Devices, methods and apparatuses for channel prediction
Embodiments of the present disclosure disclose devices, methods and apparatuses for channel prediction. A terminal device obtains model configuration information. The model configuration information indicates a first prediction model for channel prediction. The terminal device receives overfitting configuration information from a network device. The overfitting configuration information indicates configuration on reference signals for overfitting by the first prediction model. The terminal device reports channel information predicted based on the first prediction model and the overfitting configuration information.
1 . A terminal device comprising:
at least one processor; and
at least one memory including computer program codes, wherein the at least one memory and the computer program codes are configured to, with the at least one processor, cause the terminal device to:
obtain model configuration information, wherein the model configuration information indicates a first prediction model for channel prediction;
receive overfitting configuration information from a network device, wherein the overfitting configuration information indicates configuration on reference signals for overfitting the first prediction model; and
report, to the network device, channel information predicted based on the first prediction model and the overfitting configuration information.
2 . The terminal device of claim 1 , wherein the overfitting configuration information comprises one or more of:
configuration of channel information reference signals for overfitting the first prediction model;
a time duration for overfitting the first prediction model;
a start point for overfitting the first prediction model;
configuration of bandwidth or bandwidth part for overfitting the first prediction model; or
configuration of component carriers for overfitting the first prediction model,
wherein model configuration information is obtained from the network device.
3 . The terminal device of claim 1 , wherein the terminal device is configured to:
re-train the first prediction model based on the overfitting configuration information; and
obtain the channel information by performing channel prediction based on the re-trained first prediction model.
4 . The terminal device of claim 3 , wherein the terminal device is caused to re-train the first prediction model by:
collecting a first set of channel measurements on the reference signals for a time duration as indicated by the overfitting configuration information; and
re-training the first prediction model based on the first set of channel measurements.
5 . The terminal device of claim 1 , wherein the terminal device is further caused to:
collect a second set of channel measurements for a first time duration; and
predict channel information using the first prediction model or a re-trained first model for a second time duration which is after the first time duration, based on the second set of channel measurement.
6 . The terminal device of claim 1 , wherein the terminal device is further caused to:
transmit to the network device an overfitting request for the first prediction model to re-train the first prediction model; and
receive an overfitting indication for the first prediction model from the network device, and wherein the overfitting indication indicates the terminal device to re-train the first prediction model.
7 . The terminal device of claim 6 , wherein the overfitting request comprises overfitting information, and wherein the overfitting information comprises one or more of:
information on a location of the terminal device,
information on the closeness value, or
information on a difference of channel information predicted using the first prediction model and corresponding measured channel information.
8 . The terminal device of claim 6 , wherein the overfitting request further comprises a requested time duration for overfitting the first prediction model, and wherein the terminal device is further caused to:
determine the time duration as indicated by the overfitting configuration information is less than the requested time duration; and
re-train the first prediction model in a number of iterations greater than that corresponding to the time duration as indicated by the overfitting configuration information.
9 . The terminal device of claim 1 , wherein the terminal device is further caused to:
obtain a difference of channel information predicted using the first prediction model and corresponding measured channel information;
transmit to the network device an overfitting request for the first prediction model based on the obtained difference and a predetermined threshold; or
transmit to the network device information on the obtained difference.
10 . The terminal device of claim 1 , wherein the model configuration information further indicates a second inference model for obtaining a closeness value of the first prediction model, and wherein the closeness value indicates prediction accuracy level of the first prediction model.
11 . The terminal device of claim 10 , wherein the terminal device is further caused to:
obtain, by using the second inference model, the closeness value based on corresponding measured channel information;
wherein the terminal device is further caused to:
transmit to the network device an overfitting request for overfitting the first prediction model based on the closeness value and another predetermined threshold; or
transmit to the network device information on the closeness value,
wherein the closeness value is obtained, by using the second inference model, further based on the channel information predicted using the first prediction model.
12 . The terminal device of claim 11 , wherein the closeness value is obtained further based on history closeness information and reference signal characteristics.
13 . A network device, comprising:
at least one processor; and
at least one memory including computer program codes, wherein the at least one memory and the computer program codes are configured to, with the at least one processor, cause the terminal device to:
transmit overfitting configuration information to a terminal device, wherein the overfitting configuration information indicates configuration on reference signals for overfitting a first prediction model; and
receive, from the terminal device, channel information predicted based on the first prediction model and the overfitting configuration information.
14 . The network device of claim 13 , the network device is further caused to:
transmit model configuration information to the terminal device, wherein the model configuration information indicates the first prediction model for channel prediction.
15 . The network device of claim 13 , wherein the overfitting configuration information comprises one or more of:
configuration of channel information reference signals for overfitting the first prediction model;
a time duration for overfitting the first prediction model;
a start point for overfitting the first prediction model;
configuration of bandwidth or bandwidth part for overfitting the first prediction model; or
configuration of component carriers for overfitting the first prediction model.
16 . The network device of claim 13 , wherein the network device is further caused to:
transmit the reference signals for overfitting the first prediction model to the terminal device based on the overfitting configuration information.
17 . The network device of claim 13 , wherein the model configuration information further indicates a second inference model for obtaining a closeness value of the first prediction model, and wherein the closeness value indicates prediction accuracy level of the first prediction model.
18 . The network device of claim 13 , wherein the network device is further caused to:
transmit an overfitting indication, wherein the overfitting indication informs the terminal device an opportunity for re-training the first prediction model,
wherein the overfitting configuration information is transmitted after the overfitting indication is transmitted.
19 . The network device of any of claim 13 , wherein the network device is further caused to:
receive, from a terminal device, an overfitting request for the first prediction model to re-train the first prediction model, and
wherein the overfitting configuration information is transmitted in response to receiving the overfitting request; and
wherein the overfitting request comprises overfitting information, and wherein the overfitting information comprises one or more of:
information on a difference of channel information predicted using the first prediction model and corresponding measured channel information;
a closeness value of the first prediction model; or
information on a location of the terminal device.
20 . A method at a terminal device, comprising:
obtaining model configuration information, wherein the model configuration information indicates a first prediction model for channel prediction;
receiving overfitting configuration information from a network device, wherein the overfitting configuration information indicates configuration on reference signals for overfitting the first prediction model; and
reporting, to the network device, channel information predicted based on the first prediction model and the overfitting configuration information.