mapping relationship between the RS ID output by the first communication device using the model and the physical beam through the RS configuration identifier.
  • the second information may include an identifier of the beam description information used to predict the target downlink beam, so that the second communication device can know which AI/ML model the first communication device uses to infer the target downlink beam based on the identifier of the beam description information used to predict the target downlink beam.
  • the first communication device may also send an identifier of the beam description information, representing the above-mentioned model ID.
  • the second communication device can know the mapping relationship between the beam ID output by the model used by the first communication device or the RS ID and the physical beam through the identifier configured by the RS.
  • the second information may include the input beam number and output beam number information of the AI/ML model used to predict the target downlink beam, so that the second communication device can know which AI/ML model the first communication device uses to infer the target downlink beam based on the input beam number and output beam number information used to predict the target downlink beam.
  • the number of input beams can be the number of input Tx beams of the model, or the number of input Rx beams of the model, or the number of input beam pairs of the model, etc.
  • the number of output beams can be the number of output Tx beams of the model, or the number of output Rx beams of the model, or the number of output beam pairs of the model, etc.
  • the information of the ⁇ input beam number, output beam number ⁇ pair can represent the model. That is, the information of the ⁇ input beam number, output beam number ⁇ pair sent by the first communication device can correspond to the model ID, and the second communication device can know the mapping relationship between the output beam ID or RS ID of the model used by the first communication device and the physical beam through the ⁇ input beam number, output beam number ⁇ pair information.
  • FIG. 3 is a third flow chart of a downlink beam prediction method provided by an embodiment of the present disclosure. As shown in FIG. 3 , the method is applied to a terminal, including:
  • Step 300 Receive a reference signal RS configuration sent by a network device.
  • Step 301 Based on RS configuration, artificial intelligence or machine learning for downlink beam prediction AI/ML models are trained or updated.
  • Step 302 Based on the trained or updated AI/ML model, one or more target downlink beams are predicted, and the target downlink beams are used for information transmission between the network device and the terminal.
  • the network device may determine the RS configuration and send it to the terminal.
  • the first information determined by the network device may be pre-defined by the network device itself or pre-defined by a third-party device and sent to the network device.
  • the terminal After the terminal receives the RS configuration sent by the network device, it can train or update the AI/ML model for downlink beam prediction according to the RS configuration.
  • updating the model includes fine-tuning the model, that is, after the terminal receives the RS configuration, it can fine-tune the AI/ML model for downlink beam prediction according to the RS configuration.
  • the functions of the AI/ML model for downlink beam prediction may include: spatial domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • Spatial domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the optimal beam among a large number of beams; frequency domain prediction refers to measuring the beam on frequency 1 and predicting the optimal beam among another beam on frequency 2; time domain prediction refers to measuring the beam at the present moment and predicting the optimal beam at the future moment.
  • the RS measurement result can be any measurement result that can represent the downlink beam reception performance, such as RSRP, and the specific form is not limited.
  • the terminal can measure the RS and use the RS measurement results through the trained or updated model to predict one or more target downlink beams.
  • the predicted target downlink beam can be a Tx beam, or an Rx beam, or a beam pair, which is used for information transmission between the terminal and the network device.
  • the RS configuration may be transmitted via a control channel or a data channel of the Uu interface, or may be transmitted via other interfaces or other channels, without limitation to the specific circumstances.
  • the downlink beam prediction method provided by the embodiment of the present disclosure sends an RS configuration to the terminal, and the terminal can train or update the AI/ML model used for downlink beam prediction based on the received RS configuration, and predict one or more target downlink beams based on the trained or updated AI/ML model, thereby effectively saving RS transmission resources, UE measurement overhead and reducing UE measurement delay.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to the prediction set, and and a measurement set pattern, where the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the RS configuration may be one or more, and the RS configuration identifier represents different RS configurations, and different RS configurations correspond to different RS configuration identifiers.
  • the RS configuration may include the type of RS (e.g., CSI-RS, SSB, PT-RS, CRS, DMRS, etc.), RS transmission resources, and RS ID, etc.
  • the RS ID is an RS indicator, for example, the indicator of CSI-RS is CRI, and the indicator of SSB is SSBRI.
  • the measurement set refers to the RS configuration or beam in the set, which is used to obtain RS measurement results as input to the model;
  • the prediction set refers to the RS configuration or beam in the set, which is used to determine one or more downlink beams as prediction labels.
  • the terminal measures the RS sent on the beam in the measurement set, obtains the RS measurement result, and uses the RS measurement result as the input of the model; the terminal measures the RS sent on the beam in the prediction set, calculates one or more best beams using a non-AI/ML method, and uses the one or more best beams as the output labels of the model.
  • the terminal Since the terminal has different behaviors after receiving the first RS configuration belonging to the measurement set and the second RS configuration belonging to the prediction set, it is necessary to distinguish which set different RS configurations belong to.
  • the prediction set When the measurement set is a subset of the prediction set, only the prediction set may be RS configured, and the RS configuration may include a second RS configuration belonging to the prediction set and a measurement set pattern, and the measurement set pattern is used to indicate the first RS configuration belonging to the measurement set in the prediction set.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set.
  • the first pattern may indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the second pattern may indicate an RS identifier corresponding to the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the second pattern is: ⁇ 2, 4, 6, 8 ⁇ , which may indicate that the 2nd, 4th, 6th, and 8th RSs in the prediction set are the measurement set.
  • the third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the third pattern may indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the third pattern may be configured as: RSs with even numbers in the prediction set are the measurement set.
  • Different rules may be represented by a character string, for example, 2 bits may represent 4 rules.
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set.
  • the fourth pattern can indicate the beam number satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the fourth pattern is: ⁇ 2, 4, 6, 8 ⁇ , which may indicate that the 2nd, 4th, 6th, and 8th downlink beams in the prediction set are the measurement set.
  • the fifth pattern is used to indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the fifth pattern indicates the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the fifth pattern is configured as follows: downlink beams with even numbers in the prediction set are used as the measurement set.
  • Different rules can be represented by a character string, for example, 2 bits can represent 4 rules.
  • a measurement set pattern indication method may be used during RS configuration, thereby saving bit overhead for configuration.
  • the method further includes:
  • the second information is sent to the network device.
  • the terminal may send second information to the network device, where the second information is used to inform the network device of the results of the one or more target downlink beams predicted by the terminal.
  • the network device After the network device receives the second information sent by the terminal, it can learn one or more target downlink beams predicted by the terminal based on the first information and the second information, and then the terminal and the network device can transmit information on the target downlink beam (i.e., best beam), thereby improving the transmission performance of the system.
  • the target downlink beam i.e., best beam
  • the second information includes one or more of the following:
  • the second information may include the beam ID corresponding to the target downlink beam, so that the network device can learn the target downlink beam inferred by the terminal through the AI/ML model based on the beam ID corresponding to the target downlink beam.
  • the mapping relationship between the beam ID and the physical beam is known to the network device. That is, the beam ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the terminal using the first information can be recognized by the network device and correspond to the actual physical target downlink beam, so that the terminal and the network device can subsequently communicate using the target downlink beam.
  • the terminal only infers one target downlink beam through the AI/ML model, and the terminal can send the beam ID corresponding to the target downlink beam to the network device.
  • the terminal infers multiple target downlink beams through an AI/ML model, and the terminal may select a beam ID corresponding to one of the target downlink beams and send it to the network device; or, The terminal can send beam IDs corresponding to multiple target downlink beams to the network device, and the network device selects one of them.
  • the second information may include the RS ID corresponding to the target downlink beam, so that the network device can learn the target downlink beam inferred by the terminal through the AI/ML model based on the RS ID corresponding to the target downlink beam.
  • the mapping relationship between the RS ID and the physical beam is known to the network device. That is, the RS ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the terminal using the first information can be recognized by the network device and correspond to the actual physical target downlink beam, so that the terminal and the network device can subsequently communicate using the target downlink beam.
  • the terminal only infers one target downlink beam through the AI/ML model, and the terminal can send the RS ID corresponding to the target downlink beam to the network device.
  • the terminal infers multiple target downlink beams through an AI/ML model, and the terminal may select an RS ID corresponding to one of the target downlink beams and send it to a network device; or, the terminal may send RS IDs corresponding to multiple target downlink beams to the network device, and the network device may select one of them.
  • the second information may include an identifier of the RS configuration for predicting the target downlink beam, so that the network device can know which AI/ML model the terminal uses to infer the target downlink beam based on the identifier of the RS configuration for predicting the target downlink beam.
  • the terminal can also send the RS configuration identifier to represent the above model ID.
  • the network device can know the mapping relationship between the RS ID output by the model used by the terminal and the physical beam through the RS configuration identifier.
  • the second information may include the input beam number and output beam number information of the AI/ML model used to predict the target downlink beam, so that the network device can know which AI/ML model the terminal uses to infer the target downlink beam based on the input beam number and output beam number information used to predict the target downlink beam.
  • the number of input beams can be the number of input Tx beams of the model, or the number of input Rx beams of the model, or the number of input beam pairs of the model, etc.
  • the number of output beams can be the number of output Tx beams of the model, or the number of output Rx beams of the model, or the number of output beam pairs of the model, etc.
  • FIG. 4 is a fourth flow chart of a downlink beam prediction method provided in an embodiment of the present disclosure. As shown in FIG. 4 , the method is applied to a network device, including:
  • Step 400 determine the reference signal RS configuration, where the RS configuration is used to train or update an artificial intelligence or machine learning AI/ML model, where the AI/ML model is used to predict one or more target downlink beams, and the target downlink beam is used to transmit information between a network device and a terminal.
  • Step 401 Send RS configuration to the terminal.
  • the network device may determine the RS configuration and send it to the terminal.
  • the first information determined by the network device may be pre-defined by the network device itself or pre-defined by a third-party device and sent to the network device.
  • the terminal After receiving the RS configuration sent by the network device, the terminal can use the RS configuration to The AI/ML model for downlink beam prediction is trained or updated.
  • updating the model includes fine-tuning the model, that is, after the terminal receives the RS configuration, it can fine-tune the AI/ML model for downlink beam prediction according to the RS configuration.
  • the functions of the AI/ML model for downlink beam prediction may include: spatial domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • Spatial domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the optimal beam among a large number of beams; frequency domain prediction refers to measuring the beam on frequency 1 and predicting the optimal beam among another beam on frequency 2; time domain prediction refers to measuring the beam at the present moment and predicting the optimal beam at the future moment.
  • the RS measurement result can be any measurement result that can represent the downlink beam reception performance, such as RSRP, and the specific form is not limited.
  • the terminal can measure the RS and use the RS measurement results through the trained or updated model to predict one or more target downlink beams.
  • the predicted target downlink beam can be a Tx beam, or an Rx beam, or a beam pair, which is used for information transmission between the terminal and the network device.
  • the RS configuration may be transmitted via a control channel or a data channel of the Uu interface, or may be transmitted via other interfaces or other channels, without limitation to the specific circumstances.
  • the downlink beam prediction method provided by the embodiment of the present disclosure sends an RS configuration to the terminal, and the terminal can train or update the AI/ML model used for downlink beam prediction based on the received RS configuration, and predict one or more target downlink beams based on the trained or updated AI/ML model, thereby effectively saving RS transmission resources, UE measurement overhead and reducing UE measurement delay.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to the prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate the first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the RS configuration may be one or more, and the RS configuration identifier represents different RS configurations, and different RS configurations correspond to different RS configuration identifiers.
  • the RS configuration may include the type of RS (e.g., CSI-RS, SSB, PT-RS, CRS, DMRS, etc.), RS transmission resources, and RS ID, etc.
  • the RS ID is an RS indicator, for example, the indicator of CSI-RS is CRI, and the indicator of SSB is SSBRI.
  • the measurement set refers to the RS configuration or beam in the set, which is used to obtain RS measurement results as input to the model;
  • the prediction set refers to the RS configuration or beam in the set, which is used to determine one or more downlink beams as prediction labels.
  • the terminal measures the RS sent on the beam in the measurement set, obtains the RS measurement result, and uses the RS measurement result as the input of the model; the terminal measures the RS sent on the beam in the prediction set, calculates one or more best beams using a non-AI/ML method, and uses the one or more best beams as the output labels of the model.
  • the terminal Since the terminal has different behaviors after receiving the first RS configuration belonging to the measurement set and the second RS configuration belonging to the prediction set, it is necessary to distinguish which set different RS configurations belong to.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set.
  • the first pattern may indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the second pattern is used to indicate the RS configuration corresponding to the measurement set. logo.
  • the second pattern may indicate an RS identifier corresponding to the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the second pattern is: ⁇ 2, 4, 6, 8 ⁇ , which may indicate that the 2nd, 4th, 6th, and 8th RSs in the prediction set are the measurement set.
  • the third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the third pattern may indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the third pattern may be configured as: RSs with even numbers in the prediction set are the measurement set.
  • Different rules may be represented by a character string, for example, 2 bits may represent 4 rules.
  • the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set.
  • the fourth pattern is: ⁇ 2, 4, 6, 8 ⁇ , which may indicate that the 2nd, 4th, 6th, and 8th downlink beams in the prediction set are the measurement set.
  • the fifth pattern is used to indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the fifth pattern can indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set, thereby indicating the first RS configuration belonging to the measurement set in the prediction set.
  • the fifth pattern is configured as follows: downlink beams with even numbers in the prediction set are used as the measurement set.
  • Different rules can be represented by a character string, for example, 2 bits can represent 4 rules.
  • a measurement set pattern indication method may be used during RS configuration, thereby saving bit overhead for configuration.
  • the method further comprises:
  • a target downlink beam for information transmission with the terminal is determined.
  • the terminal may send second information to the network device, where the second information is used to inform the network device of the results of the one or more target downlink beams predicted by the terminal.
  • the network device After the network device receives the second information sent by the terminal, it can learn one or more target downlink beams predicted by the terminal based on the first information and the second information, and then the terminal and the network device can transmit information on the target downlink beam (i.e., best beam), thereby improving the transmission performance of the system.
  • the target downlink beam i.e., best beam
  • the second information includes one or more of the following:
  • the second information may include the beam ID corresponding to the target downlink beam, so that the network device can learn the target downlink beam inferred by the terminal through the AI/ML model based on the beam ID corresponding to the target downlink beam.
  • the mapping relationship between the beam ID and the physical beam is known to the network device. That is, the beam ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the terminal using the first information can be recognized by the network device and correspond to the actual physical target downlink beam, so that the terminal and the network device can subsequently communicate using the target downlink beam.
  • the terminal only infers one target downlink beam through the AI/ML model, and the terminal can send the beam ID corresponding to the target downlink beam to the network device.
  • the terminal infers multiple target downlink beams through an AI/ML model, and the terminal may select a beam ID corresponding to one of the target downlink beams and send it to a network device; or, the terminal may send beam IDs corresponding to multiple target downlink beams to the network device, and the network device may select one of them.
  • the second information may include the RS ID corresponding to the target downlink beam, so that the network device can know the target downlink beam inferred by the terminal through the AI/ML model according to the RS ID corresponding to the target downlink beam. bundle.
  • the mapping relationship between the RS ID and the physical beam is known to the network device. That is, the RS ID corresponding to one or more target downlink beams predicted by the AI/ML model trained or updated by the terminal using the first information can be recognized by the network device and correspond to the actual physical target downlink beam, so that the terminal and the network device can subsequently communicate using the target downlink beam.
  • the terminal only infers one target downlink beam through the AI/ML model, and the terminal can send the RS ID corresponding to the target downlink beam to the network device.
  • the terminal infers multiple target downlink beams through an AI/ML model, and the terminal may select an RS ID corresponding to one of the target downlink beams and send it to a network device; or, the terminal may send RS IDs corresponding to multiple target downlink beams to the network device, and the network device may select one of them.
  • each model can correspond to a model ID.
  • the terminal not only needs to send the beam ID or RS ID corresponding to the predicted target downlink beam, but also needs to send the model ID corresponding to the predicted target downlink beam.
  • the network device can know which AI/ML model the terminal uses to infer the target downlink beam based on the model identifier.
  • the second information may include an identifier of the RS configuration for predicting the target downlink beam, so that the network device can know which AI/ML model the terminal uses to infer the target downlink beam based on the identifier of the RS configuration for predicting the target downlink beam.
  • the terminal can also send the RS configuration identifier to represent the above model ID.
  • the network device can know the mapping relationship between the RS ID output by the model used by the terminal and the physical beam through the RS configuration identifier.
  • the second information may include the input beam number and output beam number information of the AI/ML model used to predict the target downlink beam, so that the network device can know which AI/ML model the terminal uses to infer the target downlink beam based on the input beam number and output beam number information used to predict the target downlink beam.
  • the number of input beams can be the number of input Tx beams of the model, or the number of input Rx beams of the model, or the number of input beam pairs of the model, etc.
  • the number of output beams can be the number of output Tx beams of the model, or the number of output Rx beams of the model, or the number of output beam pairs of the model, etc.
  • the ⁇ input beam number, output beam number ⁇ pair information can represent the model. That is, the ⁇ input beam number, output beam number ⁇ pair information sent by the terminal can correspond to the model ID, and the network device can know the mapping relationship between the output beam ID or RS ID of the model used by the terminal and the physical beam through the ⁇ input beam number, output beam number ⁇ pair information.
  • Embodiment 1 UE trains or updates an AI/ML model based on a data set.
  • Step 1 The UE receives one or more data sets.
  • the one or more data sets may come from a base station or a third-party device.
  • a dataset includes: a dataset identifier and one or more dataset samples.
  • the dataset samples include: multiple beam IDs, the RSRP of the beam corresponding to the beam ID (as input for training, updating or fine-tuning of the AI/ML model), and one or more optimal beam IDs (as output labels for training, updating or fine-tuning of the AI/ML model).
  • the beam here can be a beam pair, or a Tx beam, or a Rx beam.
  • Step 2 The UE trains, updates, or fine-tunes the AI/ML model based on the data set to obtain the corresponding AI/ML model.
  • the UE is trained, updated or fine-tuned with different data sets, different AI/ML models can be obtained. Different data sets can be targeted at different application scenarios, base station configurations, or AI functions.
  • the application scenarios include: urban, rural, indoor, outdoor, highway, high-speed rail, Uma, Umi, etc.;
  • base station configuration includes: base station antenna configuration, beam configuration, reference signal configuration, etc.;
  • AI functions mainly include air domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • air domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the best beam among a large number of beams
  • frequency domain prediction refers to measuring the beam on frequency 1 and predicting the best beam among another beam on frequency 2
  • time domain prediction refers to measuring the beam at the present moment and predicting the best beam at the future moment.
  • Step 3 The UE measures the RS sent on the Tx beam in the measurement set, uses the AI/ML model to predict the optimal beam, and infers one or more optimal beam IDs.
  • the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam and Rx beam pair; if it is Tx beam prediction, the UE uses one Rx beam to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam; if it is Rx beam prediction, the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on a Tx beam sent by the base station, and obtains the RSRP corresponding to each Rx beam.
  • the RS for example, CSI-RS or SSB
  • the above RSRP is used as the input of the AI/ML model to perform inference to obtain one or more optimal beam IDs.
  • beam ID refers to the beam ID of the Tx beam in the optimal beam pair predicted by beam pair, or the beam ID of the optimal Tx beam predicted by Tx beam, or the beam ID of the optimal Rx beam predicted by Rx beam.
  • One or more optimal beams are inferred, which may be one or more of the Tx beams sent by the base station, or other Tx beams of the base station instead of the Tx beams sent by the base station.
  • Step 4 The UE reports the optimal beam ID or IDs to the base station.
  • the beam ID is reported to the base station; If multiple optimal beam IDs are obtained, one beam ID is selected and reported to the base station, or all are reported to the base station and the base station selects one.
  • the UE uses different data sets to train, update or fine-tune multiple AI/ML models, the UE also needs to report the model ID corresponding to the data set to the base station. That is, the UE not only needs to report to the base station the optimal beam ID predicted, but also needs to report the ID of the model that predicts the beam.
  • the terminal can also report the dataset identification information, representing the above model ID.
  • the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information can represent the model. That is, the UE can report the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information to represent the above model ID.
  • Step 5 The base station obtains the optimal beam based on the mapping relationship between the beam ID and the physical beam, and then uses the beam to communicate with the UE.
  • the base station obtains the optimal beam based on the mapping relationship between the beam ID and the physical beam, and the base station uses the beam to communicate with the UE.
  • the base station obtains the mapping relationship between the beam ID and the physical beam of the model output obtained by the UE through training, updating or fine-tuning with the data set or the data set identification information or the ⁇ input Tx beam number, output Tx beam number ⁇ pair according to the model ID reported by the UE, or the data set identification information, or the ⁇ input Tx beam number, output Tx beam number ⁇ pair, thereby obtaining the physically actual optimal beam, and the base station uses the beam to communicate with the UE.
  • Embodiment 2 UE performs training, updating or fine-tuning of AI/ML models based on RS configuration.
  • Step 1 The UE receives one or more RS configurations.
  • the one or more RS configurations are generated by the base station and sent to the UE.
  • RS configuration includes: RS configuration on the measurement set beam and RS configuration on the prediction set beam. Including RS configuration identifier, RS type (CSI-RS, SSB, PT-RS, CRS, DMRS, etc.), RS transmission resources, RS ID, etc.
  • RS configuration identifier CSI-RS, SSB, PT-RS, CRS, DMRS, etc.
  • RS type CSI-RS, SSB, PT-RS, CRS, DMRS, etc.
  • RS transmission resources RS ID, etc.
  • the beam here can be a beam pair, or a Tx beam, or a Rx beam.
  • the following optimization methods can be used when configuring the RS:
  • the UE can obtain the RS configuration of the measurement set (ie, a subset of the RS configuration of the prediction set) based on the prediction set and the pattern, thereby saving bit overhead for configuration.
  • Alt.2 configure the measurement set pattern, including the ID number of the RS in the prediction set.
  • the configuration pattern is: ⁇ 2,4,6,8 ⁇ , which means that the 2nd, 4th, 6th, and 8th RS in the prediction set are the measurement set;
  • the pattern of configuring the measurement set is a rule for the arrangement of RS numbers in the prediction set.
  • the configuration pattern is that the RSs with even numbers in the prediction set are the measurement set.
  • Different rules can be represented by strings. For example, 2 bits can represent 4 rules.
  • Alt.4 Configure the measurement set pattern, including the ID number of the beam in the prediction set. For example, if the configuration pattern is ⁇ 2,4,6,8 ⁇ , it means that the 2nd, 4th, 6th, and 8th beams in the prediction set are the measurement set. This method requires configuring the corresponding beam ID for each RS in the prediction set.
  • the pattern of configuring the measurement set is a rule for arranging beams in the prediction set.
  • the configuration pattern is that the beams with even numbers in the prediction set are the measurement set.
  • Different rules can be represented by strings, for example, 2 bits can represent 4 rules. This method requires configuring the corresponding beam ID for each RS in the prediction set.
  • Step 2 Based on the RS configuration, the UE measures the RS configured on the beam of the measurement set and the prediction set, and trains, updates, or fine-tunes the AI/ML model according to the measurement results to obtain the corresponding AI/ML model.
  • RS configurations can be targeted at different application scenarios, base station configurations, or AI functions.
  • the application scenarios include: urban, rural, indoor, outdoor, highway, high-speed rail, Uma, Umi, etc.;
  • base station configuration includes: base station antenna configuration, beam configuration, reference signal configuration, etc.;
  • AI functions mainly include air domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • air domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the best beam among a large number of beams
  • frequency domain prediction refers to measuring the beam on frequency 1 and predicting the best beam among another beam on frequency 2
  • time domain prediction refers to measuring the beam at the present moment and predicting the best beam at the future moment.
  • Step 3 The UE measures the RS sent on the Tx beam in the measurement set, uses the AI/ML model to predict the optimal beam, and infers the RS ID corresponding to one or more optimal beams.
  • the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam and Rx beam pair; if it is Tx beam prediction, the UE uses one Rx beam to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam; if it is Rx beam prediction, the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on a Tx beam sent by the base station, and obtains the RSRP corresponding to each Rx beam.
  • the RS for example, CSI-RS or SSB
  • the above RSRP is used as the input of the AI/ML model to perform inference to obtain the RS ID corresponding to one or more optimal beams.
  • RS ID refers to the RS ID configured on the Tx beam in the optimal beam pair predicted by the beam pair, or the RS ID configured on the optimal Tx beam predicted by the Tx beam, or the RS ID of the optimal Rx beam predicted by the Rx beam.
  • One or more optimal beams are inferred, which may be one or more of the Tx beams sent by the base station, or other Tx beams belonging to the base station, which are not the Tx beams sent by the base station.
  • Step 4 The UE reports the RS ID of the optimal beam or beams to the base station.
  • the RS ID of an optimal beam is obtained by inference, the RS ID of the beam is reported to the base station; if multiple RS IDs of the optimal beam are obtained by inference, the RS ID of one beam is selected and reported. To the base station, or report both to the base station, and the base station selects one.
  • the UE uses different RS configurations to train, update or fine-tuning to obtain multiple AI/ML models, the UE also needs to report the model ID corresponding to the RS configuration to the base station. That is, the UE not only needs to report to the base station the optimal beam ID predicted, but also needs to report the ID of the model that predicts the beam.
  • the terminal can also report the RS configuration identification information, representing the above model ID.
  • the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information can represent the model. That is, the UE can report the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information to represent the above model ID.
  • Step 5 The base station obtains the optimal beam based on the mapping relationship between RS ID and physical beam, and then uses the beam to communicate with UE.
  • the base station Because the RS configuration is generated and configured by the base station, the base station obtains the optimal beam based on the mapping relationship between the RS ID and the physical beam, and then the base station uses the beam to communicate with the UE.
  • the base station obtains the mapping relationship between the beam ID and the physical beam of the model output obtained by the UE through training, updating or fine-tuning with the RS configuration or the RS configuration identification information or the ⁇ input Tx beam number, output Tx beam number ⁇ pair reported by the UE, thereby obtaining the physically actual optimal beam, and the base station uses the beam to communicate with the UE.
  • Embodiment 3 A scenario where the measurement set is a subset of the prediction set.
  • Step 1 The UE receives one or more RS configurations.
  • the one or more RS configurations are generated by the base station and sent to the UE.
  • RS configuration includes: RS configuration on the measurement set beam and RS configuration on the prediction set beam.
  • RS configuration on the measurement set beam or RS configuration on the prediction set beam, including RS configuration identifier, RS type (CSI-RS, SSB, PT-RS, CRS, DMRS, etc.), RS transmission resource, RS ID, etc.
  • RS ID that is, RS indication, for example, the indication of CSI-RS is CRI, The indication of SSB is SSBRI.
  • the beam here can be a beam pair, or a Tx beam, or a Rx beam.
  • RS configuration methods include:
  • the UE can obtain the RS configuration of the measurement set (ie, a subset of the RS configuration of the prediction set) based on the prediction set and the pattern, thereby saving bit overhead for configuration.
  • Alt.2 configure the measurement set pattern, including the ID number of the RS in the prediction set.
  • the configuration pattern is: ⁇ 2,4,6,8 ⁇ , which means that the 2nd, 4th, 6th, and 8th RS in the prediction set are the measurement set;
  • the pattern of configuring the measurement set is a rule for the arrangement of RS numbers in the prediction set.
  • the configuration pattern is that the RSs with even numbers in the prediction set are the measurement set.
  • Different rules can be represented by strings. For example, 2 bits can represent 4 rules.
  • Alt.4 Configure the measurement set pattern, including the ID number of the beam in the prediction set. For example, if the configuration pattern is ⁇ 2,4,6,8 ⁇ , it means that the 2nd, 4th, 6th, and 8th beams in the prediction set are the measurement set. This method requires configuring the corresponding beam ID for each RS in the prediction set.
  • the pattern of configuring the measurement set is a rule for arranging beams in the prediction set.
  • the configuration pattern is that the beams with even numbers in the prediction set are the measurement set.
  • Different rules can be represented by strings, for example, 2 bits can represent 4 rules. This method requires configuring the corresponding beam ID for each RS in the prediction set.
  • Step 2 Based on the RS configuration, the UE measures the RS configured on the beam of the measurement set and the prediction set, and trains, updates, or fine-tunes the AI/ML model according to the measurement results to obtain the corresponding AI/ML model.
  • the UE obtains the RS configuration of the measurement set based on the RS configuration of the prediction set and the pattern configuration of the measurement set, and measures the RS configured on the beams of the measurement set and the prediction set.
  • the measurement set refers to the RS sent on the beam in the set that the UE needs to measure, so as to obtain Input to AI/ML models (measurements of RS as input).
  • the prediction set means that the UE needs to predict one or more optimal beams in the set.
  • the UE needs to measure the prediction set and calculate one or more optimal beams using a non-AI method to obtain the output label of the training, updating or fine-tuning of the AI/ML model.
  • Step 3 The UE measures the RS sent on the Tx beam in the measurement set, uses the AI/ML model to predict the optimal beam, and infers the RS ID corresponding to one or more optimal beams.
  • the UE obtains the RS configuration of the measurement set based on the prediction set RS configuration and the pattern configuration of the measurement set, and measures the RS configured on the measurement set.
  • Step 4 The UE reports the RS ID of the optimal beam or beams to the base station.
  • Step 5 The base station obtains the optimal beam based on the mapping relationship between RS ID and physical beam, and then uses this beam to communicate with UE.
  • Embodiment 4 UE performs training, updating or fine-tuning of AI/ML models based on beam description information.
  • Step 1 The UE receives one or more beam description information.
  • the one or more beam description information may be generated by the base station and sent to the UE, or generated by a third party and sent to the base station and the UE.
  • the description information of the beam includes the beam ID or RS ID, the base station antenna configuration information of the beam corresponding to the beam ID or RS ID, the angle information of the beam, and/or the width information of the beam.
  • the above-mentioned Beam ID can be the number of the Tx beam sent by the base station, and the RS ID can be the indication of the RS sent by the base station on the Tx beam.
  • the indication of CSI-RS is CRI
  • the indication of SSB is SSBRI.
  • the beam here can be a beam pair, or a Tx beam, or a Rx beam.
  • Step 2 Based on the beam description information and the RS configured on the Tx beam, the UE performs training, updating, or fine-tuning of the AI/ML model according to the measurement results to obtain the corresponding AI/ML model.
  • the UE If the UE is based on different beam description information and measurement results, it can train, update or fine-tune different AI/ML models. Different beam description information can be targeted at different application scenarios, base station configurations, or AI functions.
  • the application scenarios include: urban, rural, indoor, outdoor, highway, high-speed rail, Uma, Umi, etc.;
  • base station configuration includes: base station antenna configuration, beam configuration, reference signal configuration, etc.;
  • AI functions mainly include air domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • air domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the best beam among a large number of beams
  • frequency domain prediction refers to measuring the beam on frequency 1 and predicting the best beam among another beam on frequency 2
  • time domain prediction refers to measuring the beam at the present moment and predicting the best beam at the future moment.
  • Step 3 The UE measures the RS sent on the Tx beam in the measurement set, uses the AI/ML model to predict the optimal beam, and infers the beam ID or RS ID corresponding to one or more optimal beams.
  • the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam and Rx beam pair; if it is Tx beam prediction, the UE uses one Rx beam to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam; if it is Rx beam prediction, the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on a Tx beam sent by the base station, and obtains the RSRP corresponding to each Rx beam.
  • the RS for example, CSI-RS or SSB
  • the above RSRP is used as the input of the AI/ML model to infer the beam ID or RS ID corresponding to one or more optimal beams.
  • beam ID or RS ID refers to the beam ID or RS ID corresponding to the Tx beam in the optimal beam pair predicted by beam pair, or the beam ID or RS ID corresponding to the optimal Tx beam predicted by Tx beam, or the beam ID or RS ID of the optimal Rx beam predicted by Rx beam.
  • One or more optimal beams are inferred, which may be one or more of the Tx beams sent by the base station, or other Tx beams belonging to the base station, which are not the Tx beams sent by the base station.
  • Step 4 The UE reports the beam ID or RS ID of the optimal beam or beams to the base station. stand.
  • the beam ID or RS ID of an optimal beam is obtained through reasoning, the beam ID or RS ID of the beam is reported to the base station; if multiple beam IDs or RS IDs of the optimal beam are obtained through reasoning, the beam ID or RS ID of one beam is selected and reported to the base station, or both are reported to the base station and the base station selects one.
  • the UE uses different beam description information to train, update or fine-tuning multiple AI/ML models, the UE also needs to report the model ID corresponding to the beam description information to the base station. That is, the UE not only needs to report to the base station the optimal beam ID predicted, but also needs to report the ID of the model that predicts the beam.
  • the terminal can also report the identification information of the beam description information, representing the above model ID.
  • the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information can represent the model. That is, the UE can report the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information to represent the above model ID.
  • Step 5 The base station obtains the optimal beam based on the mapping relationship between the beam ID or RS ID and the physical beam, and then the base station uses the beam to communicate with the UE.
  • the base station Because the beam description information is generated by the base station or a third party and sent to the base station and UE, the base station obtains the actual optimal beam based on the mapping relationship between the beam ID or RS ID and the physical beam, and the base station uses the beam to communicate with the UE.
  • the base station obtains the mapping relationship between the beam ID and the physical beam of the model output obtained by the UE using the beam description information or ⁇ input Tx beam number, output Tx beam number ⁇ pair for training, updating or fine-tuning based on the model ID reported by the UE, or the identification information of the beam description information, or the information pair of ⁇ input Tx beam number, output Tx beam number ⁇ , thereby obtaining the physically actual optimal beam, and the base station uses this beam to communicate with the UE.
  • Embodiment 5 The base station trains or updates the AI/ML model based on the data set.
  • Step 1 The base station receives one or more data sets.
  • the one or more data sets may come from the UE or a third-party device.
  • a dataset includes: a dataset identifier and one or more dataset samples.
  • the dataset samples include: multiple beam IDs, the RSRP of the beam corresponding to the beam ID (as input for training, updating or fine-tuning of the AI/ML model), and one or more optimal beam IDs (as output labels for training, updating or fine-tuning of the AI/ML model).
  • the beam here can be a beam pair, or a Tx beam, or a Rx beam.
  • Step 2 The base station trains, updates, or fine-tunes the AI/ML model based on the data set to obtain the corresponding AI/ML model.
  • the base station is trained, updated or fine-tuned with different data sets, different AI/ML models can be obtained. Different data sets can be targeted at different application scenarios, base station configurations, or AI functions.
  • the application scenarios include: urban, rural, indoor, outdoor, highway, high-speed rail, Uma, Umi, etc.;
  • base station configuration includes: base station antenna configuration, beam configuration, reference signal configuration, etc.;
  • AI functions mainly include air domain prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • air domain prediction refers to measuring a small number of beams, or beams of other reference signal types, and predicting the best beam among a large number of beams
  • frequency domain prediction refers to measuring the beam on frequency 1 and predicting the best beam among another beam on frequency 2
  • time domain prediction refers to measuring the beam at the present moment and predicting the best beam at the future moment.
  • Step 3 The UE measures the RS received on the Tx beam in the measurement set, and then feeds the measurement results back to the base station.
  • the base station uses the AI/ML model to predict the optimal beam and infer one or more optimal beam IDs.
  • the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam and Rx beam pair; if it is Tx beam prediction, the UE uses one Rx beam to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam; if it is Rx beam prediction, the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on a Tx beam sent by the base station, and obtains the RSRP corresponding to each Rx beam. The corresponding RSRP.
  • RSRP for example, CSI-RS or SSB
  • the UE reports the measurement results to the base station, which uses the AI/ML model to perform beam prediction.
  • the above RSRP is used as the input of the AI/ML model to perform inference to obtain one or more optimal beam IDs.
  • beam ID refers to the beam ID of the Rx beam in the optimal beam pair predicted by beam pair, or the beam ID of the optimal Rx beam predicted by Rx beam, or the beam ID of the optimal Rx beam predicted by Rx beam.
  • One or more optimal beams are inferred, which may be one or more of the Rx beams received by the UE, or may not be the Rx beam received by the UE, but other Rx beams of the UE.
  • Step 4 The base station sends the one or more optimal beam IDs to the UE.
  • the beam ID is sent to the UE; if multiple optimal beam IDs are obtained through reasoning, one beam ID is selected and sent to the UE, or all beam IDs are sent to the UE and the UE selects one.
  • the base station uses different data sets to train, update or fine-tuning to obtain multiple AI/ML models, the base station also needs to send the model ID corresponding to the data set to the UE. That is, the base station needs to send not only the optimal beam ID predicted by the UE, but also the ID of the model that predicts the beam.
  • the base station can also send the data set identification information, representing the above model ID.
  • the base station trains, updates or fine-tunes each ⁇ input Tx beam number, output Tx beam number ⁇ pair to obtain an AI model, then the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information can represent the model. That is, the base station can send the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information to represent the above model ID.
  • Step 5 The UE obtains the optimal beam based on the mapping relationship between the beam ID and the physical beam, and then the UE uses the beam to communicate with the base station.
  • the UE obtains the optimal beam according to the mapping relationship between the beam ID and the physical beam, and the UE communicates with the base station using the beam.
  • the UE obtains the mapping relationship between the beam ID and the physical beam of the model output obtained by the base station using the data set or the ⁇ input Tx beam number, output Tx beam number ⁇ pair for training, updating or fine-tuning based on the model ID, data set identification information, or ⁇ input Tx beam number, output Tx beam number ⁇ pair sent by the base station, thereby obtaining the physically actual optimal beam, and the UE uses the beam to communicate with the base station.
  • Embodiment 6 The base station performs training, updating or fine-tuning of the AI/ML model based on the description information of the beam.
  • Step 1 The base station receives one or more beam description information.
  • the one or more beam description information may be generated by the UE and sent to the base station, or generated by a third party and sent to the UE and the base station.
  • the description information of the beam includes the beam ID or RS ID, the UE antenna configuration information of the beam corresponding to the beam ID or RS ID, the angle information of the beam, and/or the width information of the beam.
  • the above-mentioned Beam ID can be the number of the Rx beam received by the UE, and the RS ID can be the indication of the RS received by the UE on the Rx beam, for example, the indication of CSI-RS is CRI, and the indication of SSB is SSBRI.
  • the beam here can be a beam pair, or a Tx beam, or a Rx beam.
  • Step 2 Based on the beam description information and the RS configured on the Rx beam, the base station trains, updates, or fine-tunes the AI/ML model according to the measurement results to obtain the corresponding AI/ML model.
  • the base station is based on different beam description information and measurement results, different AI/ML models can be trained, updated or fine-tuned. Different beam description information can be targeted at different application scenarios, base station configurations, or AI functions.
  • application scenarios include: urban, rural, indoor, outdoor, highway, high-speed rail, Uma, Umi, etc.; base station configuration includes: base station antenna configuration, beam configuration, reference signal configuration, etc.; AI functions mainly include airspace prediction, frequency domain prediction, time domain prediction, downlink beam pair prediction, downlink Tx beam prediction, downlink Rx beam prediction, etc.
  • airspace prediction refers to measuring a small amount of The frequency domain prediction is to measure the beam on frequency 1 and predict the best beam in the beam on another frequency 2; the time domain prediction is to measure the beam at the present moment and predict the best beam at the future moment.
  • Step 3 The UE measures the RS sent on the Tx beam in the measurement set, and then feeds the measurement results back to the base station.
  • the base station uses the AI/ML model to predict the optimal beam and infers the beam ID or RS ID corresponding to one or more optimal beams.
  • the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam and Rx beam pair; if it is Tx beam prediction, the UE uses one Rx beam to measure the RS (for example, CSI-RS or SSB) on each Tx beam sent by the base station, and obtains the RSRP corresponding to each Tx beam; if it is Rx beam prediction, the UE uses different Rx beams to measure the RS (for example, CSI-RS or SSB) on a Tx beam sent by the base station, and obtains the RSRP corresponding to each Rx beam.
  • the RS for example, CSI-RS or SSB
  • the UE reports the measurement results to the base station, which uses the AI/ML model to perform beam prediction.
  • the above RSRP is used as the input of the AI/ML model to infer the beam ID or RS ID corresponding to one or more optimal beams.
  • beam ID or RS ID refers to the beam ID or RS ID corresponding to the Rx beam in the optimal beam pair predicted by beam pair, or the beam ID or RS ID corresponding to the optimal Rx beam predicted by Rx beam, or the beam ID or RS ID of the optimal Rx beam predicted by Rx beam.
  • One or more optimal beams are inferred, which may be one or more of the Rx beams received by the UE, or may not be the Rx beam received by the UE, but other Rx beams of the UE.
  • Step 4 The base station sends the beam ID or RS ID of the one or more optimal beams to the UE.
  • the beam ID or RS ID of an optimal beam is obtained by reasoning, the beam ID or RS ID of the beam is sent to the UE; if multiple optimal beam IDs or RS IDs are obtained by reasoning, one beam ID or RS ID is selected and sent to the UE, or both are sent to the UE and the UE selects one.
  • the base station uses different beam description information for training, updating or fine-tuning to obtain multiple For AI/ML models, the base station also needs to send the model ID corresponding to the beam description information to the UE. That is, the base station not only needs to send the optimal beam ID predicted by the UE, but also needs to send the ID of the model that predicts the beam.
  • the UE can also send the identification information of the beam description information, representing the above model ID.
  • the base station trains, updates or fine-tunes each ⁇ input Tx beam number, output Tx beam number ⁇ pair to obtain an AI model, then the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information can represent the model. That is, the base station can send the ⁇ input Tx beam number, output Tx beam number ⁇ pair of information to represent the above model ID.
  • Step 5 The UE obtains the optimal beam based on the mapping relationship between the beam ID or RS ID and the physical beam, and then the UE communicates with the base station using the beam.
  • the UE Because the beam description information is generated by the UE or a third party and sent to the UE and the base station, the UE obtains the actual optimal beam based on the mapping relationship between the beam ID or RS ID and the physical beam, and the UE uses the beam to communicate with the base station.
  • the UE obtains the mapping relationship between the beam ID and the physical beam of the model output by the base station using the beam description information or ⁇ input Tx beam number, output Tx beam number ⁇ for training, updating or fine-tuning based on the model ID sent by the base station, or the identification information of the beam description information, or the information pair of ⁇ input Tx beam number, output Tx beam number ⁇ , thereby obtaining the physically actual optimal beam, and the UE uses the beam to communicate with the base station.
  • the methods and devices provided in the various embodiments of the present disclosure are based on the same application concept. Since the methods and devices solve problems based on similar principles, the implementation of the devices and methods can refer to each other, and the repeated parts will not be repeated.
  • FIG5 is a schematic diagram of the structure of a first communication device provided in an embodiment of the present disclosure.
  • the first communication device includes a memory 520, a transceiver 510, and a processor 500; wherein the processor 500 and the memory 520 may also be arranged physically separately.
  • the memory 520 is used to store computer programs; the transceiver 510 is used to send and receive data under the control of the processor 500.
  • the transceiver 510 is used to receive and send data under the control of the processor 500 .
  • the bus architecture may include any number of interconnected buses and bridges, specifically various circuits of one or more processors represented by processor 500 and memory represented by memory 520 are linked together.
  • the bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described in this disclosure.
  • the bus interface provides an interface.
  • the transceiver 510 may be a plurality of components, including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which may include a wireless channel, a wired channel, an optical cable, and other transmission media.
  • the processor 500 is responsible for managing the bus architecture and general processing, and the memory 520 can store data used by the processor 500 when performing operations.
  • the processor 500 can be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or a complex programmable logic device (CPLD).
  • the processor can also adopt a multi-core architecture.
  • the processor 500 calls the computer program stored in the memory 520 to execute any of the methods provided in the embodiments of the present disclosure according to the obtained executable instructions, for example:
  • the first information includes one or more of a data set, a reference signal RS configuration, and beam description information;
  • an artificial intelligence or machine learning AI/ML model for downlink beam prediction is trained or updated
  • one or more target downlink beams are predicted, and the target downlink beams are used for information transmission between the first communication device and the second communication device.
  • receiving the first information includes:
  • the first information sent by the second communication device is received.
  • the dataset contains one or more of the following:
  • One or more data set samples each data set sample including multiple beam identifiers, RS measurement results of the beam corresponding to each beam identifier, and one or more beam identifiers used as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • a fourth pattern where the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • the fifth pattern is used to indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the method further includes:
  • FIG6 is a schematic diagram of the structure of a second communication device provided in an embodiment of the present disclosure.
  • the second communication device includes a memory 620, a transceiver 610, and a processor 600; wherein the processor 600 and the memory 620 may also be physically arranged separately.
  • the memory 620 is used to store computer programs; the transceiver 610 is used to send and receive data under the control of the processor 600.
  • the transceiver 610 is used to receive and send data under the control of the processor 600 .
  • the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 600 and various circuits of memory represented by memory 620 are linked together.
  • the bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described in this disclosure.
  • the bus interface provides an interface.
  • the transceiver 610 may be a plurality of components, including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which may include a wireless channel, a wired channel, an optical cable, and other transmission media.
  • the processor 600 is responsible for managing the bus architecture and general processing, and the memory 620 can store data used by the processor 600 when performing operations.
  • the processor 600 may be a CPU, an ASIC, an FPGA or a CPLD, and the processor may also adopt a multi-core architecture.
  • the processor 600 calls the computer program stored in the memory 620 to execute any of the methods provided in the embodiments of the present disclosure according to the obtained executable instructions, for example:
  • the first information includes one or more of a data set, an RS configuration, and a beam description information; the first information is used to train or update an artificial intelligence or machine learning AI/ML model, The AI/ML model is used to predict one or more target downlink beams, where the target downlink beams are used for information transmission between the second communication device and the first communication device;
  • First information is sent to a first communication device.
  • the dataset contains one or more of the following:
  • Each data set sample including multiple beam identifiers, a reference signal measurement result of a beam corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • a fourth pattern where the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • the fifth pattern is used to indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the method further comprises:
  • the second information includes one or more of the following:
  • FIG7 is a schematic diagram of the structure of a terminal provided in an embodiment of the present disclosure.
  • the terminal includes a memory 720, a transceiver 710, and a processor 700; wherein the processor 700 and the memory 720 may also be arranged physically separately.
  • the memory 720 is used to store computer programs; the transceiver 710 is used to send and receive data under the control of the processor 700.
  • the transceiver 710 is used to receive and send data under the control of the processor 700 .
  • the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 700 and memory represented by memory 720.
  • the bus architecture may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described in this disclosure.
  • the bus interface provides an interface.
  • the transceiver 710 may be a plurality of components, That is, it includes a transmitter and a receiver, and provides a unit for communicating with various other devices on a transmission medium, and these transmission media include transmission media such as wireless channels, wired channels, and optical cables.
  • the user interface 730 can also be an interface that can be connected to external or internal devices, and the connected devices include but are not limited to a keypad, a display, a speaker, a microphone, a joystick, etc.
  • the processor 700 is responsible for managing the bus architecture and general processing, and the memory 720 can store data used by the processor 700 when performing operations.
  • the processor 700 may be a CPU, an ASIC, an FPGA or a CPLD, and the processor may also adopt a multi-core architecture.
  • the processor 700 calls the computer program stored in the memory 720 to execute any of the methods provided in the embodiments of the present disclosure according to the obtained executable instructions, for example:
  • one or more target downlink beams are predicted, and the target downlink beams are used for information transmission between network equipment and terminals.
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to the prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate the first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • a fourth pattern where the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • the fifth pattern is used to indicate the beam sequence satisfied by the RS configuration belonging to the measurement set. Number arrangement pattern.
  • the method further includes:
  • FIG8 is a schematic diagram of the structure of a network device provided in an embodiment of the present disclosure.
  • the network device includes a memory 820, a transceiver 810, and a processor 800; wherein the processor 800 and the memory 820 may also be arranged physically separately.
  • the memory 820 is used to store computer programs; the transceiver 810 is used to send and receive data under the control of the processor 800.
  • the transceiver 810 is used to receive and send data under the control of the processor 800 .
  • the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 800 and various circuits of memory represented by memory 820 are linked together.
  • the bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described in this disclosure.
  • the bus interface provides an interface.
  • the transceiver 810 may be a plurality of components, including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which may include a wireless channel, a wired channel, an optical cable, and other transmission media.
  • the processor 800 is responsible for managing the bus architecture and general processing, and the memory 820 can store data used by the processor 800 when performing operations.
  • the processor 800 may be a CPU, an ASIC, an FPGA or a CPLD, and the processor may also adopt a multi-core architecture.
  • the processor 800 calls the computer program stored in the memory 820 to execute any of the methods provided in the embodiments of the present disclosure according to the obtained executable instructions, for example:
  • RS configuration is used to train or update an artificial intelligence or machine learning AI/ML model, where the AI/ML model is used to predict one or more target downlink beams, where the target downlink beams are used to transmit information between a network device and a terminal;
  • the RS configuration includes an RS configuration identifier, a second RS configuration belonging to the prediction set, and a measurement set pattern, where the measurement set pattern is used to indicate the first RS configuration belonging to the measurement set;
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • a fourth pattern where the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • the fifth pattern is used to indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the method further comprises:
  • the second information includes one or more of the following:
  • the number of input beams and the number of output beams of the AI/ML model used to predict the target downlink beam interest are the number of input beams and the number of output beams of the AI/ML model used to predict the target downlink beam interest.
  • first communication device can implement all the method steps implemented in the above-mentioned method embodiment, and can achieve the same technical effect.
  • the parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.
  • FIG. 9 is a schematic diagram of a structure of a downlink beam prediction device provided in an embodiment of the present disclosure.
  • the device is applied to a first communication device. As shown in FIG. 9 , the device includes:
  • a first receiving unit 900 is configured to receive first information, where the first information includes one or more of a data set, a reference signal RS configuration, and a beam description information;
  • a first model unit 910 is used to train or update an artificial intelligence or machine learning AI/ML model for downlink beam prediction based on the first information
  • the first prediction unit 920 is used to predict one or more target downlink beams based on the trained or updated AI/ML model, and the target downlink beams are used for information transmission between the first communication device and the second communication device.
  • receiving the first information includes:
  • the first information sent by the second communication device is received.
  • the dataset contains one or more of the following:
  • One or more data set samples each data set sample including multiple beam identifiers, RS measurement results of the beam corresponding to each beam identifier, and one or more beam identifiers used as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:
  • a first pattern where the first pattern is used to indicate whether each second RS configuration belonging to the prediction set belongs to the measurement set;
  • a second pattern where the second pattern is used to indicate an RS identifier corresponding to an RS configuration belonging to a measurement set
  • a third pattern is used to indicate the RS sequence number arrangement rule satisfied by the RS configuration belonging to the measurement set;
  • a fourth pattern where the fourth pattern is used to indicate a beam identifier corresponding to an RS configuration belonging to a measurement set
  • the fifth pattern is used to indicate the beam number arrangement rule satisfied by the RS configuration belonging to the measurement set.
  • the beam description information includes one or more of the following:
  • the first information is associated with one or more of an application scenario, a network device configuration, and a model function.
  • the apparatus further includes a first sending unit, configured to send second information to the second communication device, where the second information includes one or more of the following:
  • FIG. 10 is a second structural diagram of a downlink beam prediction device provided in an embodiment of the present disclosure. The device is applied to a second communication device, as shown in FIG10, and includes:
  • the first determination unit 1000 is used to determine first information, where the first information includes one or more of a data set, an RS configuration, and a beam description information; the first information is used to train or update an artificial intelligence or machine learning AI/ML model, where the AI/ML model is used to predict one or more target downlink beams, and the target downlink beam is used for information transmission between the second communication device and the first communication device;
  • the second sending unit 1010 is configured to send first information to the first communication device.
  • the dataset contains one or more of the following:
  • Each data set sample including multiple beam identifiers, a reference signal measurement result of a beam corresponding to each beam identifier, and one or more beam identifiers serving as prediction labels.
  • the RS configuration includes one or more of the following:
  • the measurement set pattern is used to indicate a first RS configuration belonging to the measurement set
  • the first RS configuration is used to obtain RS measurement results as model input; the second RS configuration is used to determine one or more downlink beams as prediction labels.
  • the measurement set pattern includes one or more of the following:

Landscapes

  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Radio Transmission System (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本公开实施例提供一种下行波束预测方法、设备、装置及存储介质,该方法包括:第一通信设备接收第一信息,所述第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;基于所述第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述第一通信设备与第二通信设备之间的信息传输。

Description

下行波束预测方法、设备、装置及存储介质
相关申请的交叉引用
本申请要求于2022年09月30日提交的申请号为202211215784.0,发明名称为“下行波束预测方法、设备、装置及存储介质”的中国专利申请的优先权,其通过引用方式全部并入本文。
技术领域
本公开涉及无线通信技术领域,尤其涉及一种下行波束预测方法、设备、装置及存储介质。
背景技术
在新无线(New Radio,NR)下行波束管理场景中,确定最优下行波束的方式,需要基站循环在不同方向发送Tx beam(发送波束),终端(User Equipment,UE)使用Rx beam(接收波束)接收Tx beam,并测量所有Tx beam上发送的信道状况信息参考信号(Channel State Information Reference Signal,CSI-RS)或同步信号块(Synchronization Signal Block,SSB)信号,选出接收性能(如参考信号接收功率(Reference Signal Received Power,RSRP))最好的波束(best beam),用于后续基站使用该下行波束给终端传输信息。
相关技术的实现方案,CSI-RS或SSB信号需要在所有Tx beam上发送,资源消耗较大。并且,终端需要分别测量所有Tx beam上发送的CSI-RS或SSB信号,因此终端实现较为复杂,并且测量开销较大。
发明内容
针对相关技术存在的问题,本公开实施例提供一种下行波束预测方法、设备、装置及存储介质。
第一方面,本公开实施例提供一种下行波束预测方法,应用于第一通信 设备,包括:
接收第一信息,所述第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
基于所述第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述第一通信设备与第二通信设备之间的信息传输。
可选地,所述接收第一信息,包括:
接收所述第二通信设备发送的所述第一信息。
可选地,所述数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
可选地,所述RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,所述预测出一个或多个目标下行波束之后,所述方法还包括:
向所述第二通信设备发送第二信息,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的数据集的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的波束描述信息的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第二方面,本公开实施例还提供一种下行波束预测方法,应用于第二通信设备,包括:
确定第一信息,所述第一信息包括数据集、RS配置、波束描述信息中的一种或多种;所述第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行 波束用于所述第二通信设备与第一通信设备之间的信息传输;
向所述第一通信设备发送所述第一信息。
可选地,所述数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
可选地,所述RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,所述方法还包括:
接收所述第一通信设备发送的第二信息;
基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述第一通信设备之间进行信息传输的目标下行波束;
其中,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的数据集的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的波束描述信息的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第三方面,本公开实施例还提供一种下行波束预测方法,应用于终端,包括:
接收网络设备发送的参考信号RS配置;
基于所述RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述网络设备与所述终端之间的信息传输。
可选地,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS 配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述预测出一个或多个目标下行波束之后,所述方法还包括:
向所述网络设备发送第二信息,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第四方面,本公开实施例还提供一种下行波束预测方法,应用于网络设备,包括:
确定参考信号RS配置,所述RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述网络设备与终端之间的信息传输;
向终端发送所述RS配置。
可选地,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述方法还包括:
接收终端发送的第二信息;
基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述终端之间进行信息传输的目标下行波束;
其中,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第五方面,本公开实施例还提供一种第一通信设备,包括存储器,收发 机,处理器;
存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
接收第一信息,所述第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
基于所述第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述第一通信设备与第二通信设备之间的信息传输。
可选地,所述接收第一信息,包括:
接收所述第二通信设备发送的所述第一信息。
可选地,所述数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
可选地,所述RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,所述预测出一个或多个目标下行波束之后,所述操作还包括:
向所述第二通信设备发送第二信息,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的数据集的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的波束描述信息的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第六方面,本公开实施例还提供一种第二通信设备,包括存储器,收发机,处理器;
存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收 发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
确定第一信息,所述第一信息包括数据集、RS配置、波束描述信息中的一种或多种;所述第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述第二通信设备与第一通信设备之间的信息传输;
向所述第一通信设备发送所述第一信息。
可选地,所述数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
可选地,所述RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,所述操作还包括:
接收所述第一通信设备发送的第二信息;
基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述第一通信设备之间进行信息传输的目标下行波束;
其中,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的数据集的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的波束描述信息的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第七方面,本公开实施例还提供一种终端,包括存储器,收发机,处理器;
存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
接收网络设备发送的参考信号RS配置;
基于所述RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述网络设备与所述终端之间的信息传输。
可选地,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述预测出一个或多个目标下行波束之后,所述操作还包括:
向所述网络设备发送第二信息,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第八方面,本公开实施例还提供一种网络设备,包括存储器,收发机,处理器;
存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
确定参考信号RS配置,所述RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述网络设备与终端之间的信息传输;
向终端发送所述RS配置。
可选地,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述操作还包括:
接收终端发送的第二信息;
基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述终端之间进行信息传输的目标下行波束;
其中,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第九方面,本公开实施例还提供一种下行波束预测装置,应用于第一通信设备,包括:
第一接收单元,用于接收第一信息,所述第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
第一模型单元,用于基于所述第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
第一预测单元,用于基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述第一通信设备与第二通信设备之间的信息传输。
可选地,所述接收第一信息,包括:
接收所述第二通信设备发送的所述第一信息。
可选地,所述数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
可选地,所述RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,所述预测出一个或多个目标下行波束之后,所述装置还包括:
第一发送单元,用于向所述第二通信设备发送第二信息,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的数据集的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的波束描述信息的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第十方面,本公开实施例还提供一种下行波束预测装置,应用于第二通信设备,包括:
第一确定单元,用于确定第一信息,所述第一信息包括数据集、RS配置、波束描述信息中的一种或多种;所述第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述第二通信设备与第一通信设备之间的信息传输;
第二发送单元,用于向所述第一通信设备发送所述第一信息。
可选地,所述数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
可选地,所述RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,所述装置还包括:
第二接收单元,用于接收所述第一通信设备发送的第二信息;
基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述第一通信设备之间进行信息传输的目标下行波束;
其中,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的数据集的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的波束描述信息的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数 量信息。
第十一方面,本公开实施例还提供一种下行波束预测装置,应用于终端,包括:
第三接收单元,用于接收网络设备发送的参考信号RS配置;
第二模型单元,用于基于所述RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
第二预测单元,用于基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述网络设备与所述终端之间的信息传输。
可选地,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,所述预测出一个或多个目标下行波束之后,所述装置还包括:
第三发送单元,用于向所述网络设备发送第二信息,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第十二方面,本公开实施例还提供一种下行波束预测装置,应用于网络设备,包括:
第二确定单元,用于确定参考信号RS配置,所述RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述网络设备与终端之间的信息传输;
第四发送单元,用于向终端发送所述RS配置。
可选地,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,所述测量集合图样包括以下一种或多种:
第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波 束序号排列规律。
可选地,所述装置还包括:
第四接收单元,用于接收终端发送的第二信息;
基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述终端之间进行信息传输的目标下行波束;
其中,所述第二信息中包含以下一项或多项:
所述目标下行波束对应的波束标识;
所述目标下行波束对应的RS标识;
用于预测所述目标下行波束的AI/ML模型的标识;
用于预测所述目标下行波束的RS配置的标识;
用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
第十三方面,本公开实施例还提供一种计算机可读存储介质,所述计算机可读存储介质存储有计算机程序,所述计算机程序用于使计算机执行如上所述第一方面所述的下行波束预测方法,或执行如上所述第二方面所述的下行波束预测方法,或执行如上所述第三方面所述的下行波束预测方法,或执行如上所述第四方面所述的下行波束预测方法。
第十四方面,本公开实施例还提供一种通信设备,所述通信设备中存储有计算机程序,所述计算机程序用于使通信设备执行如上所述第一方面所述的下行波束预测方法,或执行如上所述第二方面所述的下行波束预测方法,或执行如上所述第三方面所述的下行波束预测方法,或执行如上所述第四方面所述的下行波束预测方法。
第十五方面,本公开实施例还提供一种处理器可读存储介质,所述处理器可读存储介质存储有计算机程序,所述计算机程序用于使处理器执行如上所述第一方面所述的下行波束预测方法,或执行如上所述第二方面所述的下行波束预测方法,或执行如上所述第三方面所述的下行波束预测方法,或执行如上所述第四方面所述的下行波束预测方法。
第十六方面,本公开实施例还提供一种芯片产品,所述芯片产品中存储 有计算机程序,所述计算机程序用于使芯片产品执行如上所述第一方面所述的下行波束预测方法,或执行如上所述第二方面所述的下行波束预测方法,或执行如上所述第三方面所述的下行波束预测方法,或执行如上所述第四方面所述的下行波束预测方法。
本公开实施例提供的下行波束预测方法、设备、装置及存储介质,通过向第一通信设备发送数据集、RS配置、波束描述信息中的一种或多种,第一通信设备可以基于这些信息来对用于下行波束预测的AI/ML模型进行训练或更新,并基于训练或更新后的AI/ML模型预测出一个或多个目标下行波束,从而可以有效节省RS发送资源、UE测量开销和降低UE测量时延。
附图说明
为了更清楚地说明本公开实施例或相关技术中的技术方案,下面将对实施例或相关技术描述中所需要使用的附图作一简单地介绍,显而易见地,下面描述中的附图是本公开的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1是本公开实施例提供的下行波束预测方法的流程示意图之一;
图2是本公开实施例提供的下行波束预测方法的流程示意图之二;
图3是本公开实施例提供的下行波束预测方法的流程示意图之三;
图4是本公开实施例提供的下行波束预测方法的流程示意图之四;
图5是本公开实施例提供的第一通信设备的结构示意图;
图6是本公开实施例提供的第二通信设备的结构示意图;
图7是本公开实施例提供的终端的结构示意图;
图8是本公开实施例提供的网络设备的结构示意图;
图9是本公开实施例提供的下行波束预测装置的结构示意图之一;
图10是本公开实施例提供的下行波束预测装置的结构示意图之二;
图11是本公开实施例提供的下行波束预测装置的结构示意图之三;
图12是本公开实施例提供的下行波束预测装置的结构示意图之四。
具体实施方式
本公开实施例中术语“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这三种情况。字符“/”一般表示前后关联对象是一种“或”的关系。
本公开实施例中术语“多个”是指两个或两个以上,其它量词与之类似。
下面将结合本公开实施例中的附图,对本公开实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本公开一部分实施例,并不是全部的实施例。基于本公开中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本公开保护的范围。
为了便于更加清晰地理解本公开各实施例的技术方案,首先对本公开各实施例相关的一些技术内容进行介绍。
1、三种计算下行(Downlink,DL)最优波束(best beam)的方式。
一种是测量所有下行波束对(DL beam pair)(基站的一个Tx beam和UE的一个Rx beam组成的一个beam pair)的RSRP,得到最大的RSRP对应的beam pair即为最优beam pair,并告知基站,后续基站采用该beam pair中的Tx beam给UE发送信息,UE采用该beam pair中的Rx beam接收信息。
另一种是UE固定或选择一个最好的DL Rx beam,接收测量基站发送的所有DL Tx beam的接收功率RSRP,得到最大的RSRP对应的Tx beam即为最优Tx beam,并告知基站,后续基站采用该Tx beam给UE发送信息。
再一种是基站固定或选择一个最好的DL Tx beam,UE采用所有DL Rx beam接收测量基站发送的该DL Tx beam的接收功率RSRP,得到最大的RSRP对应的Rx beam即为最优Rx beam,后续基站采用该Tx beam给UE发送信息时,UE采用该最优Rx beam接收。
如上所述,对于DL beam pair测量方式,UE需要使用全部的Rx beam接收基站发送的每个Tx beam的CSI-RS/SSB进行测量。例如,UE有4个Rx beam,基站有32个Tx beam,如果UE采用全部Rx beam分别接收每个Tx beam,则需要测量4*32=128个CSI-RS/SSB,即每个Rx beam都需要测量所有的Tx  beam,才能计算出,最好的beam pair。对于DL Tx beam测量方式,UE固定或选择一个最好的Rx beam,测量基站发送的所有Tx beam,需要测量32个Tx beam,才能计算出,最好的Tx beam。对于DL Rx beam测量方式,也存在同样的问题。
考虑到上述计算方式需要测量全部的Rx beam和Tx beam,用于测量的参考信号占用传输资源较大,UE测量复杂度高,测量消耗较大,UE测量时延较高。本公开提出利用人工智能(Artificial Intelligence,AI)或机器学习(Machine Learning,AI)技术对下行波束进行预测,对于DL beam pair测量方式,只需要测量一部分DL beam pair,例如只发送32个Tx beam中的8个,则UE只需要测量4*8=32个beam pair上的CSI-RS/SSB,就可以准确预测出32个Tx beam和4个Rx beam组成的128个beam pair中接收性能最好的beam pair;对于DL Tx beam测量方式,UE固定或选择一个最好的Rx beam,只需要测量8个Tx beam,就可以准确预测出32个Tx beam中接收性能最好的Tx beam。当然本公开还可以用于其他场景,例如测量其他便于测量的发送SSB的宽beam,或其他频段的beam,从而预测发送CSI-RS的窄beam,或者预测高频率的beam中的best beam;又例如利用以前测量过的beam测量结果,预测出基站未来发送的Tx beam或beam pair中接收性能最好的beam(pair);又例如,DL Rx beam预测,基站固定或选择一个最好的Tx beam,连续发送参考信号(Reference Signal,RS),UE使用不同的DL Rx beam接收测量RS,通过测量结果,预测出最好的Rx beam。从而可以节省RS发送资源、UE测量开销和降低UE测量时延。
图1为本公开实施例提供的下行波束预测方法的流程示意图之一,如图1所示,该方法应用于第一通信设备,包括:
步骤100、接收第一信息,第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种。
步骤101、基于第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新。
步骤102、基于训练或更新后的AI/ML模型,预测出一个或多个目标下 行波束,目标下行波束用于第一通信设备与第二通信设备之间的信息传输。
具体地,第一通信设备可以是终端或网络设备(例如基站),相应地,第二通信设备可以是与第一通信设备之间进行信息传输的网络设备或终端。
例如,在第一通信设备为终端的情况下,第二通信设备可以为网络设备。终端可以接收第一信息(网络设备或第三方设备发送的),根据接收到的第一信息,终端可以对用于下行波束预测的AI/ML模型进行训练或更新,从而预测出一个或多个目标下行波束,预测出的目标下行波束用于终端和网络设备之间的信息传输。
在第一通信设备为网络设备的情况下,第二通信设备可以为终端。网络设备可以接收第一信息(终端或第三方设备发送的),根据接收到的第一信息,网络设备可以对用于下行波束预测的AI/ML模型进行训练或更新,从而预测出一个或多个目标下行波束,预测出的目标下行波束用于终端和网络设备之间的信息传输。
为了避免终端预测出的目标下行波束无法被网络设备识别,或者网络设备预测出的目标下行波束无法被终端识别的问题,本公开实施例中,第二通信设备可以向第一通信设备发送第一信息,第一信息可以包括数据集、参考信号RS配置、波束描述信息中的一种或多种。第一通信设备接收到第一信息后,可以根据第一信息对用于下行波束预测的AI/ML模型进行训练或更新。其中,对模型进行更新包括对模型进行微调(Fine-tuning),即第一通信设备接收到第一信息后,可以根据第一信息对用于下行波束预测的AI/ML模型进行微调。该用于下行波束预测的AI/ML模型的功能可以包括:空域预测,频域预测,时域预测,下行beam pair预测,下行Tx beam预测,下行Rx beam预测等等。空域预测是指测量少量的波束,或者其他参考信号类型的波束,预测大量波束中的最优波束;频域预测是指测量频率1上的波束,预测另一个频率2上的波束中的最优波束;时域预测是指测量现在时刻的波束,预测未来时刻的最优波束。
其中,RS测量结果可以是RSRP等任意可以表示下行波束接收性能的测量结果,具体形式不做限定。
得到训练或更新后的模型后,第一通信设备可以根据RS测量结果,通过训练或更新后的模型预测出一个或多个目标下行波束,预测出的目标下行波束可以是Tx beam,或者Rx beam,或者beam pair,其用于第一通信设备和第二通信设备之间的信息传输。
第一通信设备接收到的第一信息可以是由第二通信设备或者第三方设备预先定义的。
其中,在第三方设备预先定义第一信息的情况下,一种实施方式中,第三方设备可以将第一信息发送给第一通信设备和第二通信设备;或者,第三方设备可以将第一信息发送给第二通信设备,第二通信设备接收第一信息后,再发送给第一通信设备。
第一信息可以是通过Uu接口的控制信道或者数据信道传送的,也可以通过其他接口、其他信道来传送,具体情形不做限制。
本公开实施例提供的下行波束预测方法,通过向第一通信设备发送数据集、RS配置、波束描述信息中的一种或多种,第一通信设备可以基于这些信息来对用于下行波束预测的AI/ML模型进行训练或更新,并基于训练或更新后的AI/ML模型预测出一个或多个目标下行波束,从而可以有效节省RS发送资源、UE测量开销和降低UE测量时延。
可选地,接收第一信息,包括:
接收第二通信设备发送的第一信息。
具体地,当第一通信设备为终端时,终端接收到的第一信息可以是由网络设备发送给终端的。一种可能的实现方式中,网络设备预先定义并将第一信息发送至终端。例如,网络设备预先定义一个RS配置,并将RS配置发送至终端,终端根据网络设备发送的RS配置来对用于下行波束预测的AI/ML模型进行训练或更新。
当第一通信设备为网络设备时,网络设备接收到的第一信息可以是终端发送的。一种可能的实现方式中,终端发送的第一信息可以包括数据集和/或波束描述信息,不包括RS配置。终端预先定义第一信息并将第一信息发送至网络设备。例如,终端预先定义一个数据集,并将数据集发送至网络设备, 网络设备根据终端发送的数据集来对用于下行波束预测的AI/ML模型进行训练或更新。
可选地,数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
具体地,数据集可以是一个或多个,数据集标识代表了不同的数据集,不同的数据集对应的数据集标识不同。
数据集样本中包括了多个波束标识(beam ID),这些beam ID对应的波束的RS测量结果,以及一个或多个作为预测标签的beam ID。可以理解的是,终端或网络设备或第三方设备预先定义数据集的过程中,将每个下行波束对应一个beam ID。一个下行波束可以是一个Tx beam,或者一个Rx beam,或者一个beam pair。
在对用于下行波束预测的AI/ML模型进行训练或更新的过程中,第一通信设备可以将数据集样本中的多个beam ID对应的下行波束的RS测量结果作为模型的输入;将数据集样本中的一个或多个作为预测标签的beam ID作为模型的输出标签。
可选地,RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
具体地,RS配置可以是一个或多个,RS配置标识代表了不同的RS配置,不同的RS配置对应的RS配置标识不同。RS配置可以包括RS的种类(例如CSI-RS、SSB、相位跟踪参考信号(Phase-tracking Reference Signal,PT-RS)、小区参考信号(Cell Reference Signal,CRS)、解调参考信号(Demodulation  Reference Signal,DMRS)等)、RS发送资源和RS标识(RS ID)等。其中,RS ID为RS指示符(RS Indication),例如,CSI-RS的指示符为CRI,SSB的指示符为SSBRI。
测量集合指的是集合中的RS配置或者波束,用于获取RS测量结果作为模型的输入;预测集合指的是集合中的RS配置或者波束,用于确定一个或多个下行波束作为预测标签。
例如,第一通信设备测量该测量集合中beam上发送的RS,得到RS的测量结果,并将RS测量结果作为模型的输入;第一通信设备通过测量该预测集合中beam上发送的RS,用非AI/ML的方法计算出一个或多个best beam,并将这一个或多个best beam作为模型的输出标签。
由于第一通信设备接收属于测量集合的第一RS配置和属于预测集合的第二RS配置之后的行为有区别,因此,需要区分不同的RS配置分别属于哪个集合。
一种实施方式中,可以用RS配置的标识ID区分,例如ID=0是测量集合的配置,ID=1是预测集合的配置;或者ID是奇数代表测量集合的配置,ID是偶数代表预测集合的配置。
在测量集合为预测集合子集的情况下,可以只对预测集合进行RS配置,RS配置中可以包括属于预测集合的第二RS配置和测量集合图样,通过测量集合图样来指示预测集合中属于测量集合的第一RS配置。
可选地,测量集合图样包括以下一种或多种:
(1)第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合。
具体地,可以通过第一图样指示属于预测集合的每个第二RS配置是否属于测量集合,进而指示预测集合中属于测量集合的第一RS配置。
一种实施方式中,第一图样可以是在每个RS资源配置中增加1比特,例如当该比特=1时,表示该RS既属于预测集合,也属于测量集合;当该比特=0时,表示该RS只属于预测集合。
一种实施方式中,第一图样可以是与RS资源配置个数相等的字符串或 位图(bitmap),例如,当对应位置的比特=1时,表示该RS既属于预测集合,也属于测量集合;当对应位置的比特=0时,表示该RS只属于预测集合,那么01011可以表示预测集合中的第2、4、5个RS为测量集合。
(2)第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识。
具体地,可以通过第二图样指示属于测量集合的RS配置所对应的RS标识,进而指示预测集合中属于测量集合的第一RS配置。
例如,第二图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个RS为测量集合。
(3)第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律。
具体地,可以通过第三图样指示属于测量集合的RS配置所满足的RS序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,可以配置第三图样为:预测集合中的偶数序号的RS为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。
(4)第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通过第四图样指示属于测量集合的RS配置所满足的波束序号,进而指示预测集合中属于测量集合的第一RS配置。
例如,第四图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个下行波束为测量集合。
(5)第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通过第五图样指示属于测量集合的RS配置所满足的波束序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,配置第五图样为:预测集合中的偶数序号的下行波束为测量集合。 可以用字符串表示不同的规律,例如2比特可以表示4种规律。
对于测量集合为预测集合的子集的情况,在RS配置时可以采用测量集合图样指示的方式,从而节省用于配置的比特开销。
可选地,波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
具体地,第一通信设备还可以基于波束描述信息对用于下行波束预测的AI/ML模型进行训练或更新,波束描述信息中可以包含beam ID、RS ID、网络设备的天线配置信息、波束的角度信息、波束的宽度信息中的一项或多项。可选地,波束描述信息可以作为辅助信息与数据集和/或RS配置一起使用,对用于下行波束预测的AI/ML模型进行训练或更新,以及后续预测流程。
一个波束描述信息中包括一个或多个下行波束的描述信息。一个下行波束可以是一个Tx beam,或者一个Rx beam,或者一个beam pair。可选地,在一个波束描述信息中包括多个下行波束的描述信息时,可以按照列表的方式,分别对多个下行波束进行描述。
可以理解的是,终端或网络设备或第三方设备预先定义波束描述信息的过程中,将每个下行波束对应一个beam ID。RS ID为在该下行波束上发送的RS的指示符,例如,CSI-RS的指示符为CRI,SSB的指示符为SSBRI。对一个下行波束进行描述时,还可以用该下行波束的网络设备的天线配置信息,该下行波束的角度信息和该下行波束的宽度信息来对该下行波束进行描述。
可选地,预测出一个或多个目标下行波束之后,该方法还包括:
向第二通信设备发送第二信息。
具体地,在第一通信设备预测出一个或多个目标下行波束后,第一通信设备可以向第二通信设备发送第二信息,第二信息用于将第一通信设备预测出的一个或多个目标下行波束的结果告知第二通信设备。
第二通信设备接收到第一通信设备发送的第二信息后,可以根据第一信息和第二信息得知第一通信设备预测出的一个或多个目标下行波束,进而第一通信设备和第二通信设备之间可以在目标下行波束(即best beam)上传输信息,提升系统的传输性能。
其中,第二信息中包含以下一项或多项:
(1)目标下行波束对应的波束标识。
第二信息中可以包括目标下行波束对应的beam ID,从而第二通信设备可以根据目标下行波束对应的beam ID得知第一通信设备通过AI/ML模型推测出的目标下行波束。
因为第一信息是第二通信设备定义的或者是第三方设备定义但第二通信设备已知的,所以beam ID和物理波束的映射关系是第二通信设备已知的,即第一通信设备使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的beam ID,可以被第二通信设备识别,并对应到物理实际的目标下行波束,从而第一通信设备和第二通信设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,第一通信设备通过AI/ML模型只推测出一个目标下行波束,第一通信设备可以将该目标下行波束对应的beam ID发送给第二通信设备。
一种实施方式中,第一通信设备通过AI/ML模型推测出多个目标下行波束,第一通信设备可以选择其中一个目标下行波束对应的beam ID将其发送给第二通信设备;或者,第一通信设备可以将多个目标下行波束对应的beam ID发送给第二通信设备,由第二通信设备选择其中一个。
(2)目标下行波束对应的RS标识。
第二信息中可以包括目标下行波束对应的RS ID,从而第二通信设备可以根据目标下行波束对应的RS ID得知第一通信设备通过AI/ML模型推测出的目标下行波束。
因为第一信息是第二通信设备定义的或者是第三方设备定义但第二通信设备已知的,所以RS ID和物理波束的映射关系是第二通信设备已知的,即 第一通信设备使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的RS ID,可以被第二通信设备识别,并对应到物理实际的目标下行波束,从而第一通信设备和第二通信设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,第一通信设备通过AI/ML模型只推测出一个目标下行波束,第一通信设备可以将该目标下行波束对应的RS ID发送给第二通信设备。
一种实施方式中,第一通信设备通过AI/ML模型推测出多个目标下行波束,第一通信设备可以选择其中一个目标下行波束对应的RS ID将其发送给第二通信设备;或者,第一通信设备可以将多个目标下行波束对应的RS ID发送给第二通信设备,由第二通信设备选择其中一个。
(3)用于预测目标下行波束的AI/ML模型的标识。
在训练或更新得到多个不同的AI/ML模型的情况下,每个模型都可以对应一个模型标识(model ID),此时,第一通信设备不仅需要发送预测得到的目标下行波束对应的beam ID或者RS ID,还需要发送预测得到目标下行波束对应的model ID,从而第二通信设备可以根据模型的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
(4)用于预测目标下行波束的数据集的标识。
在有多个不同的数据集用于训练或更新得到多个不同的AI/ML模型的情况下,第二信息中可以包括用于预测目标下行波束的数据集的标识,从而第二通信设备可以根据用于预测目标下行波束的数据集的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
例如,如果数据集和训练、更新的模型是一一对应的,那么第一通信设备也可以发送数据集的标识,代表上述model ID,第二通信设备通过数据集的标识,就可以知道第一通信设备使用该模型输出的beam ID和物理beam的映射关系。
(5)用于预测目标下行波束的RS配置的标识。
在有多个不同的RS配置用于训练或更新得到多个不同的AI/ML模型的 情况下,第二信息中可以包括用于预测目标下行波束的RS配置的标识,从而第二通信设备可以根据用于预测目标下行波束的RS配置的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
例如,如果RS配置和训练或更新的模型是一一对应的,那么第一通信设备也可以发送RS配置的标识,代表上述model ID,第二通信设备通过RS配置的标识,就可以知道第一通信设备使用该model输出的RS ID和物理beam的映射关系。
(6)用于预测目标下行波束的波束描述信息的标识。
在有多个不同的波束描述信息用于训练或更新得到多个不同的AI/ML模型的情况下,第二信息中可以包括用于预测目标下行波束的波束描述信息的标识,从而第二通信设备可以根据用于预测目标下行波束的波束描述信息的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
例如,如果波束描述信息和训练或更新的模型是一一对应的,那么第一通信设备也可以发送波束描述信息的标识,代表上述model ID,第二通信设备通过RS配置的标识,就可以知道第一通信设备使用该model输出的beam ID或RS ID和物理beam的映射关系。
(7)用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
在训练或更新得到多个不同的AI/ML模型,且每个AI/ML模型所输入波束数量与输出波束数量信息不同的情况下,第二信息中可以包括用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息,从而第二通信设备可以根据用于预测目标下行波束的输入波束数量与输出波束数量信息得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
其中,输入波束数量可以是模型的输入Tx beam数,或者模型的输入Rx beam数,或者模型的输入beam pair数等;输出波束数量可以是模型的输出Tx beam数,或者模型的输出Rx beam数,或者模型的输出beam pair数等。
例如,如果第一通信设备针对每个{输入beam数,输出beam数}对进行训练或更新得到一个AI/ML model,那么{输入beam数,输出beam数}对的 信息可以代表model。即,第一通信设备发送{输入beam数,输出beam数}对的信息可以对应model ID,第二通信设备通过{输入beam数,输出beam数}对信息,就可以知道第一通信设备使用该model的输出的beam ID或RS ID和物理beam的映射关系。
可选地,第一信息可以与应用场景、网络设备配置、模型功能中的一项或多项相关联。
具体地,不同的数据集、不同的RS配置和不同的beam描述信息,分别可以针对不同的应用场景,或者基站配置,或者AI功能。
应用场景包括:城市,农村,室内,室外,高速公路,高铁,城市宏小区(Uma),城市微小区(Umi)等等;基站配置包括:基站天线配置,波束配置,参考信号配置等等;AI功能,主要包括空域预测、频域预测、时域预测、下行beam pair预测、下行Tx beam预测、下行Rx beam预测等等,其中,空域预测是指测量少量的beam,或者其他参考信号类型的beam,预测大量beam中的best beam;频域预测是指测量频率1上的beam,预测另一个频率2上的beam中的best beam;时域预测是指测量现在时刻的beam,预测未来时刻的best beam。
通过第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联,可以得到不同场景下使用的用于下行波束预测的AI/ML模型。
图2为本公开实施例提供的下行波束预测方法的流程示意图之二,如图2所示,该方法应用于第二通信设备,包括:
步骤200、确定第一信息,第一信息包括数据集、RS配置、波束描述信息中的一种或多种;第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,AI/ML模型用于预测一个或多个目标下行波束,目标下行波束用于第二通信设备与第一通信设备之间的信息传输。
步骤201、向第一通信设备发送第一信息。
具体地,第一通信设备可以是终端或网络设备(例如基站),相应地,第二通信设备可以是与第一通信设备之间进行信息传输的网络设备或终端。
例如,在第一通信设备为终端的情况下,第二通信设备可以为网络设备。 终端可以接收第一信息(网络设备或第三方设备发送的),根据接收到的第一信息,终端可以对用于下行波束预测的AI/ML模型进行训练或更新,从而预测出一个或多个目标下行波束,预测出的目标下行波束用于终端和网络设备之间的信息传输。
在第一通信设备为网络设备的情况下,第二通信设备为终端。网络设备可以接收第一信息(终端或第三方设备发送的),根据接收到的第一信息,网络设备可以对用于下行波束预测的AI/ML模型进行训练或更新,从而预测出一个或多个目标下行波束,预测出的目标下行波束用于终端和网络设备之间的信息传输。
为了避免终端预测出的目标下行波束无法被网络设备识别,或者网络设备预测出的目标下行波束无法被终端识别的问题,本公开实施例中,第二通信设备可以向第一通信设备发送第一信息,第一信息可以包括数据集、参考信号RS配置、波束描述信息中的一种或多种。
第二通信设备确定第一信息可以是第二通信设备自己预先定义的;或者,由第三方设备预先定义并发送给第二通信设备的。
第二通信设备确定了第一信息后,可以将第一信息发送给第一通信设备。第一通信设备接收到第一信息后,可以根据第一信息对用于下行波束预测的AI/ML模型进行训练或更新。其中,对模型进行更新包括对模型进行微调,即第一通信设备接收到第一信息后,可以根据第一信息对用于下行波束预测的AI/ML模型进行微调。该用于下行波束预测的AI/ML模型的功能可以包括:空域预测,频域预测,时域预测,下行beam pair预测,下行Tx beam预测,下行Rx beam预测等等。空域预测是指测量少量的波束,或者其他参考信号类型的波束,预测大量波束中的最优波束;频域预测是指测量频率1上的波束,预测另一个频率2上的波束中的最优波束;时域预测是指测量现在时刻的波束,预测未来时刻的最优波束。
其中,RS测量结果可以是RSRP等任意可以表示下行波束接收性能的测量结果,具体形式不做限定。
得到训练或更新后的模型后,第一通信设备可以根据RS测量结果,通 过训练或更新后的模型预测出一个或多个目标下行波束,预测出的目标下行波束可以是Tx beam,或者Rx beam,或者beam pair,用于第一通信设备和第二通信设备之间的信息传输。
第一信息可以是通过Uu接口的控制信道或者数据信道传送的,也可以通过其他接口、其他信道来传送,具体情形不做限制。
本公开实施例提供的下行波束预测方法,通过向第一通信设备发送数据集、RS配置、波束描述信息中的一种或多种,第一通信设备可以基于这些信息来对用于下行波束预测的AI/ML模型进行训练或更新,并基于训练或更新后的AI/ML模型预测出一个或多个目标下行波束,从而可以有效节省RS发送资源、UE测量开销和降低UE测量时延。
可选地,数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
具体地,数据集可以是一个或多个,数据集标识代表了不同的数据集,不同的数据集对应的数据集标识不同。
数据集样本中包括了多个beam ID,这些beam ID对应的波束的RS测量结果,以及一个或多个作为预测标签的beam ID。可以理解的是,终端或网络设备或第三方设备预先定义数据集的过程中,将每个下行波束对应一个beam ID。一个下行波束可以是一个Tx beam,或者一个Rx beam,或者一个beam pair。
在对用于下行波束预测的AI/ML模型进行训练或更新的过程中,第一通信设备可以将数据集样本中的多个beam ID对应的下行波束的RS测量结果作为模型的输入;将数据集样本中的一个或多个作为预测标签的beam ID作为模型的输出标签。
可选地,RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
具体地,RS配置可以是一个或多个,RS配置标识代表了不同的RS配置,不同的RS配置对应的RS配置标识不同。RS配置可以包括RS的种类(例如CSI-RS、SSB、PT-RS、CRS、DMRS等)、RS发送资源和RS ID等。其中,RS ID为RS指示符,例如,CSI-RS的指示符为CRI,SSB的指示符为SSBRI。
测量集合指的是集合中的RS配置或者波束,用于获取RS测量结果作为模型的输入;预测集合指的是集合中的RS配置或者波束,用于确定一个或多个下行波束作为预测标签。
例如,第一通信设备测量该测量集合中beam上发送的RS,得到RS的测量结果,并将RS测量结果作为模型的输入;第一通信设备通过测量该预测集合中beam上发送的RS,用非AI/ML的方法计算出一个或多个best beam,并将这一个或多个best beam作为模型的输出标签。
由于第一通信设备接收属于测量集合的第一RS配置和属于预测集合的第二RS配置之后的行为有区别,因此,需要区分不同的RS配置分别属于哪个集合。
一种实施方式中,可以用RS配置的标识ID区分,例如ID=0是测量集合的配置,ID=1是预测集合的配置;或者ID是奇数代表测量集合的配置,ID是偶数代表预测集合的配置。
在测量集合为预测集合子集的情况下,可以只对预测集合进行RS配置,RS配置中可以包括属于预测集合的第二RS配置和测量集合图样,通过测量集合图样来指示预测集合中属于测量集合的第一RS配置。
可选地,测量集合图样包括以下一种或多种:
(1)第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合。
具体地,可以通过第一图样指示属于预测集合的每个第二RS配置是否 属于测量集合,进而指示预测集合中属于测量集合的第一RS配置。
一种实施方式中,第一图样可以是在每个RS资源配置中增加1比特,例如当该比特=1时,表示该RS既属于预测集合,也属于测量集合;当该比特=0时,表示该RS只属于预测集合。
一种实施方式中,第一图样可以是与RS资源配置个数相等的字符串或位图,例如,当对应位置的比特=1时,表示该RS既属于预测集合,也属于测量集合;当对应位置的比特=0时,表示该RS只属于预测集合,那么01011可以表示预测集合中的第2、4、5个RS为测量集合。
(2)第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识。
具体地,可以通过第二图样指示属于测量集合的RS配置所对应的RS标识,进而指示预测集合中属于测量集合的第一RS配置。
例如,第二图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个RS为测量集合。
(3)第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律。
具体地,可以通过第三图样指示属于测量集合的RS配置所满足的RS序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,可以配置第三图样为:预测集合中的偶数序号的RS为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。
(4)第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通过第四图样指示属于测量集合的RS配置所满足的波束序号,进而指示预测集合中属于测量集合的第一RS配置。
例如,第四图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个下行波束为测量集合。
(5)第五图样,第五图样用于指示属于测量集合的RS配置所满足的波 束序号排列规律。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通过第五图样指示属于测量集合的RS配置所满足的波束序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,配置第五图样为:预测集合中的偶数序号的下行波束为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。
对于测量集合为预测集合的子集的情况,在RS配置时可以采用测量集合图样指示的方式,从而节省用于配置的比特开销。
可选地,波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
具体地,第一通信设备还可以基于波束描述信息对用于下行波束预测的AI/ML模型进行训练或更新,波束描述信息中可以包含beam ID、RS ID、网络设备的天线配置信息、波束的角度信息、波束的宽度信息中的一项或多项。可选地,波束描述信息可以作为辅助信息与数据集和/或RS配置一起使用,对用于下行波束预测的AI/ML模型进行训练或更新,以及后续预测流程。
一个波束描述信息中包括一个或多个下行波束的描述信息。一个下行波束可以是一个Tx beam,或者一个Rx beam,或者一个beam pair。可选地,在一个波束描述信息中包括多个下行波束的描述信息时,可以按照列表的方式,分别对多个下行波束进行描述。
可以理解的是,终端或网络设备或第三方设备预先定义波束描述信息的过程中,将每个下行波束对应一个beam ID。RS ID为在该下行波束上发送的RS的指示符,例如,CSI-RS的指示符为CRI,SSB的指示符为SSBRI。对一个下行波束进行描述时,还可以用该下行波束的网络设备的天线配置信息,该下行波束的角度信息和该下行波束的宽度信息来对该下行波束进行描述。
可选地,第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
具体地,不同的数据集、不同的RS配置和不同的beam描述信息,分别可以针对不同的应用场景,或者基站配置,或者AI功能。
应用场景包括:城市,农村,室内,室外,高速公路,高铁,Uma,Umi等等;基站配置包括:基站天线配置,波束配置,参考信号配置等等;AI功能,主要包括空域预测、频域预测、时域预测、下行beam pair预测、下行Tx beam预测、下行Rx beam预测等等,其中,空域预测是指测量少量的beam,或者其他参考信号类型的beam,预测大量beam中的best beam;频域预测是指测量频率1上的beam,预测另一个频率2上的beam中的best beam;时域预测是指测量现在时刻的beam,预测未来时刻的best beam。
通过第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联,可以得到不同场景下使用的用于下行波束预测的AI/ML模型。
可选地,该方法还包括:
接收第一通信设备发送的第二信息;
基于第二信息,以及波束标识和物理波束之间的映射关系,确定用于与第一通信设备之间进行信息传输的目标下行波束。
具体地,在第一通信设备预测出一个或多个目标下行波束后,第一通信设备可以向第二通信设备发送第二信息,第二信息用于将第一通信设备预测出的一个或多个目标下行波束的结果告知第二通信设备。
第二通信设备接收到第一通信设备发送的第二信息后,可以根据第一信息和第二信息得知第一通信设备预测出的一个或多个目标下行波束,进而第一通信设备和第二通信设备之间可以在目标下行波束(即best beam)上传输信息,提升系统的传输性能。
其中,第二信息中包含以下一项或多项:
(1)目标下行波束对应的波束标识。
第二信息中可以包括目标下行波束对应的beam ID,从而第二通信设备可以根据目标下行波束对应的beam ID得知第一通信设备通过AI/ML模型推 测出的目标下行波束。
一种实施方式中,第一通信设备通过AI/ML模型只推测出一个目标下行波束,第一通信设备可以将该目标下行波束对应的beam ID发送给第二通信设备。
因为第一信息是第二通信设备定义的或者是第三方设备定义但第二通信设备已知的,所以beam ID和物理波束的映射关系是第二通信设备已知的,即第一通信设备使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的beam ID,可以被第二通信设备识别,并对应到物理实际的目标下行波束,从而第一通信设备和第二通信设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,第一通信设备通过AI/ML模型推测出多个目标下行波束,第一通信设备可以选择其中一个目标下行波束对应的beam ID将其发送给第二通信设备;或者,第一通信设备可以将多个目标下行波束对应的beam ID发送给第二通信设备,由第二通信设备选择其中一个。
(2)目标下行波束对应的RS标识。
第二信息中可以包括目标下行波束对应的RS ID,从而第二通信设备可以根据目标下行波束对应的RS ID得知第一通信设备通过AI/ML模型推测出的目标下行波束。
因为第一信息是第二通信设备定义的或者是第三方设备定义但第二通信设备已知的,所以RS ID和物理波束的映射关系是第二通信设备已知的,即第一通信设备使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的RS ID,可以被第二通信设备识别,并对应到物理实际的目标下行波束,从而第一通信设备和第二通信设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,第一通信设备通过AI/ML模型只推测出一个目标下行波束,第一通信设备可以将该目标下行波束对应的RS ID发送给第二通信设备。
一种实施方式中,第一通信设备通过AI/ML模型推测出多个目标下行波 束,第一通信设备可以选择其中一个目标下行波束对应的RS ID将其发送给第二通信设备;或者,第一通信设备可以将多个目标下行波束对应的RS ID发送给第二通信设备,由第二通信设备选择其中一个。
(3)用于预测目标下行波束的AI/ML模型的标识。
在训练或更新得到多个不同的AI/ML模型的情况下,每个模型都可以对应一个model ID,此时,第一通信设备不仅需要发送预测得到的目标下行波束对应的beam ID或者RS ID,还需要发送预测得到目标下行波束对应的model ID,从而第二通信设备可以根据模型的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
(4)用于预测目标下行波束的数据集的标识。
在有多个不同的数据集用于训练或更新得到多个不同的AI/ML模型的情况下,第二信息中可以包括用于预测目标下行波束的数据集的标识,从而第二通信设备可以根据用于预测目标下行波束的数据集的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
例如,如果数据集和训练、更新的模型是一一对应的,那么第一通信设备也可以发送数据集的标识,代表上述model ID,第二通信设备通过数据集的标识,就可以知道第一通信设备使用该模型输出的beam ID和物理beam的映射关系。
(5)用于预测目标下行波束的RS配置的标识。
在有多个不同的RS配置用于训练或更新得到多个不同的AI/ML模型的情况下,第二信息中可以包括用于预测目标下行波束的RS配置的标识,从而第二通信设备可以根据用于预测目标下行波束的RS配置的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
例如,如果RS配置和训练或更新的模型是一一对应的,那么第一通信设备也可以发送RS配置的标识,代表上述model ID,第二通信设备通过RS配置的标识,就可以知道第一通信设备使用该model输出的RS ID和物理beam的映射关系。
(6)用于预测目标下行波束的波束描述信息的标识。
在有多个不同的波束描述信息用于训练或更新得到多个不同的AI/ML模型的情况下,第二信息中可以包括用于预测目标下行波束的波束描述信息的标识,从而第二通信设备可以根据用于预测目标下行波束的波束描述信息的标识得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
例如,如果波束描述信息和训练或更新的模型是一一对应的,那么第一通信设备也可以发送波束描述信息的标识,代表上述model ID,第二通信设备通过RS配置的标识,就可以知道第一通信设备使用该model输出的beam ID或RS ID和物理beam的映射关系。
(7)用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
在训练或更新得到多个不同的AI/ML模型,且每个AI/ML模型所输入波束数量与输出波束数量信息不同的情况下,第二信息中可以包括用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息,从而第二通信设备可以根据用于预测目标下行波束的输入波束数量与输出波束数量信息得知第一通信设备通过哪个AI/ML模型推测出的目标下行波束。
其中,输入波束数量可以是模型的输入Tx beam数,或者模型的输入Rx beam数,或者模型的输入beam pair数等;输出波束数量可以是模型的输出Tx beam数,或者模型的输出Rx beam数,或者模型的输出beam pair数等。
例如,如果第一通信设备针对每个{输入beam数,输出beam数}对进行训练或更新得到一个AI/ML model,那么{输入beam数,输出beam数}对的信息可以代表model。即,第一通信设备发送{输入beam数,输出beam数}对的信息可以对应model ID,第二通信设备通过{输入beam数,输出beam数}对信息,就可以知道第一通信设备使用该model的输出的beam ID或RS ID和物理beam的映射关系。
图3为本公开实施例提供的下行波束预测方法的流程示意图之三,如图3所示,该方法应用于终端,包括:
步骤300、接收网络设备发送的参考信号RS配置。
步骤301、基于RS配置,对用于下行波束预测的人工智能或机器学习 AI/ML模型进行训练或更新。
步骤302、基于训练或更新后的AI/ML模型,预测出一个或多个目标下行波束,目标下行波束用于网络设备与终端之间的信息传输。
具体地,网络设备可以确定RS配置,并发送给终端。网络设备确定第一信息可以是网络设备自己预先定义的或者由第三方设备预先定义并发送给网络设备的。
终端接收到网络设备发送的RS配置后,可以根据RS配置对用于下行波束预测的AI/ML模型进行训练或更新。其中,对模型进行更新包括对模型进行微调,即终端接收到RS配置后,可以根据RS配置对用于下行波束预测的AI/ML模型进行微调。该用于下行波束预测的AI/ML模型的功能可以包括:空域预测,频域预测,时域预测,下行beam pair预测,下行Tx beam预测,下行Rx beam预测等等。空域预测是指测量少量的波束,或者其他参考信号类型的波束,预测大量波束中的最优波束;频域预测是指测量频率1上的波束,预测另一个频率2上的波束中的最优波束;时域预测是指测量现在时刻的波束,预测未来时刻的最优波束。
其中,RS测量结果可以是RSRP等任意可以表示下行波束接收性能的测量结果,具体形式不做限定。
得到训练或更新后的模型后,终端可以对RS进行测量,并将得到的RS测量结果通过训练或更新后的模型,预测出一个或多个目标下行波束,预测出的目标下行波束可以是Tx beam,或者Rx beam,或者beam pair,用于终端和网络设备之间的信息传输。
RS配置可以是通过Uu接口的控制信道或者数据信道传送的,也可以通过其他接口、其他信道来传送,具体情形不做限制。
本公开实施例提供的下行波束预测方法,通过向终端发送RS配置,终端可以基于接收到的RS配置来对用于下行波束预测的AI/ML模型进行训练或更新,并基于训练或更新后的AI/ML模型预测出一个或多个目标下行波束,从而可以有效节省RS发送资源、UE测量开销和降低UE测量时延。
可选地,RS配置中包含RS配置标识、属于预测集合的第二RS配置以 及测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
具体地,RS配置可以是一个或多个,RS配置标识代表了不同的RS配置,不同的RS配置对应的RS配置标识不同。RS配置可以包括RS的种类(例如CSI-RS、SSB、PT-RS、CRS、DMRS等)、RS发送资源和RS ID等。其中,RS ID为RS指示符,例如,CSI-RS的指示符为CRI,SSB的指示符为SSBRI。
测量集合指的是集合中的RS配置或者波束,用于获取RS测量结果作为模型的输入;预测集合指的是集合中的RS配置或者波束,用于确定一个或多个下行波束作为预测标签。
例如,终端测量该测量集合中beam上发送的RS,得到RS的测量结果,并将RS测量结果作为模型的输入;终端通过测量该预测集合中beam上发送的RS,用非AI/ML的方法计算出一个或多个best beam,并将这一个或多个best beam作为模型的输出标签。
由于终端接收属于测量集合的第一RS配置和属于预测集合的第二RS配置之后的行为有区别,因此,需要区分不同的RS配置分别属于哪个集合。
一种实施方式中,可以用RS配置的标识ID区分,例如ID=0是测量集合的配置,ID=1是预测集合的配置;或者ID是奇数代表测量集合的配置,ID是偶数代表预测集合的配置。
在测量集合为预测集合子集的情况下,可以只对预测集合进行RS配置,RS配置中可以包括属于预测集合的第二RS配置和测量集合图样,通过测量集合图样来指示预测集合中属于测量集合的第一RS配置。
可选地,测量集合图样包括以下一种或多种:
(1)第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合。
具体地,可以通过第一图样指示属于预测集合的每个第二RS配置是否属于测量集合,进而指示预测集合中属于测量集合的第一RS配置。
一种实施方式中,第一图样可以是在每个RS资源配置中增加1比特, 例如当该比特=1时,表示该RS既属于预测集合,也属于测量集合;当该比特=0时,表示该RS只属于预测集合。
一种实施方式中,第一图样可以是与RS资源配置个数相等的字符串或位图,例如,当对应位置的比特=1时,表示该RS既属于预测集合,也属于测量集合;当对应位置的比特=0时,表示该RS只属于预测集合,那么01011可以表示预测集合中的第2、4、5个RS为测量集合。
(2)第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识。
具体地,可以通过第二图样指示属于测量集合的RS配置所对应的RS标识,进而指示预测集合中属于测量集合的第一RS配置。
例如,第二图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个RS为测量集合。
(3)第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律。
具体地,可以通过第三图样指示属于测量集合的RS配置所满足的RS序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,可以配置第三图样为:预测集合中的偶数序号的RS为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。
(4)第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通过第四图样指示属于测量集合的RS配置所满足的波束序号,进而指示预测集合中属于测量集合的第一RS配置。
例如,第四图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个下行波束为测量集合。
(5)第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通 过第五图样指示属于测量集合的RS配置所满足的波束序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,配置第五图样为:预测集合中的偶数序号的下行波束为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。
对于测量集合为预测集合的子集的情况,在RS配置时可以采用测量集合图样指示的方式,从而节省用于配置的比特开销。
可选地,预测出一个或多个目标下行波束之后,该方法还包括:
向网络设备发送第二信息。
具体地,在终端预测出一个或多个目标下行波束后,终端可以向网络设备发送第二信息,第二信息用于将终端预测出的一个或多个目标下行波束的结果告知网络设备。
网络设备接收到终端发送的第二信息后,可以根据第一信息和第二信息得知终端预测出的一个或多个目标下行波束,进而终端和网络设备之间可以在目标下行波束(即best beam)上传输信息,提升系统的传输性能。
其中,第二信息中包含以下一项或多项:
(1)目标下行波束对应的波束标识。
第二信息中可以包括目标下行波束对应的beam ID,从而网络设备可以根据目标下行波束对应的beam ID得知终端通过AI/ML模型推测出的目标下行波束。
因为第一信息是网络设备定义的或者是第三方设备定义但网络设备已知的,所以beam ID和物理波束的映射关系是网络设备已知的,即终端使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的beam ID,可以被网络设备识别,并对应到物理实际的目标下行波束,从而终端和网络设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,终端通过AI/ML模型只推测出一个目标下行波束,终端可以将该目标下行波束对应的beam ID发送给网络设备。
一种实施方式中,终端通过AI/ML模型推测出多个目标下行波束,终端可以选择其中一个目标下行波束对应的beam ID将其发送给网络设备;或者, 终端可以将多个目标下行波束对应的beam ID发送给网络设备,由网络设备选择其中一个。
(2)目标下行波束对应的RS标识。
第二信息中可以包括目标下行波束对应的RS ID,从而网络设备可以根据目标下行波束对应的RS ID得知终端通过AI/ML模型推测出的目标下行波束。
因为第一信息是网络设备定义的或者是第三方设备定义但网络设备已知的,所以RS ID和物理波束的映射关系是网络设备已知的,即终端使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的RS ID,可以被网络设备识别,并对应到物理实际的目标下行波束,从而终端和网络设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,终端通过AI/ML模型只推测出一个目标下行波束,终端可以将该目标下行波束对应的RS ID发送给网络设备。
一种实施方式中,终端通过AI/ML模型推测出多个目标下行波束,终端可以选择其中一个目标下行波束对应的RS ID将其发送给网络设备;或者,终端可以将多个目标下行波束对应的RS ID发送给网络设备,由网络设备选择其中一个。
(3)用于预测目标下行波束的AI/ML模型的标识。
在训练或更新得到多个不同的AI/ML模型的情况下,每个模型都可以对应一个model ID,此时,终端不仅需要发送预测得到的目标下行波束对应的beam ID或者RS ID,还需要发送预测得到目标下行波束对应的model ID,从而网络设备可以根据模型的标识得知终端通过哪个AI/ML模型推测出的目标下行波束。
(4)用于预测目标下行波束的RS配置的标识。
在有多个不同的RS配置用于训练或更新得到多个不同的AI/ML模型的情况下,第二信息中可以包括用于预测目标下行波束的RS配置的标识,从而网络设备可以根据用于预测目标下行波束的RS配置的标识得知终端通过哪个AI/ML模型推测出的目标下行波束。
例如,如果RS配置和训练或更新的模型是一一对应的,那么终端也可以发送RS配置的标识,代表上述model ID,网络设备通过RS配置的标识,就可以知道终端使用该model输出的RS ID和物理beam的映射关系。
(5)用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
在训练或更新得到多个不同的AI/ML模型,且每个AI/ML模型所输入波束数量与输出波束数量信息不同的情况下,第二信息中可以包括用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息,从而网络设备可以根据用于预测目标下行波束的输入波束数量与输出波束数量信息得知终端通过哪个AI/ML模型推测出的目标下行波束。
其中,输入波束数量可以是模型的输入Tx beam数,或者模型的输入Rx beam数,或者模型的输入beam pair数等;输出波束数量可以是模型的输出Tx beam数,或者模型的输出Rx beam数,或者模型的输出beam pair数等。
例如,如果终端针对每个{输入beam数,输出beam数}对进行训练或更新得到一个AI/ML model,那么{输入beam数,输出beam数}对的信息可以代表model。即,终端发送{输入beam数,输出beam数}对的信息可以对应model ID,网络设备通过{输入beam数,输出beam数}对信息,就可以知道终端使用该model的输出的beam ID或RS ID和物理beam的映射关系。
图4为本公开实施例提供的下行波束预测方法的流程示意图之四,如图4所示,该方法应用于网络设备,包括:
步骤400、确定参考信号RS配置,RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,AI/ML模型用于预测一个或多个目标下行波束,目标下行波束用于网络设备与终端之间的信息传输。
步骤401、向终端发送RS配置。
具体地,网络设备可以确定RS配置,并发送给终端。网络设备确定第一信息可以是网络设备自己预先定义的或者由第三方设备预先定义并发送给网络设备的。
终端接收到网络设备发送的RS配置后,可以根据RS配置对用于下行波 束预测的AI/ML模型进行训练或更新。其中,对模型进行更新包括对模型进行微调,即终端接收到RS配置后,可以根据RS配置对用于下行波束预测的AI/ML模型进行微调。该用于下行波束预测的AI/ML模型的功能可以包括:空域预测,频域预测,时域预测,下行beam pair预测,下行Tx beam预测,下行Rx beam预测等等。空域预测是指测量少量的波束,或者其他参考信号类型的波束,预测大量波束中的最优波束;频域预测是指测量频率1上的波束,预测另一个频率2上的波束中的最优波束;时域预测是指测量现在时刻的波束,预测未来时刻的最优波束。
其中,RS测量结果可以是RSRP等任意可以表示下行波束接收性能的测量结果,具体形式不做限定。
得到训练或更新后的模型后,终端可以对RS进行测量,并将得到的RS测量结果通过训练或更新后的模型,预测出一个或多个目标下行波束,预测出的目标下行波束可以是Tx beam,或者Rx beam,或者beam pair,用于终端和网络设备之间的信息传输。
RS配置可以是通过Uu接口的控制信道或者数据信道传送的,也可以通过其他接口、其他信道来传送,具体情形不做限制。
本公开实施例提供的下行波束预测方法,通过向终端发送RS配置,终端可以基于接收到的RS配置来对用于下行波束预测的AI/ML模型进行训练或更新,并基于训练或更新后的AI/ML模型预测出一个或多个目标下行波束,从而可以有效节省RS发送资源、UE测量开销和降低UE测量时延。
可选地,RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
具体地,RS配置可以是一个或多个,RS配置标识代表了不同的RS配置,不同的RS配置对应的RS配置标识不同。RS配置可以包括RS的种类(例如CSI-RS、SSB、PT-RS、CRS、DMRS等)、RS发送资源和RS ID等。其中,RS ID为RS指示符,例如,CSI-RS的指示符为CRI,SSB的指示符为SSBRI。
测量集合指的是集合中的RS配置或者波束,用于获取RS测量结果作为模型的输入;预测集合指的是集合中的RS配置或者波束,用于确定一个或多个下行波束作为预测标签。
例如,终端测量该测量集合中beam上发送的RS,得到RS的测量结果,并将RS测量结果作为模型的输入;终端通过测量该预测集合中beam上发送的RS,用非AI/ML的方法计算出一个或多个best beam,并将这一个或多个best beam作为模型的输出标签。
由于终端接收属于测量集合的第一RS配置和属于预测集合的第二RS配置之后的行为有区别,因此,需要区分不同的RS配置分别属于哪个集合。
一种实施方式中,可以用RS配置的标识ID区分,例如ID=0是测量集合的配置,ID=1是预测集合的配置;或者ID是奇数代表测量集合的配置,ID是偶数代表预测集合的配置。
在测量集合为预测集合子集的情况下,可以只对预测集合进行RS配置,RS配置中可以包括属于预测集合的第二RS配置和测量集合图样,通过测量集合图样来指示预测集合中属于测量集合的第一RS配置。
可选地,测量集合图样包括以下一种或多种:
(1)第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合。
具体地,可以通过第一图样指示属于预测集合的每个第二RS配置是否属于测量集合,进而指示预测集合中属于测量集合的第一RS配置。
一种实施方式中,第一图样可以是在每个RS资源配置中增加1比特,例如当该比特=1时,表示该RS既属于预测集合,也属于测量集合;当该比特=0时,表示该RS只属于预测集合。
一种实施方式中,第一图样可以是与RS资源配置个数相等的字符串或位图,例如,当对应位置的比特=1时,表示该RS既属于预测集合,也属于测量集合;当对应位置的比特=0时,表示该RS只属于预测集合,那么01011可以表示预测集合中的第2、4、5个RS为测量集合。
(2)第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS 标识。
具体地,可以通过第二图样指示属于测量集合的RS配置所对应的RS标识,进而指示预测集合中属于测量集合的第一RS配置。
例如,第二图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个RS为测量集合。
(3)第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律。
具体地,可以通过第三图样指示属于测量集合的RS配置所满足的RS序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,可以配置第三图样为:预测集合中的偶数序号的RS为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。
(4)第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通过第四图样指示属于测量集合的RS配置所满足的波束序号,进而指示预测集合中属于测量集合的第一RS配置。
例如,第四图样为:{2,4,6,8},可以表示预测集合中的第2、4、6、8个下行波束为测量集合。
(5)第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
具体地,在预测集合中为每个RS配置对应的beam ID的情况下,可以通过第五图样指示属于测量集合的RS配置所满足的波束序号排列规律,进而指示预测集合中属于测量集合的第一RS配置。
例如,配置第五图样为:预测集合中的偶数序号的下行波束为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。
对于测量集合为预测集合的子集的情况,在RS配置时可以采用测量集合图样指示的方式,从而节省用于配置的比特开销。
可选地,该方法还包括:
接收终端发送的第二信息;
基于第二信息,以及波束标识和物理波束之间的映射关系,确定用于与终端之间进行信息传输的目标下行波束。
具体地,在终端预测出一个或多个目标下行波束后,终端可以向网络设备发送第二信息,第二信息用于将终端预测出的一个或多个目标下行波束的结果告知网络设备。
网络设备接收到终端发送的第二信息后,可以根据第一信息和第二信息得知终端预测出的一个或多个目标下行波束,进而终端和网络设备之间可以在目标下行波束(即best beam)上传输信息,提升系统的传输性能。
其中,第二信息中包含以下一项或多项:
(1)目标下行波束对应的波束标识。
第二信息中可以包括目标下行波束对应的beam ID,从而网络设备可以根据目标下行波束对应的beam ID得知终端通过AI/ML模型推测出的目标下行波束。
因为第一信息是网络设备定义的或者是第三方设备定义但网络设备已知的,所以beam ID和物理波束的映射关系是网络设备已知的,即终端使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的beam ID,可以被网络设备识别,并对应到物理实际的目标下行波束,从而终端和网络设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,终端通过AI/ML模型只推测出一个目标下行波束,终端可以将该目标下行波束对应的beam ID发送给网络设备。
一种实施方式中,终端通过AI/ML模型推测出多个目标下行波束,终端可以选择其中一个目标下行波束对应的beam ID将其发送给网络设备;或者,终端可以将多个目标下行波束对应的beam ID发送给网络设备,由网络设备选择其中一个。
(2)目标下行波束对应的RS标识。
第二信息中可以包括目标下行波束对应的RS ID,从而网络设备可以根据目标下行波束对应的RS ID得知终端通过AI/ML模型推测出的目标下行波 束。
因为第一信息是网络设备定义的或者是第三方设备定义但网络设备已知的,所以RS ID和物理波束的映射关系是网络设备已知的,即终端使用第一信息训练或更新的AI/ML模型的预测出的一个或多个目标下行波束对应的RS ID,可以被网络设备识别,并对应到物理实际的目标下行波束,从而终端和网络设备之间后续可以用该目标下行波束进行通信。
一种实施方式中,终端通过AI/ML模型只推测出一个目标下行波束,终端可以将该目标下行波束对应的RS ID发送给网络设备。
一种实施方式中,终端通过AI/ML模型推测出多个目标下行波束,终端可以选择其中一个目标下行波束对应的RS ID将其发送给网络设备;或者,终端可以将多个目标下行波束对应的RS ID发送给网络设备,由网络设备选择其中一个。
(3)用于预测目标下行波束的AI/ML模型的标识。
在训练或更新得到多个不同的AI/ML模型的情况下,每个模型都可以对应一个model ID,此时,终端不仅需要发送预测得到的目标下行波束对应的beam ID或者RS ID,还需要发送预测得到目标下行波束对应的model ID,从而网络设备可以根据模型的标识得知终端通过哪个AI/ML模型推测出的目标下行波束。
(4)用于预测目标下行波束的RS配置的标识。
在有多个不同的RS配置用于训练或更新得到多个不同的AI/ML模型的情况下,第二信息中可以包括用于预测目标下行波束的RS配置的标识,从而网络设备可以根据用于预测目标下行波束的RS配置的标识得知终端通过哪个AI/ML模型推测出的目标下行波束。
例如,如果RS配置和训练或更新的模型是一一对应的,那么终端也可以发送RS配置的标识,代表上述model ID,网络设备通过RS配置的标识,就可以知道终端使用该model输出的RS ID和物理beam的映射关系。
(5)用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
在训练或更新得到多个不同的AI/ML模型,且每个AI/ML模型所输入波束数量与输出波束数量信息不同的情况下,第二信息中可以包括用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息,从而网络设备可以根据用于预测目标下行波束的输入波束数量与输出波束数量信息得知终端通过哪个AI/ML模型推测出的目标下行波束。
其中,输入波束数量可以是模型的输入Tx beam数,或者模型的输入Rx beam数,或者模型的输入beam pair数等;输出波束数量可以是模型的输出Tx beam数,或者模型的输出Rx beam数,或者模型的输出beam pair数等。
例如,如果终端针对每个{输入beam数,输出beam数}对进行训练或更新得到一个AI/ML model,那么{输入beam数,输出beam数}对的信息可以代表model。即,终端发送{输入beam数,输出beam数}对的信息可以对应model ID,网络设备通过{输入beam数,输出beam数}对信息,就可以知道终端使用该model的输出的beam ID或RS ID和物理beam的映射关系。
本公开各实施例提供的方法是基于同一申请构思的,因此各方法的实施可以相互参见,重复之处不再赘述。
以下通过具体应用场景的实施例对本公开各上述实施例提供的方法进行举例说明。
实施例一:UE基于数据集进行AI/ML模型的训练或更新。
步骤1:UE接收一个或多个数据集。
该一个或多个数据集,可以来自于基站或第三方设备。
数据集包括:数据集标识,一个或多个数据集样本。
数据集样本包括:多个beam ID,以及该beam ID对应的beam的RSRP(作为AI/ML模型(model)的训练、更新或Fine-tuning的输入),一个或多个最优的beam ID(作为AI/ML模型的训练、更新或Fine-tuning的输出标签)。
不同的beam ID实际上对应不同的物理beam(即实际发送的beam)。
这里的beam可以是beam pair,或者Tx beam,或者Rx beam。
步骤2:UE基于数据集进行AI/ML模型的训练、更新或Fine-tuning,得到对应的AI/ML模型。
如果UE用不同的数据集训练、更新或Fine-tuning,可以得到不同的AI/ML模型。不同的数据集,可以针对不同的应用场景,或者基站配置,或者AI功能。
其中,应用场景包括:城市,农村,室内,室外,高速公路,高铁,Uma,Umi等等;基站配置包括:基站天线配置,波束配置,参考信号配置等等;AI功能,主要包括空域预测、频域预测、时域预测、下行beam pair预测、下行Tx beam预测、下行Rx beam预测等等。其中,空域预测是指测量少量的beam,或者其他参考信号类型的beam,预测大量beam中的best beam;频域预测是指测量频率1上的beam,预测另一个频率2上的beam中的best beam;时域预测是指测量现在时刻的beam,预测未来时刻的best beam。
步骤3:UE对测量集合中的Tx beam上发送的RS进行测量,使用AI/ML模型进行最优波束预测,推理得到一个或多个最优beam ID。
如果是beam pair预测,UE使用不同的Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam和Rx beam pair对应的RSRP;如果是Tx beam预测,UE使用一个Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam对应的RSRP;如果是Rx beam预测,UE使用不同的Rx beam对基站发送的一个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Rx beam对应的RSRP。
上述RSRP作为AI/ML模型的输入,进行推理得到一个或多个最优beam ID。
其中,beam ID是指beam pair预测的最优beam pair中的Tx beam的beam ID,或Tx beam预测的最优Tx beam的beam ID,或Rx beam预测的最优Rx beam的beam ID。
推理得到一个或多个最优beam,可以是基站发送的Tx beam中的一个或多个,也可以不是基站发送的Tx beam,是基站的其他Tx beam。
步骤4:UE将该一个或多个最优beam ID上报给基站。
如果推理得到一个最优beam ID,则将该beam ID上报给基站;如果推理 得到多个最优beam ID,则选择一个beam ID上报给基站,或者都上报给基站,由基站选择一个。
如果UE用不同的数据集训练、更新或Fine-tuning得到多个AI/ML模型,UE还需要将数据集对应的model ID上报给基站,即UE不仅需要上报给基站预测得到的最优beam ID,也需要上报预测该beam的model的ID。
如果数据集和训练、更新或Fine-tuning的model是一一对应的,那么终端也可以上报数据集标识信息,代表上述model ID。
如果UE针对每个{输入Tx beam数,输出Tx beam数}对进行训练、更新或Fine-tuning得到一个AI model,那么{输入Tx beam数,输出Tx beam数}对信息可以代表model。即UE可以上报{输入Tx beam数,输出Tx beam数}对信息,代表上述model ID。
步骤5:基站根据beam ID和物理beam的映射关系,得到物理实际的最优beam,从而基站用该beam和UE进行通信。
因为数据集是基站定义的或者是第三方定义,但基站已知的(第三方需要告知基站该数据集中的beam ID和物理实际beam的对应关系),所以基站根据beam ID和物理beam的映射关系,得到物理实际的最优beam,从而基站用该beam和UE进行通信。
如果UE用不同的数据集训练、更新或Fine-tuning得到多个AI/ML模型,基站根据UE上报的model ID,或数据集标识信息,或{输入Tx beam数,输出Tx beam数}对信息,得到UE使用该数据集或{输入Tx beam数,输出Tx beam数}对训练、更新或Fine-tuning得到的model输出的beam ID和物理beam的映射关系,从而得到物理实际的最优beam,基站用该beam和UE进行通信。
实施例二:UE基于RS配置进行AI/ML模型的训练、更新或Fine-tuning。
步骤1:UE接收一个或多个RS配置。
该一个或多个RS配置,是由基站生成,并发送给UE的。
RS配置包括:测量集合beam上的RS配置,以及预测集合beam上的RS配置。测量集合beam上的RS配置,或预测集合beam上的RS配置,包 括RS配置标识,RS种类(CSI-RS,SSB,PT-RS,CRS,DMRS等),RS发送资源,RS ID等。
这里的beam可以是beam pair,或者Tx beam,或者Rx beam。
特别的,对于UE测量一部分beam,预测全部beam的场景,即测量集合是预测集合的子集,在RS配置时,可以有如下优化方法:
只配置预测集合的RS,并且配置测量集合的图样(pattern),这样UE就可以基于预测集合和图样,得到测量集合的RS配置(即预测集合RS配置的子集),从而节省用于配置的比特开销。
配置测量集合的图样的方法:
Alt.1:在预测集合的RS配置中,每个RS资源配置中增加1比特,例如,当该比特=1时,表示该RS即属于预测集合,也属于测量集合;当该比特=0时,表示该RS只属于预测集合;
Alt.2:配置测量集合图样,包括预测集合中RS的ID号,例如配置图样为:{2,4,6,8},即表示预测集合中的第2,4,6,8个RS为测量集合;
Alt.3:配置测量集合的图样是预测集合中RS序号排列的一种规律,例如配置图样为预测集合中的偶数序号的RS为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律;
Alt.4:配置测量集合图样,包括预测集合中beam的ID号,例如配置图样为:{2,4,6,8},即表示预测集合中的第2,4,6,8个beam为测量集合。此方法需要在预测集合为每个RS配置对应的beam ID;
Alt.5:配置测量集合的图样是预测集合中beam排列的一种规律,例如配置图样为预测集合中的偶数序号的beam为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。此方法需要在预测集合为每个RS配置对应的beam ID。
步骤2:UE基于RS配置,对测量集合和预测集合的beam上配置的RS进行测量,根据测量结果进行AI/ML模型的训练、更新或Fine-tuning,得到对应的AI/ML模型。
如果UE基于不同的RS配置和测量结果,训练、更新或Fine-tuning,可 以得到不同的AI/ML模型。不同的RS配置,可以针对不同的应用场景,或者基站配置,或者AI功能。
其中,应用场景包括:城市,农村,室内,室外,高速公路,高铁,Uma,Umi等等;基站配置包括:基站天线配置,波束配置,参考信号配置等等;AI功能,主要包括空域预测、频域预测、时域预测、下行beam pair预测、下行Tx beam预测、下行Rx beam预测等等。其中,空域预测是指测量少量的beam,或者其他参考信号类型的beam,预测大量beam中的best beam;频域预测是指测量频率1上的beam,预测另一个频率2上的beam中的best beam;时域预测是指测量现在时刻的beam,预测未来时刻的best beam。
步骤3:UE对测量集合中的Tx beam上发送的RS进行测量,使用AI/ML模型进行最优波束预测,推理得到一个或多个最优beam对应的RS ID。
如果是beam pair预测,UE使用不同的Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam和Rx beam pair对应的RSRP;如果是Tx beam预测,UE使用一个Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam对应的RSRP;如果是Rx beam预测,UE使用不同的Rx beam对基站发送的一个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Rx beam对应的RSRP。
上述RSRP作为AI/ML模型的输入,进行推理得到一个或多个最优beam对应的RS ID。
其中,RS ID是指beam pair预测的最优beam pair中的Tx beam上配置的RS ID,或Tx beam预测的最优Tx beam上配置的RS ID,或Rx beam预测的最优Rx beam的RS ID。
推理得到一个或多个最优beam,可以是基站发送的Tx beam中的一个或多个,也可以不是基站发送的Tx beam,属于基站的其他Tx beam。
步骤4:UE将该一个或多个最优beam的RS ID上报给基站。
如果推理得到一个最优beam的RS ID,则将该beam的RS ID上报给基站;如果推理得到多个最优beam的RS ID,则选择一个beam的RS ID上报 给基站,或者都上报给基站,由基站选择一个。
如果UE用不同的RS配置训练、更新或Fine-tuning得到多个AI/ML模型,UE还需要将RS配置对应的model ID上报给基站,即UE不仅需要上报给基站预测得到的最优beam ID,也需要上报预测该beam的model的ID。
如果RS配置和训练、更新或Fine-tuning的model是一一对应的,那么终端也可以上报RS配置标识信息,代表上述model ID。
如果UE针对每个{输入Tx beam数,输出Tx beam数}对进行训练、更新或Fine-tuning得到一个AI model,那么{输入Tx beam数,输出Tx beam数}对信息可以代表model。即UE可以上报{输入Tx beam数,输出Tx beam数}对信息,代表上述model ID。
步骤5:基站根据RS ID和物理beam的映射关系,得到物理实际的最优beam,从而基站用该beam和UE进行通信。
因为RS配置是基站生成并配置的,所以基站根据RS ID和物理beam的映射关系,得到物理实际的最优beam,从而基站用该beam和UE进行通信。
如果UE用不同的RS配置训练、更新或Fine-tuning得到多个AI/ML模型,基站根据UE上报的model ID,或RS配置标识信息,或{输入Tx beam数,输出Tx beam数}对信息,得到UE使用该RS配置或{输入Tx beam数,输出Tx beam数}对训练、更新或Fine-tuning得到的model输出的beam ID和物理beam的映射关系,从而得到物理实际的最优beam,基站用该beam和UE进行通信。
实施例三:测量集合是预测集合的子集的场景。
步骤1:UE接收一个或多个RS配置。
该一个或多个RS配置,是由基站生成,并发送给UE的。
RS配置包括:测量集合beam上的RS配置,以及预测集合beam上的RS配置。
测量集合beam上的RS配置,或预测集合beam上的RS配置,包括RS配置标识,RS种类(CSI-RS,SSB,PT-RS,CRS,DMRS等),RS发送资源,RS ID等。RS ID,即RS indication,例如CSI-RS的indication为CRI, SSB的indication为SSBRI。
这里的beam可以是beam pair,或者Tx beam,或者Rx beam。
RS配置方法包括:
只配置预测集合的RS,并且配置测量集合的图样,这样UE就可以基于预测集合和图样,得到测量集合的RS配置(即预测集合RS配置的子集),从而节省用于配置的比特开销。
配置测量集合的图样的方法:
Alt.1:在预测集合的RS配置中,每个RS资源配置中增加1比特,例如,当该比特=1时,表示该RS即属于预测集合,也属于测量集合;当该比特=0时,表示该RS只属于预测集合;
Alt.2:配置测量集合图样,包括预测集合中RS的ID号,例如配置图样为:{2,4,6,8},即表示预测集合中的第2,4,6,8个RS为测量集合;
Alt.3:配置测量集合的图样是预测集合中RS序号排列的一种规律,例如配置图样为预测集合中的偶数序号的RS为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律;
Alt.4:配置测量集合图样,包括预测集合中beam的ID号,例如配置图样为:{2,4,6,8},即表示预测集合中的第2,4,6,8个beam为测量集合。此方法需要在预测集合为每个RS配置对应的beam ID;
Alt.5:配置测量集合的图样是预测集合中beam排列的一种规律,例如配置图样为预测集合中的偶数序号的beam为测量集合。可以用字符串表示不同的规律,例如2比特可以表示4种规律。此方法需要在预测集合为每个RS配置对应的beam ID。
步骤2:UE基于RS配置,对测量集合和预测集合的beam上配置的RS进行测量,根据测量结果进行AI/ML模型的训练、更新或Fine-tuning,得到对应的AI/ML模型。
UE基于预测集合RS配置和测量集合的图样配置,得到测量集合的RS配置,并对测量集合和预测集合的beam上配置的RS进行测量。
测量集合是指UE需要测量该集合中的beam上发送的RS,从而得到 AI/ML模型的输入(RS的测量结果作为输入)。
预测集合是指UE需要在该集合中预测出一个或多个最优beam。
在AI/ML模型的训练、更新或Fine-tuning过程中,UE需要测量该预测集合,并用非AI的方法计算出一个或多个最优beam,得到AI/ML模型的训练、更新或Fine-tuning的输出标签。
步骤3:UE对测量集合中的Tx beam上发送的RS进行测量,使用AI/ML模型进行最优波束预测,推理得到一个或多个最优beam对应的RS ID。
UE基于预测集合RS配置和测量集合的图样配置,得到测量集合的RS配置,并对测量集合上配置的RS进行测量。
步骤4:UE将该一个或多个最优beam的RS ID上报给基站。
步骤5:基站根据RS ID和物理beam的映射关系,得到物理实际的最优beam,从而基站用该beam和UE进行通信。
实施例四:UE基于beam的描述信息进行AI/ML模型的训练、更新或Fine-tuning。
步骤1:UE接收一个或多个beam描述信息。
该一个或多个beam描述信息,可以是基站生成并发给UE,或者是第三方生成,并发给基站和UE。
Beam的描述信息包括,beam ID或者RS ID,以及该beam ID或者RS ID对应的beam的基站天线配置信息,beam的角度信息,和/或,beam的宽度信息。
如果有多个beam,可以按照beam ID列表的方式,分别描述。
上述Beam ID可以是基站发送Tx beam的编号,RS ID可以是基站在Tx beam上发送的RS的indication,例如,CSI-RS的indication为CRI,SSB的indication为SSBRI。
这里的beam可以是beam pair,或者Tx beam,或者Rx beam。
步骤2:UE基于beam描述信息,以及对Tx beam上配置的RS进行测量,根据测量结果进行AI/ML模型的训练、更新或Fine-tuning,得到对应的AI/ML模型。
如果UE基于不同的beam描述信息和测量结果,可以训练、更新或Fine-tuning得到不同的AI/ML模型。不同的beam描述信息,可以针对不同的应用场景,或者基站配置,或者AI功能。
其中,应用场景包括:城市,农村,室内,室外,高速公路,高铁,Uma,Umi等等;基站配置包括:基站天线配置,波束配置,参考信号配置等等;AI功能,主要包括空域预测、频域预测、时域预测、下行beam pair预测、下行Tx beam预测、下行Rx beam预测等等。其中,空域预测是指测量少量的beam,或者其他参考信号类型的beam,预测大量beam中的best beam;频域预测是指测量频率1上的beam,预测另一个频率2上的beam中的best beam;时域预测是指测量现在时刻的beam,预测未来时刻的best beam。
步骤3:UE对测量集合中的Tx beam上发送的RS进行测量,使用AI/ML模型进行最优波束预测,推理得到一个或多个最优beam对应的beam ID或者RS ID。
如果是beam pair预测,UE使用不同的Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam和Rx beam pair对应的RSRP;如果是Tx beam预测,UE使用一个Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam对应的RSRP;如果是Rx beam预测,UE使用不同的Rx beam对基站发送的一个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Rx beam对应的RSRP。
上述RSRP作为AI/ML模型的输入,进行推理得到一个或多个最优beam对应的beam ID或者RS ID。
其中,beam ID或者RS ID是指beam pair预测的最优beam pair中的Tx beam对应的beam ID或者RS ID,或Tx beam预测的最优Tx beam对应的beam ID或者RS ID,或Rx beam预测的最优Rx beam的beam ID或者RS ID。
推理得到一个或多个最优beam,可以是基站发送的Tx beam中的一个或多个,也可以不是基站发送的Tx beam,属于基站的其他Tx beam。
步骤4:UE将该一个或多个最优beam的beam ID或者RS ID上报给基 站。
如果推理得到一个最优beam的beam ID或者RS ID,则将该beam的beam ID或者RS ID上报给基站;如果推理得到多个最优beam的beam ID或者RS ID,则选择一个beam的beam ID或者RS ID上报给基站,或者都上报给基站,由基站选择一个。
如果UE用不同的beam描述信息训练、更新或Fine-tuning得到多个AI/ML模型,UE还需要将beam描述信息对应的model ID上报给基站,即UE不仅需要上报给基站预测得到的最优beam ID,也需要上报预测该beam的model的ID。
如果beam描述信息和训练、更新或Fine-tuning的model是一一对应的,那么终端也可以上报beam描述信息的标识信息,代表上述model ID。
如果UE针对每个{输入Tx beam数,输出Tx beam数}对进行训练、更新或Fine-tuning得到一个AI model,那么{输入Tx beam数,输出Tx beam数}对信息可以代表model。即UE可以上报{输入Tx beam数,输出Tx beam数}对信息,代表上述model ID。
步骤5:基站根据beam ID或者RS ID和物理beam的映射关系,得到物理实际的最优beam,从而基站用该beam和UE进行通信。
因为beam描述信息是基站生成或者第三方生成并发给基站和UE的,所以基站根据beam ID或者RS ID和物理beam的映射关系,得到物理实际的最优beam,从而基站用该beam和UE进行通信。
如果UE用不同的beam描述信息训练、更新或Fine-tuning得到多个AI/ML模型,基站根据UE上报的model ID,或beam描述信息的标识信息,或{输入Tx beam数,输出Tx beam数}对信息,得到UE使用该beam描述信息或{输入Tx beam数,输出Tx beam数}对训练、更新或Fine-tuning得到的model输出的beam ID和物理beam的映射关系,从而得到物理实际的最优beam,基站用该beam和UE进行通信。
实施例五:基站基于数据集进行AI/ML模型的训练或更新。
步骤1:基站接收一个或多个数据集。
该一个或多个数据集,可以来自于UE或第三方设备。
数据集包括:数据集标识,一个或多个数据集样本。
数据集样本包括:多个beam ID,以及该beam ID对应的beam的RSRP(作为AI/ML模型(model)的训练、更新或Fine-tuning的输入),一个或多个最优的beam ID(作为AI/ML模型的训练、更新或Fine-tuning的输出标签)。
不同的beam ID实际上对应不同的物理beam(即实际发送的beam)。
这里的beam可以是beam pair,或者Tx beam,或者Rx beam。
步骤2:基站基于数据集进行AI/ML模型的训练、更新或Fine-tuning,得到对应的AI/ML模型。
如果基站用不同的数据集训练、更新或Fine-tuning,可以得到不同的AI/ML模型。不同的数据集,可以针对不同的应用场景,或者基站配置,或者AI功能。
其中,应用场景包括:城市,农村,室内,室外,高速公路,高铁,Uma,Umi等等;基站配置包括:基站天线配置,波束配置,参考信号配置等等;AI功能,主要包括空域预测、频域预测、时域预测、下行beam pair预测、下行Tx beam预测、下行Rx beam预测等等。其中,空域预测是指测量少量的beam,或者其他参考信号类型的beam,预测大量beam中的best beam;频域预测是指测量频率1上的beam,预测另一个频率2上的beam中的best beam;时域预测是指测量现在时刻的beam,预测未来时刻的best beam。
步骤3:UE对测量集合中的Tx beam上接收的RS进行测量,然后将测量结果反馈给基站,基站使用AI/ML模型进行最优波束预测,推理得到一个或多个最优beam ID。
如果是beam pair预测,UE使用不同的Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam和Rx beam pair对应的RSRP;如果是Tx beam预测,UE使用一个Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam对应的RSRP;如果是Rx beam预测,UE使用不同的Rx beam对基站发送的一个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Rx beam 对应的RSRP。
UE上报测量结果给基站,由基站使用AI/ML模型进行波束预测。
上述RSRP作为AI/ML模型的输入,进行推理得到一个或多个最优beam ID。
其中,beam ID是指beam pair预测的最优beam pair中的Rx beam的beam ID,或Rx beam预测的最优Rx beam的beam ID,或Rx beam预测的最优Rx beam的beam ID。
推理得到一个或多个最优beam,可以是UE接收的Rx beam中的一个或多个,也可以不是UE接收的Rx beam,是UE的其他Rx beam。
步骤4:基站将该一个或多个最优beam ID发送给UE。
如果推理得到一个最优beam ID,则将该beam ID发送给UE;如果推理得到多个最优beam ID,则选择一个beam ID发送给UE,或者都发送给UE,由UE选择一个。
如果基站用不同的数据集训练、更新或Fine-tuning得到多个AI/ML模型,基站还需要将数据集对应的model ID发送给UE,即基站不仅需要发送给UE预测得到的最优beam ID,也需要发送预测该beam的model的ID。
如果数据集和训练、更新或Fine-tuning的model是一一对应的,那么基站也可以发送数据集标识信息,代表上述model ID。
如果基站针对每个{输入Tx beam数,输出Tx beam数}对进行训练、更新或Fine-tuning得到一个AI model,那么{输入Tx beam数,输出Tx beam数}对信息可以代表model。即基站可以发送{输入Tx beam数,输出Tx beam数}对信息,代表上述model ID。
步骤5:UE根据beam ID和物理beam的映射关系,得到物理实际的最优beam,从而UE用该beam和基站进行通信。
因为数据集是UE定义的或者是第三方定义但UE已知的(第三方需要告知UE该数据集中的beam ID和物理实际beam的对应关系),所以UE根据beam ID和物理beam的映射关系,得到物理实际的最优beam,从而UE用该beam和基站进行通信。
如果基站用不同的数据集训练、更新或Fine-tuning得到多个AI/ML模型,UE根据基站发送的model ID,或数据集标识信息,或{输入Tx beam数,输出Tx beam数}对信息,得到基站使用该数据集或{输入Tx beam数,输出Tx beam数}对训练、更新或Fine-tuning得到的model输出的beam ID和物理beam的映射关系,从而得到物理实际的最优beam,UE用该beam和基站进行通信。
实施例六:基站基于beam的描述信息进行AI/ML模型的训练、更新或Fine-tuning。
步骤1:基站接收一个或多个beam描述信息。
该一个或多个beam描述信息,可以是UE生成并发给基站,或者是第三方生成,并发给UE和基站。
Beam的描述信息包括,beam ID或者RS ID,以及该beam ID或者RS ID对应的beam的UE天线配置信息,beam的角度信息,和/或,beam的宽度信息。
如果有多个beam,可以按照beam ID列表的方式,分别描述。
上述Beam ID可以是UE接收Rx beam的编号,RS ID可以是UE在Rx beam上接收的RS的indication,例如,CSI-RS的indication为CRI,SSB的indication为SSBRI。
这里的beam可以是beam pair,或者Tx beam,或者Rx beam。
步骤2:基站基于beam描述信息,以及对Rx beam上配置的RS进行测量,根据测量结果进行AI/ML模型的训练、更新或Fine-tuning,得到对应的AI/ML模型。
如果基站基于不同的beam描述信息和测量结果,可以训练、更新或Fine-tuning得到不同的AI/ML模型。不同的beam描述信息,可以针对不同的应用场景,或者基站配置,或者AI功能。
其中,应用场景包括:城市,农村,室内,室外,高速公路,高铁,Uma,Umi等等;基站配置包括:基站天线配置,波束配置,参考信号配置等等;AI功能,主要包括空域预测、频域预测、时域预测、下行beam pair预测、下行Tx beam预测、下行Rx beam预测等等。其中,空域预测是指测量少量 的beam,或者其他参考信号类型的beam,预测大量beam中的best beam;频域预测是指测量频率1上的beam,预测另一个频率2上的beam中的best beam;时域预测是指测量现在时刻的beam,预测未来时刻的best beam。
步骤3:UE对测量集合中的Tx beam上发送的RS进行测量,然后将测量结果反馈给基站,基站使用AI/ML模型进行最优波束预测,推理得到一个或多个最优beam对应的beam ID或者RS ID。
如果是beam pair预测,UE使用不同的Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam和Rx beam pair对应的RSRP;如果是Tx beam预测,UE使用一个Rx beam对基站发送的每个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Tx beam对应的RSRP;如果是Rx beam预测,UE使用不同的Rx beam对基站发送的一个Tx beam上的RS(例如,CSI-RS或者SSB)测量,得到每个Rx beam对应的RSRP。
UE上报测量结果给基站,由基站使用AI/ML模型进行波束预测。
上述RSRP作为AI/ML模型的输入,进行推理得到一个或多个最优beam对应的beam ID或者RS ID。
其中,beam ID或者RS ID是指beam pair预测的最优beam pair中的Rx beam对应的beam ID或者RS ID,或Rx beam预测的最优Rx beam对应的beam ID或者RS ID,或Rx beam预测的最优Rx beam的beam ID或者RS ID。
推理得到一个或多个最优beam,可以是UE接收的Rx beam中的一个或多个,也可以不是UE接收的Rx beam,是UE的其他Rx beam。
步骤4:基站将该一个或多个最优beam的beam ID或者RS ID发送给UE。
如果推理得到一个最优beam的beam ID或者RS ID,则将该beam的beam ID或者RS ID发送给UE;如果推理得到多个最优beam的beam ID或者RS ID,则选择一个beam的beam ID或者RS ID发送给UE,或者都发送给UE,由UE选择一个。
如果基站用不同的beam描述信息训练、更新或Fine-tuning得到多个 AI/ML模型,基站还需要将beam描述信息对应的model ID发送给UE,即基站不仅需要发送给UE预测得到的最优beam ID,也需要发送预测该beam的model的ID。
如果beam描述信息和训练、更新或Fine-tuning的model是一一对应的,那么UE也可以发送beam描述信息的标识信息,代表上述model ID。
如果基站针对每个{输入Tx beam数,输出Tx beam数}对进行训练、更新或Fine-tuning得到一个AI model,那么{输入Tx beam数,输出Tx beam数}对信息可以代表model。即基站可以发送{输入Tx beam数,输出Tx beam数}对信息,代表上述model ID。
步骤5:UE根据beam ID或者RS ID和物理beam的映射关系,得到物理实际的最优beam,从而UE用该beam和基站进行通信。
因为beam描述信息是UE生成或者第三方生成并发给UE和基站的,所以UE根据beam ID或者RS ID和物理beam的映射关系,得到物理实际的最优beam,从而UE用该beam和基站进行通信。
如果基站用不同的beam描述信息训练、更新或Fine-tuning得到多个AI/ML模型,UE根据基站发送的model ID,或beam描述信息的标识信息,或{输入Tx beam数,输出Tx beam数}对信息,得到基站使用该beam描述信息或{输入Tx beam数,输出Tx beam数}对训练、更新或Fine-tuning得到的model输出的beam ID和物理beam的映射关系,从而得到物理实际的最优beam,UE用该beam和基站进行通信。
本公开各实施例提供的方法和装置是基于同一申请构思的,由于方法和装置解决问题的原理相似,因此装置和方法的实施可以相互参见,重复之处不再赘述。
图5为本公开实施例提供的第一通信设备的结构示意图,如图5所示,该第一通信设备包括存储器520,收发机510和处理器500;其中,处理器500与存储器520也可以物理上分开布置。
存储器520,用于存储计算机程序;收发机510,用于在处理器500的控制下收发数据。
具体地,收发机510用于在处理器500的控制下接收和发送数据。
其中,在图5中,总线架构可以包括任意数量的互联的总线和桥,具体由处理器500代表的一个或多个处理器和存储器520代表的存储器的各种电路链接在一起。总线架构还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路链接在一起,这些都是本领域所公知的,因此,本公开不再对其进行进一步描述。总线接口提供接口。收发机510可以是多个元件,即包括发送机和接收机,提供用于在传输介质上与各种其他装置通信的单元,这些传输介质包括无线信道、有线信道、光缆等传输介质。
处理器500负责管理总线架构和通常的处理,存储器520可以存储处理器500在执行操作时所使用的数据。
处理器500可以是中央处理器(Central Processing Unit,CPU)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现场可编程门阵列(Field-Programmable Gate Array,FPGA)或复杂可编程逻辑器件(Complex Programmable Logic Device,CPLD),处理器也可以采用多核架构。
处理器500通过调用存储器520存储的计算机程序,用于按照获得的可执行指令执行本公开实施例提供的任一所述方法,例如:
接收第一信息,第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
基于第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
基于训练或更新后的AI/ML模型,预测出一个或多个目标下行波束,目标下行波束用于第一通信设备与第二通信设备之间的信息传输。
可选地,接收第一信息,包括:
接收第二通信设备发送的第一信息。
可选地,数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
可选地,RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,预测出一个或多个目标下行波束之后,该方法还包括:
向第二通信设备发送第二信息,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的数据集的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的波束描述信息的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
图6为本公开实施例提供的第二通信设备的结构示意图,如图6所示,该第二通信设备包括存储器620,收发机610和处理器600;其中,处理器600与存储器620也可以物理上分开布置。
存储器620,用于存储计算机程序;收发机610,用于在处理器600的控制下收发数据。
具体地,收发机610用于在处理器600的控制下接收和发送数据。
其中,在图6中,总线架构可以包括任意数量的互联的总线和桥,具体由处理器600代表的一个或多个处理器和存储器620代表的存储器的各种电路链接在一起。总线架构还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路链接在一起,这些都是本领域所公知的,因此,本公开不再对其进行进一步描述。总线接口提供接口。收发机610可以是多个元件,即包括发送机和接收机,提供用于在传输介质上与各种其他装置通信的单元,这些传输介质包括无线信道、有线信道、光缆等传输介质。
处理器600负责管理总线架构和通常的处理,存储器620可以存储处理器600在执行操作时所使用的数据。
处理器600可以是CPU、ASIC、FPGA或CPLD,处理器也可以采用多核架构。
处理器600通过调用存储器620存储的计算机程序,用于按照获得的可执行指令执行本公开实施例提供的任一所述方法,例如:
确定第一信息,第一信息包括数据集、RS配置、波束描述信息中的一种或多种;第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新, AI/ML模型用于预测一个或多个目标下行波束,目标下行波束用于第二通信设备与第一通信设备之间的信息传输;
向第一通信设备发送第一信息。
可选地,数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
可选地,RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,该方法还包括:
接收第一通信设备发送的第二信息;
基于第二信息,以及波束标识和物理波束之间的映射关系,确定用于与第一通信设备之间进行信息传输的目标下行波束;
其中,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的数据集的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的波束描述信息的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
图7为本公开实施例提供的终端的结构示意图,如图7所示,该终端包括存储器720,收发机710和处理器700;其中,处理器700与存储器720也可以物理上分开布置。
存储器720,用于存储计算机程序;收发机710,用于在处理器700的控制下收发数据。
具体地,收发机710用于在处理器700的控制下接收和发送数据。
其中,在图7中,总线架构可以包括任意数量的互联的总线和桥,具体由处理器700代表的一个或多个处理器和存储器720代表的存储器的各种电路链接在一起。总线架构还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路链接在一起,这些都是本领域所公知的,因此,本公开不再对其进行进一步描述。总线接口提供接口。收发机710可以是多个元件, 即包括发送机和接收机,提供用于在传输介质上与各种其他装置通信的单元,这些传输介质包括无线信道、有线信道、光缆等传输介质。针对不同的用户设备,用户接口730还可以是能够外接内接需要设备的接口,连接的设备包括但不限于小键盘、显示器、扬声器、麦克风、操纵杆等。
处理器700负责管理总线架构和通常的处理,存储器720可以存储处理器700在执行操作时所使用的数据。
处理器700可以是CPU、ASIC、FPGA或CPLD,处理器也可以采用多核架构。
处理器700通过调用存储器720存储的计算机程序,用于按照获得的可执行指令执行本公开实施例提供的任一所述方法,例如:
接收网络设备发送的参考信号RS配置;
基于RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
基于训练或更新后的AI/ML模型,预测出一个或多个目标下行波束,目标下行波束用于网络设备与终端之间的信息传输。
可选地,RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序 号排列规律。
可选地,预测出一个或多个目标下行波束之后,该方法还包括:
向网络设备发送第二信息,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
图8为本公开实施例提供的网络设备的结构示意图,如图8所示,该网络设备包括存储器820,收发机810和处理器800;其中,处理器800与存储器820也可以物理上分开布置。
存储器820,用于存储计算机程序;收发机810,用于在处理器800的控制下收发数据。
具体地,收发机810用于在处理器800的控制下接收和发送数据。
其中,在图8中,总线架构可以包括任意数量的互联的总线和桥,具体由处理器800代表的一个或多个处理器和存储器820代表的存储器的各种电路链接在一起。总线架构还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路链接在一起,这些都是本领域所公知的,因此,本公开不再对其进行进一步描述。总线接口提供接口。收发机810可以是多个元件,即包括发送机和接收机,提供用于在传输介质上与各种其他装置通信的单元,这些传输介质包括无线信道、有线信道、光缆等传输介质。
处理器800负责管理总线架构和通常的处理,存储器820可以存储处理器800在执行操作时所使用的数据。
处理器800可以是CPU、ASIC、FPGA或CPLD,处理器也可以采用多核架构。
处理器800通过调用存储器820存储的计算机程序,用于按照获得的可执行指令执行本公开实施例提供的任一所述方法,例如:
确定参考信号RS配置,RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,AI/ML模型用于预测一个或多个目标下行波束,目标下行波束用于网络设备与终端之间的信息传输;
向终端发送RS配置。
可选地,RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,该方法还包括:
接收终端发送的第二信息;
基于第二信息,以及波束标识和物理波束之间的映射关系,确定用于与终端之间进行信息传输的目标下行波束;
其中,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信 息。
在此需要说明的是,本公开实施例提供的上述第一通信设备、第二通信设备、终端和网络设备,能够实现上述方法实施例所实现的所有方法步骤,且能够达到相同的技术效果,在此不再对本实施例中与方法实施例相同的部分及有益效果进行具体赘述。
图9为本公开实施例提供的下行波束预测装置的结构示意图之一,该装置应用于第一通信设备,如图9所示,该装置包括:
第一接收单元900,用于接收第一信息,第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
第一模型单元910,用于基于第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
第一预测单元920,用于基于训练或更新后的AI/ML模型,预测出一个或多个目标下行波束,目标下行波束用于第一通信设备与第二通信设备之间的信息传输。
可选地,接收第一信息,包括:
接收第二通信设备发送的第一信息。
可选地,数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
可选地,RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,该装置还包括第一发送单元,用于向第二通信设备发送第二信息,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的数据集的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的波束描述信息的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
图10为本公开实施例提供的下行波束预测装置的结构示意图之二,该装 置应用于第二通信设备,如图10所示,该装置包括:
第一确定单元1000,用于确定第一信息,第一信息包括数据集、RS配置、波束描述信息中的一种或多种;第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,AI/ML模型用于预测一个或多个目标下行波束,目标下行波束用于第二通信设备与第一通信设备之间的信息传输;
第二发送单元1010,用于向第一通信设备发送第一信息。
可选地,数据集中包含以下一项或多项:
数据集标识;
一个或多个数据集样本,数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
可选地,RS配置中包含以下一项或多项:
RS配置标识;
属于测量集合的第一RS配置;
属于预测集合的第二RS配置;
测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,波束描述信息中包含以下一项或多项:
波束标识;
RS标识;
网络设备的天线配置信息;
波束的角度信息;
波束的宽度信息。
可选地,第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
可选地,该装置还包括第二接收单元,用于接收第一通信设备发送的第二信息;
第一确定单元1000还用于基于第二信息,以及波束标识和物理波束之间的映射关系,确定用于与第一通信设备之间进行信息传输的目标下行波束;
其中,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的数据集的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的波束描述信息的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
图11为本公开实施例提供的下行波束预测装置的结构示意图之三,该装置应用于终端,如图11所示,该装置包括:
第三接收单元1100,用于接收网络设备发送的参考信号RS配置;
第二模型单元1110,用于基于RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
第二预测单元1120,用于基于训练或更新后的AI/ML模型,预测出一个或多个目标下行波束,目标下行波束用于网络设备与终端之间的信息传输。
可选地,RS配置中包含RS配置标识、属于预测集合的第二RS配置以 及测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,该装置还包括第三发送单元,用于向网络设备发送第二信息,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
图12为本公开实施例提供的下行波束预测装置的结构示意图之四,该装置应用于网络设备,如图12所示,该装置包括:
第二确定单元1200,用于确定参考信号RS配置,RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,AI/ML模型用于预测一个或多个目标下行波束,目标下行波束用于网络设备与终端之间的信息传输;
第四发送单元1210,用于向终端发送RS配置。
可选地,RS配置中包含RS配置标识、属于预测集合的第二RS配置以 及测量集合图样,测量集合图样用于指示属于测量集合的第一RS配置;
其中,第一RS配置用于获取RS测量结果作为模型输入;第二RS配置用于确定一个或多个下行波束作为预测标签。
可选地,测量集合图样包括以下一种或多种:
第一图样,第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
第二图样,第二图样用于指示属于测量集合的RS配置所对应的RS标识;
第三图样,第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
第四图样,第四图样用于指示属于测量集合的RS配置所对应的波束标识;
第五图样,第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
可选地,该装置还包括第四接收单元,用于接收终端发送的第二信息;
第二确定单元1200还用于基于第二信息,以及波束标识和物理波束之间的映射关系,确定用于与终端之间进行信息传输的目标下行波束;
其中,第二信息中包含以下一项或多项:
目标下行波束对应的波束标识;
目标下行波束对应的RS标识;
用于预测目标下行波束的AI/ML模型的标识;
用于预测目标下行波束的RS配置的标识;
用于预测目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
需要说明的是,本公开实施例中对单元的划分是示意性的,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。另外,在本公开各个实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用软件功能单元的形式实现。
所述集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个处理器可读取存储介质中。基于这样的理解,本公开的技术方案本质上或者说对相关技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)或处理器(processor)执行本公开各个实施例所述方法的全部或部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
在此需要说明的是,本公开实施例提供的上述装置,能够实现上述方法实施例所实现的所有方法步骤,且能够达到相同的技术效果,在此不再对本实施例中与方法实施例相同的部分及有益效果进行具体赘述。
另一方面,本公开实施例还提供一种计算机可读存储介质,所述计算机可读存储介质存储有计算机程序,所述计算机程序用于使计算机执行上述各实施例提供的下行波束预测方法。
在此需要说明的是,本公开实施例提供的计算机可读存储介质,能够实现上述方法实施例所实现的所有方法步骤,且能够达到相同的技术效果,在此不再对本实施例中与方法实施例相同的部分及有益效果进行具体赘述。
所述计算机可读存储介质可以是计算机能够存取的任何可用介质或数据存储设备,包括但不限于磁性存储器(例如软盘、硬盘、磁带、磁光盘(MO)等)、光学存储器(例如CD、DVD、BD、HVD等)、以及半导体存储器(例如ROM、EPROM、EEPROM、非易失性存储器(NAND FLASH)、固态硬盘(SSD))等。
本公开实施例提供的技术方案可以适用于多种系统,尤其是5G系统。例如适用的系统可以是全球移动通讯(global system of mobile communication,GSM)系统、码分多址(code division multiple access,CDMA)系统、宽带码分多址(Wideband Code Division Multiple Access,WCDMA)通用分组无线业务(general packet radio service,GPRS)系统、长期演进(long term evolution, LTE)系统、LTE频分双工(frequency division duplex,FDD)系统、LTE时分双工(time division duplex,TDD)系统、高级长期演进(long term evolution advanced,LTE-A)系统、通用移动系统(universal mobile telecommunication system,UMTS)、全球互联微波接入(worldwide interoperability for microwave access,WiMAX)系统、5G新空口(New Radio,NR)系统等。这多种系统中均包括终端设备和网络设备。系统中还可以包括核心网部分,例如演进的分组系统(Evloved Packet System,EPS)、5G系统(5GS)等。
本公开实施例涉及的终端,可以是指向用户提供语音和/或数据连通性的设备,具有无线连接功能的手持式设备、或连接到无线调制解调器的其他处理设备等。在不同的系统中,终端的名称可能也不相同,例如在5G系统中,终端可以称为用户设备(User Equipment,UE)。无线终端设备可以经无线接入网(Radio Access Network,RAN)与一个或多个核心网(Core Network,CN)进行通信,无线终端设备可以是移动终端设备,如移动电话(或称为“蜂窝”电话)和具有移动终端设备的计算机,例如,可以是便携式、袖珍式、手持式、计算机内置的或者车载的移动装置,它们与无线接入网交换语言和/或数据。例如,个人通信业务(Personal Communication Service,PCS)电话、无绳电话、会话发起协议(Session Initiated Protocol,SIP)话机、无线本地环路(Wireless Local Loop,WLL)站、个人数字助理(Personal Digital Assistant,PDA)等设备。无线终端设备也可以称为系统、订户单元(subscriber unit)、订户站(subscriber station),移动站(mobile station)、移动台(mobile)、远程站(remote station)、接入点(access point)、远程终端设备(remote terminal)、接入终端设备(access terminal)、用户终端设备(user terminal)、用户代理(user agent)、用户装置(user device),本公开实施例中并不限定。
本公开实施例涉及的网络设备,可以是基站,该基站可以包括多个为终端提供服务的小区。根据具体应用场合不同,基站又可以称为接入点,或者可以是接入网中在空中接口上通过一个或多个扇区与无线终端设备通信的设备,或者其它名称。网络设备可用于将收到的空中帧与网际协议(Internet Protocol,IP)分组进行相互更换,作为无线终端设备与接入网的其余部分之 间的路由器,其中接入网的其余部分可包括网际协议(IP)通信网络。网络设备还可协调对空中接口的属性管理。例如,本公开实施例涉及的网络设备可以是全球移动通信系统(Global System for Mobile communications,GSM)或码分多址接入(Code Division Multiple Access,CDMA)中的网络设备(Base Transceiver Station,BTS),也可以是带宽码分多址接入(Wide-band Code Division Multiple Access,WCDMA)中的网络设备(NodeB),还可以是长期演进(long term evolution,LTE)系统中的演进型网络设备(evolutional Node B,eNB或e-NodeB)、5G网络架构(next generation system)中的5G基站(gNB),也可以是家庭演进基站(Home evolved Node B,HeNB)、中继节点(relay node)、家庭基站(femto)、微微基站(pico)等,本公开实施例中并不限定。在一些网络结构中,网络设备可以包括集中单元(centralized unit,CU)节点和分布单元(distributed unit,DU)节点,集中单元和分布单元也可以地理上分开布置。
网络设备与终端之间可以各自使用一或多根天线进行多输入多输出(Multi Input Multi Output,MIMO)传输,MIMO传输可以是单用户MIMO(Single User MIMO,SU-MIMO)或多用户MIMO(Multiple User MIMO,MU-MIMO)。根据根天线组合的形态和数量,MIMO传输可以是2D-MIMO、3D-MIMO、FD-MIMO或massive-MIMO,也可以是分集传输或预编码传输或波束赋形传输等。
本领域内的技术人员应明白,本公开的实施例可提供为方法、系统、或计算机程序产品。因此,本公开可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且,本公开可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器和光学存储器等)上实施的计算机程序产品的形式。
本公开是参照根据本公开实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机可执行指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机可执行指令到通用计算机、专用计 算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器,使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些处理器可执行指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的处理器可读存储器中,使得存储在该处理器可读存储器中的指令产生包括指令装置的制造品,该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些处理器可执行指令也可装载到计算机或其他可编程数据处理设备上,使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理,从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
显然,本领域的技术人员可以对本公开进行各种改动和变型而不脱离本公开的精神和范围。这样,倘若本公开的这些修改和变型属于本公开权利要求及其等同技术的范围之内,则本公开也意图包含这些改动和变型在内。

Claims (70)

  1. 一种下行波束预测方法,应用于第一通信设备,包括:
    接收第一信息,所述第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
    基于所述第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
    基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述第一通信设备与第二通信设备之间的信息传输。
  2. 根据权利要求1所述的下行波束预测方法,其中,所述接收第一信息,包括:
    接收所述第二通信设备发送的所述第一信息。
  3. 根据权利要求1或2所述的下行波束预测方法,其中,所述数据集中包含以下一项或多项:
    数据集标识;
    一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
  4. 根据权利要求1或2所述的下行波束预测方法,其中,所述RS配置中包含以下一项或多项:
    RS配置标识;
    属于测量集合的第一RS配置;
    属于预测集合的第二RS配置;
    测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  5. 根据权利要求4所述的下行波束预测方法,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是 否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  6. 根据权利要求1或2所述的下行波束预测方法,其中,所述波束描述信息中包含以下一项或多项:
    波束标识;
    RS标识;
    网络设备的天线配置信息;
    波束的角度信息;
    波束的宽度信息。
  7. 根据权利要求1所述的下行波束预测方法,其中,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
  8. 根据权利要求1至7任一所述的下行波束预测方法,其中,所述预测出一个或多个目标下行波束之后,所述方法还包括:
    向所述第二通信设备发送第二信息,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的数据集的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的波束描述信息的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  9. 一种下行波束预测方法,应用于第二通信设备,包括:
    确定第一信息,所述第一信息包括数据集、RS配置、波束描述信息中的一种或多种;所述第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述第二通信设备与第一通信设备之间的信息传输;
    向所述第一通信设备发送所述第一信息。
  10. 根据权利要求9所述的下行波束预测方法,其中,所述数据集中包含以下一项或多项:
    数据集标识;
    一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
  11. 根据权利要求9所述的下行波束预测方法,其中,所述RS配置中包含以下一项或多项:
    RS配置标识;
    属于测量集合的第一RS配置;
    属于预测集合的第二RS配置;
    测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  12. 根据权利要求11所述的下行波束预测方法,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS 标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  13. 根据权利要求9所述的下行波束预测方法,其中,所述波束描述信息中包含以下一项或多项:
    波束标识;
    RS标识;
    网络设备的天线配置信息;
    波束的角度信息;
    波束的宽度信息。
  14. 根据权利要求9至13任一项所述的下行波束预测方法,其中,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
  15. 根据权利要求9至14任一所述的下行波束预测方法,其中,所述方法还包括:
    接收所述第一通信设备发送的第二信息;
    基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述第一通信设备之间进行信息传输的目标下行波束;
    其中,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的数据集的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的波束描述信息的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  16. 一种下行波束预测方法,应用于终端,包括:
    接收网络设备发送的参考信号RS配置;
    基于所述RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
    基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述网络设备与所述终端之间的信息传输。
  17. 根据权利要求16所述的下行波束预测方法,其中,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  18. 根据权利要求17所述的下行波束预测方法,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  19. 根据权利要求16至18任一所述的下行波束预测方法,其中,所述预测出一个或多个目标下行波束之后,所述方法还包括:
    向所述网络设备发送第二信息,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  20. 一种下行波束预测方法,应用于网络设备,包括:
    确定参考信号RS配置,所述RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述网络设备与终端之间的信息传输;
    向终端发送所述RS配置。
  21. 根据权利要求20所述的下行波束预测方法,其中,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  22. 根据权利要求21所述的下行波束预测方法,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  23. 根据权利要求20至22任一所述的下行波束预测方法,其中,所述方法还包括:
    接收终端发送的第二信息;
    基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述终端之间进行信息传输的目标下行波束;
    其中,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  24. 一种第一通信设备,包括存储器,收发机,处理器;
    存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
    接收第一信息,所述第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
    基于所述第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
    基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述第一通信设备与第二通信设备之间的信息传输。
  25. 根据权利要求24所述的第一通信设备,其中,所述接收第一信息,包括:
    接收所述第二通信设备发送的所述第一信息。
  26. 根据权利要求24或25所述的第一通信设备,其中,所述数据集中包含以下一项或多项:
    数据集标识;
    一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波 束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
  27. 根据权利要求24或25所述的第一通信设备,其中,所述RS配置中包含以下一项或多项:
    RS配置标识;
    属于测量集合的第一RS配置;
    属于预测集合的第二RS配置;
    测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  28. 根据权利要求27所述的第一通信设备,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  29. 根据权利要求24或25所述的第一通信设备,其中,所述波束描述信息中包含以下一项或多项:
    波束标识;
    RS标识;
    网络设备的天线配置信息;
    波束的角度信息;
    波束的宽度信息。
  30. 根据权利要求24所述的第一通信设备,其中,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
  31. 根据权利要求24至30任一所述的第一通信设备,其中,所述预测出一个或多个目标下行波束之后,所述操作还包括:
    向所述第二通信设备发送第二信息,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的数据集的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的波束描述信息的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  32. 一种第二通信设备,包括存储器,收发机,处理器;
    存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
    确定第一信息,所述第一信息包括数据集、RS配置、波束描述信息中的一种或多种;所述第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述第二通信设备与第一通信设备之间的信息传输;
    向所述第一通信设备发送所述第一信息。
  33. 根据权利要求32所述的第二通信设备,其中,所述数据集中包含以下一项或多项:
    数据集标识;
    一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标 识。
  34. 根据权利要求32所述的第二通信设备,其中,所述RS配置中包含以下一项或多项:
    RS配置标识;
    属于测量集合的第一RS配置;
    属于预测集合的第二RS配置;
    测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  35. 根据权利要求34所述的第二通信设备,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  36. 根据权利要求32所述的第二通信设备,其中,所述波束描述信息中包含以下一项或多项:
    波束标识;
    RS标识;
    网络设备的天线配置信息;
    波束的角度信息;
    波束的宽度信息。
  37. 根据权利要求32至36任一项所述的第二通信设备,其中,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
  38. 根据权利要求32至37任一所述的第二通信设备,其中,所述操作还包括:
    接收所述第一通信设备发送的第二信息;
    基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述第一通信设备之间进行信息传输的目标下行波束;
    其中,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的数据集的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的波束描述信息的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  39. 一种终端,包括存储器,收发机,处理器;
    存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
    接收网络设备发送的参考信号RS配置;
    基于所述RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
    基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述网络设备与所述终端之间的信息传输。
  40. 根据权利要求39所述的终端,其中,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  41. 根据权利要求40所述的终端,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  42. 根据权利要求39至41任一所述的终端,其中,所述预测出一个或多个目标下行波束之后,所述操作还包括:
    向所述网络设备发送第二信息,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  43. 一种网络设备,包括存储器,收发机,处理器;
    存储器,用于存储计算机程序;收发机,用于在所述处理器的控制下收发数据;处理器,用于读取所述存储器中的计算机程序并执行以下操作:
    确定参考信号RS配置,所述RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束, 所述目标下行波束用于所述网络设备与终端之间的信息传输;
    向终端发送所述RS配置。
  44. 根据权利要求43所述的网络设备,其中,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  45. 根据权利要求43所述的网络设备,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  46. 根据权利要求43至45任一所述的网络设备,其中,所述操作还包括:
    接收终端发送的第二信息;
    基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述终端之间进行信息传输的目标下行波束;
    其中,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  47. 一种下行波束预测装置,应用于第一通信设备,包括:
    第一接收单元,用于接收第一信息,所述第一信息包括数据集、参考信号RS配置、波束描述信息中的一种或多种;
    第一模型单元,用于基于所述第一信息,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
    第一预测单元,用于基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述第一通信设备与第二通信设备之间的信息传输。
  48. 根据权利要求47所述的下行波束预测装置,其中,所述接收第一信息,包括:
    接收所述第二通信设备发送的所述第一信息。
  49. 根据权利要求47或48所述的下行波束预测装置,其中,所述数据集中包含以下一项或多项:
    数据集标识;
    一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的RS测量结果以及一个或多个作为预测标签的波束标识。
  50. 根据权利要求47或48所述的下行波束预测装置,其中,所述RS配置中包含以下一项或多项:
    RS配置标识;
    属于测量集合的第一RS配置;
    属于预测集合的第二RS配置;
    测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  51. 根据权利要求50所述的下行波束预测装置,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  52. 根据权利要求47或48所述的下行波束预测装置,其中,所述波束描述信息中包含以下一项或多项:
    波束标识;
    RS标识;
    网络设备的天线配置信息;
    波束的角度信息;
    波束的宽度信息。
  53. 根据权利要求47所述的下行波束预测装置,其中,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
  54. 根据权利要求47至53任一所述的下行波束预测装置,其中,所述预测出一个或多个目标下行波束之后,所述装置还包括:
    第一发送单元,用于向所述第二通信设备发送第二信息,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的数据集的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的波束描述信息的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  55. 一种下行波束预测装置,应用于第二通信设备,包括:
    第一确定单元,用于确定第一信息,所述第一信息包括数据集、RS配置、波束描述信息中的一种或多种;所述第一信息用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述第二通信设备与第一通信设备之间的信息传输;
    第二发送单元,用于向所述第一通信设备发送所述第一信息。
  56. 根据权利要求55所述的下行波束预测装置,其中,所述数据集中包含以下一项或多项:
    数据集标识;
    一个或多个数据集样本,所述数据集样本中包括多个波束标识、每个波束标识对应波束的参考信号测量结果以及一个或多个作为预测标签的波束标识。
  57. 根据权利要求55所述的下行波束预测装置,其中,所述RS配置中包含以下一项或多项:
    RS配置标识;
    属于测量集合的第一RS配置;
    属于预测集合的第二RS配置;
    测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  58. 根据权利要求57所述的下行波束预测装置,其中,所述测量集合图 样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  59. 根据权利要求55所述的下行波束预测装置,其中,所述波束描述信息中包含以下一项或多项:
    波束标识;
    RS标识;
    网络设备的天线配置信息;
    波束的角度信息;
    波束的宽度信息。
  60. 根据权利要求55至59任一项所述的下行波束预测装置,其中,所述第一信息与应用场景、网络设备配置、模型功能中的一项或多项相关联。
  61. 根据权利要求55至60任一所述的下行波束预测装置,其中,所述装置还包括:
    第二接收单元,用于接收所述第一通信设备发送的第二信息;
    基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述第一通信设备之间进行信息传输的目标下行波束;
    其中,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的数据集的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的波束描述信息的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  62. 一种下行波束预测装置,应用于终端,包括:
    第三接收单元,用于接收网络设备发送的参考信号RS配置;
    第二模型单元,用于基于所述RS配置,对用于下行波束预测的人工智能或机器学习AI/ML模型进行训练或更新;
    第二预测单元,用于基于训练或更新后的所述AI/ML模型,预测出一个或多个目标下行波束,所述目标下行波束用于所述网络设备与所述终端之间的信息传输。
  63. 根据权利要求62所述的下行波束预测装置,其中,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  64. 根据权利要求63所述的下行波束预测装置,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  65. 根据权利要求62至64任一所述的下行波束预测装置,其中,所述预测出一个或多个目标下行波束之后,所述装置还包括:
    第三发送单元,用于向所述网络设备发送第二信息,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  66. 一种下行波束预测装置,应用于网络设备,包括:
    第二确定单元,用于确定参考信号RS配置,所述RS配置用于对人工智能或机器学习AI/ML模型进行训练或更新,所述AI/ML模型用于预测一个或多个目标下行波束,所述目标下行波束用于所述网络设备与终端之间的信息传输;
    第四发送单元,用于向终端发送所述RS配置。
  67. 根据权利要求66所述的下行波束预测装置,其中,所述RS配置中包含RS配置标识、属于预测集合的第二RS配置以及测量集合图样,所述测量集合图样用于指示属于测量集合的第一RS配置;
    其中,所述第一RS配置用于获取RS测量结果作为模型输入;所述第二RS配置用于确定一个或多个下行波束作为预测标签。
  68. 根据权利要求67所述的下行波束预测装置,其中,所述测量集合图样包括以下一种或多种:
    第一图样,所述第一图样用于指示属于预测集合的每个第二RS配置是否属于测量集合;
    第二图样,所述第二图样用于指示属于测量集合的RS配置所对应的RS 标识;
    第三图样,所述第三图样用于指示属于测量集合的RS配置所满足的RS序号排列规律;
    第四图样,所述第四图样用于指示属于测量集合的RS配置所对应的波束标识;
    第五图样,所述第五图样用于指示属于测量集合的RS配置所满足的波束序号排列规律。
  69. 根据权利要求66至68任一所述的下行波束预测装置,其中,所述装置还包括:
    第四接收单元,用于接收终端发送的第二信息;
    基于所述第二信息,以及波束标识和物理波束之间的映射关系,确定用于与所述终端之间进行信息传输的目标下行波束;
    其中,所述第二信息中包含以下一项或多项:
    所述目标下行波束对应的波束标识;
    所述目标下行波束对应的RS标识;
    用于预测所述目标下行波束的AI/ML模型的标识;
    用于预测所述目标下行波束的RS配置的标识;
    用于预测所述目标下行波束的AI/ML模型的输入波束数量与输出波束数量信息。
  70. 一种计算机可读存储介质,所述计算机可读存储介质存储有计算机程序,所述计算机程序用于使计算机执行权利要求1至8任一项所述的方法,或执行权利要求9至15任一项所述的方法,或执行权利要求16至19任一项所述的方法,或执行权利要求20至23任一项所述的方法。
PCT/CN2023/118271 2022-09-30 2023-09-12 下行波束预测方法、设备、装置及存储介质 WO2024067067A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211215784.0A CN117858251A (zh) 2022-09-30 2022-09-30 下行波束预测方法、设备、装置及存储介质
CN202211215784.0 2022-09-30

Publications (1)

Publication Number Publication Date
WO2024067067A1 true WO2024067067A1 (zh) 2024-04-04

Family

ID=90476061

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/118271 WO2024067067A1 (zh) 2022-09-30 2023-09-12 下行波束预测方法、设备、装置及存储介质

Country Status (2)

Country Link
CN (1) CN117858251A (zh)
WO (1) WO2024067067A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3496290A1 (en) * 2016-09-30 2019-06-12 China Academy of Telecommunications Technology Beam processing method, base station, and mobile terminal
CN112073106A (zh) * 2020-08-14 2020-12-11 清华大学 毫米波波束预测方法及装置、电子设备、可读存储介质
CN113783593A (zh) * 2021-07-30 2021-12-10 中国信息通信研究院 一种基于深度强化学习的波束选择方法和系统
WO2022069054A1 (en) * 2020-10-01 2022-04-07 Telefonaktiebolaget Lm Ericsson (Publ) Adaptive beam management in telecommunications network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3496290A1 (en) * 2016-09-30 2019-06-12 China Academy of Telecommunications Technology Beam processing method, base station, and mobile terminal
CN112073106A (zh) * 2020-08-14 2020-12-11 清华大学 毫米波波束预测方法及装置、电子设备、可读存储介质
WO2022069054A1 (en) * 2020-10-01 2022-04-07 Telefonaktiebolaget Lm Ericsson (Publ) Adaptive beam management in telecommunications network
CN113783593A (zh) * 2021-07-30 2021-12-10 中国信息通信研究院 一种基于深度强化学习的波束选择方法和系统

Also Published As

Publication number Publication date
CN117858251A (zh) 2024-04-09

Similar Documents

Publication Publication Date Title
JP2021501537A (ja) ビーム検出の方法および装置
WO2018059470A1 (zh) 传输信息的方法和设备
WO2022042294A1 (zh) 波束指示方法、网络设备、终端、装置及存储介质
WO2024093656A1 (zh) 一种无线资源管理rrm测量的配置方法及装置
WO2022237637A1 (zh) 一种信息处理方法、装置、终端及网络设备
WO2024067067A1 (zh) 下行波束预测方法、设备、装置及存储介质
CN114759964A (zh) 一种信息处理方法、装置及网络侧设备
WO2024067098A1 (zh) 模型信息上报方法、设备、装置及存储介质
WO2024169462A1 (zh) 模型监控方法、装置、终端及网络侧设备
WO2018137698A1 (zh) 功率控制方法及装置
WO2024140078A1 (zh) 连续条件小区接入方法、装置及存储介质
WO2024046092A1 (zh) 目标小区的选择方法、配置方法、信息传输方法及设备
WO2024027642A1 (zh) 一种信息传输方法、装置及设备
WO2024066620A1 (zh) 波束指示方法、装置及存储介质
WO2024046272A1 (zh) 信息传输方法、装置及存储介质
WO2023151453A1 (zh) 信息传输方法、装置及存储介质
WO2024152826A1 (zh) 波束失败检测方法、装置、终端、第一节点及网络设备
WO2024099243A1 (zh) 模型监测方法、装置、终端及网络侧设备
WO2024066866A1 (zh) 小区切换方法、终端、网络设备、装置和存储介质
WO2024032807A1 (zh) 信息传输方法、装置、终端及网络侧设备
WO2024139264A1 (zh) 波束指示方法、装置及设备
WO2024093624A1 (zh) 测量方法、装置、终端及网络设备
WO2024032764A1 (zh) 信息处理方法、装置、节点设备及介质
WO2024032308A1 (zh) Ta传输的方法及装置
WO2022206999A1 (zh) 定位方法、装置及存储介质

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: 23870305

Country of ref document: EP

Kind code of ref document: A1