US 12,438,575 B2
Apparatuses and methods for spatial beam prediction with multiple assistance information
Amir Ahmadian Tehrani, Munich (DE); Sajad Rezaie, Aalborg (DK); and Keeth Saliya Jayasinghe Laddu, Espoo (FI)
Assigned to Nokia Technologies Oy, Espoo (FI)
Filed by Nokia Technologies Oy, Espoo (FI)
Filed on Nov. 13, 2023, as Appl. No. 18/507,852.
Claims priority of application No. 20226023 (FI), filed on Nov. 14, 2022.
Prior Publication US 2024/0162943 A1, May 16, 2024
Int. Cl. H04B 7/02 (2018.01); G06N 20/00 (2019.01); H04B 7/0417 (2017.01); H04B 7/0456 (2017.01); H04B 7/06 (2006.01)
CPC H04B 7/0417 (2013.01) [G06N 20/00 (2019.01); H04B 7/046 (2013.01); H04B 7/063 (2013.01)] 9 Claims
OG exemplary drawing
 
1. An apparatus, comprising:
at least one processor; and
at least one memory comprising instructions which, when executed by the at least one processor, cause the apparatus at least to:
configure subspaces and corresponding machine learning models trained for spatial beam prediction based on a dataset comprising multiple assistance information parameters associated with a base station or a client node, wherein a number of the configured subspaces is based on a number of the multiple assistance information parameters, and wherein a plurality of the configured subspaces comprise at least one of the multiple assistance information parameters;
select a first subset of the configured subspaces based on predefined selection probabilities of the configured subspaces;
send a request to report one or more of the multiple assistance information parameters to the base station or to the client node;
obtain first inputs for the machine learning models, wherein the first inputs comprise beam measurements;
obtain second inputs for the machine learning models, wherein the second inputs comprise available assistance information parameters received from the base station or the client node based on the request;
execute one or more machine learning models based on the first inputs and the second inputs for the selected first subset of the configured subspaces, wherein subspaces comprised in the plurality of configured subspaces and not comprising one or more of the available assistance information parameters are excluded from the execution;
select a preferred output for the spatial beam prediction among outputs of the executed machine learning models based on a predetermined selection method; and
determine at least one of a downlink transmission beam alignment or a reception beam alignment based on the preferred output.