US 12,250,071 B2
Link adaptation optimized with machine learning
Christian Skärby, Stockholm (SE); Henrik Nyberg, Stockholm (SE); Raimundas Gaigalas, Hässelby (SE); and Tor Kvernvik, Täby (SE)
Assigned to TELEFONAKTIEBOLAGET LM ERICSSON (PUBL), Stockholm (SE)
Appl. No. 17/440,209
Filed by Telefonaktiebolaget LM Ericsson (publ), Stockholm (SE)
PCT Filed Mar. 18, 2019, PCT No. PCT/SE2019/050238
§ 371(c)(1), (2) Date Sep. 17, 2021,
PCT Pub. No. WO2020/190181, PCT Pub. Date Sep. 24, 2020.
Prior Publication US 2022/0149980 A1, May 12, 2022
Int. Cl. H04L 1/00 (2006.01); G06N 3/02 (2006.01)
CPC H04L 1/0017 (2013.01) [G06N 3/02 (2013.01); H04L 1/0003 (2013.01); H04L 1/0009 (2013.01); H04L 1/0026 (2013.01); H04L 1/0034 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A method for dynamically selecting a link adaptation policy (LAP), the method comprising:
generating a machine learning (ML) model, wherein generating the ML model comprises providing training data to an ML algorithm;
a first transmission point (TRP) transmitting first data to a user equipment (UE) using a first LAP, wherein the first TRP serves at least a first cell;
receiving a channel quality report transmitted by the UE, the channel quality report comprising channel quality information indicating a quality of a channel between the UE and the first TRP;
obtaining additional information, wherein the additional information comprises: neighbor cell information about a second cell served by a second TRP, distance information indicating a distance between the UE and the first TRP, and/or gain information indicating a radio propagation gain between the UE and the serving node;
using the channel quality information, the additional information, and the ML model to select a LAP from a set of predefined LAPs, the set of predefined LAPs comprising the first LAP and a second LAP; and
the first TRP transmitting second data to the UE using the selected LAP, wherein
the set of predefined LAPs is a set of predefined block error rate (BLER) targets,
the set of predefined BLER targets comprises a first BLER target and a second BLER target, and
using the channel quality information (CQI), the additional information, and the ML model to select a LAP from the set of predefined LAPs comprises:
i) inputting the COI and the additional information into the ML model, wherein the ML model is configured to use the inputted COI and additional information to produce a predicted value for the first BLER target;
ii) determining whether the predicted value for the first BLER target is greater than a predicted value for the second BLER target; and
iii) selecting the first BLER target as a result of determining that the predicted value for the first BLER target is greater than the predicted value for the second BLER target.