US 12,250,258 B2
Optimized model transmission
Bin Sun, Luleå (SE); Yifei Jin, Solna (SE); and Rafia Inam, Västerås (SE)
Assigned to TELEFONAKTIEBOLAGET LM ERICSSON (PUBL), Stockholm (SE)
Appl. No. 17/918,241
Filed by Telefonaktiebolaget LM Ericsson (publ), Stockholm (SE)
PCT Filed May 12, 2020, PCT No. PCT/EP2020/063190
§ 371(c)(1), (2) Date Oct. 11, 2022,
PCT Pub. No. WO2021/228373, PCT Pub. Date Nov. 18, 2021.
Prior Publication US 2023/0148106 A1, May 11, 2023
Int. Cl. H04L 65/752 (2022.01); H04L 67/14 (2022.01); H04W 28/08 (2023.01)
CPC H04L 65/752 (2022.05) [H04L 67/14 (2013.01); H04W 28/0925 (2020.05); H04W 28/0967 (2020.05)] 12 Claims
OG exemplary drawing
 
1. A method for operating a policy control entity in a cellular network, the method comprising:
determining a quality-of-service (QoS) parameter for a data packet session in which one trained model from a plurality of trained models is downloaded to a mobile entity, wherein the plurality of trained models differ from one another by a data size and a number of features used by the trained model to carry out a certain task;
determining at least one capacity parameter of the mobile entity describing a processing capacity of the mobile entity;
determining a network transmission parameter of the cellular network describing transmission capabilities of the cellular network for transmitting the one trained model;
determining the one trained model from the plurality of trained models based on a dataset which maps different capacity parameters and transmission capabilities to the plurality of trained models, on the determined capacity parameter, and on the network transmission parameter;
determining routing information indicating where the one trained model is accessible for a transmission to the mobile entity; and
transmitting the routing information to a session management entity configured to manage the data packet session in the cellular network;
wherein the plurality of trained models differs from one another by the number of features and by an amount of compression of the features with which the corresponding features of the plurality of trained models are transmitted through the cellular network;
wherein the dataset indicates a compression parameter in dependence on the network transmission parameter and the capacity parameter;
wherein different importance levels in at least some of the plurality of trained models are given to the different features; and
wherein the dataset indicates that features with a higher importance level are to be transmitted to the mobile entity with a lower compression compared to the features with a lower importance level.