US 12,452,740 B2
Communication system for machine learning metadata
Anna Pantelidou, Massy (FR); Cinzia Sartori, Pullach (DE); Malgorzata Tomala, Wroclaw (PL); and Hakon Helmers, Sceaux (FR)
Appl. No. 18/020,542
Filed by NOKIA TECHNOLOGIES OY, Espoo (FI)
PCT Filed Aug. 11, 2020, PCT No. PCT/FI2020/050524
§ 371(c)(1), (2) Date Feb. 9, 2023,
PCT Pub. No. WO2022/034259, PCT Pub. Date Feb. 17, 2022.
Prior Publication US 2023/0300686 A1, Sep. 21, 2023
Int. Cl. H04W 36/00 (2009.01); G06N 20/00 (2019.01)
CPC H04W 36/0033 (2013.01) [G06N 20/00 (2019.01); H04W 36/0083 (2013.01)] 1 Claim
OG exemplary drawing
 
1. An apparatus for a terminal, the apparatus comprising:
at least one processor; and
at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to perform:
signalling, to a network apparatus, metadata about at least one machine learning model accessible for execution and/or training by the terminal, wherein the metadata includes metadata about at least one machine learning model accessible for training and execution by the terminal, and wherein signalling comprises signalling the metadata as part of a measurement report to a source network apparatus for signalling to the network apparatus;
receiving signalling via a radio resource control connection reconfiguration message indicating whether or not the terminal should execute and train the at least one machine learning model after the terminal is handed over to the network apparatus, wherein the receiving comprises receiving for each of said at least one machine learning algorithms an indication whether or not the terminal should execute and train that machine learning model after the terminal is handed over to the network apparatus;
abandoning at least one of the at least one machine learning models in response to receiving signalling;
receiving a request to provide at least one of the machine learning algorithms to a network apparatus;
responding to the request with at least one of the requested machine learning models;
starting a timer after which any machine learning models which are not in execution will be erased;
transmitting an indication to the network apparatus that any machine learning models which are not in execution will be erased;
if the terminal receives an instruction from the network apparatus, when the timer expires, not erasing at least one of the machine learning models which are not in execution based on the instruction; and
if the terminal does not receive an instruction from the network apparatus, when the timer expires, erasing the machine learning models which are not in execution based on the instruction.