Granted Patent B2

US 12,489,686 · App. 18/284,907 · Granted Dec 2, 2025

Inventors: Rerngvit Yanggratoke (Järfälla, SE); Serveh Shalmashi (Enebyberg, SE); Paul Stjernholm (Lidingö, SE)

Assignee: TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

H04L41/147H04W24/02H04W28/24

Abstract

Claims (77)

1 . A method for a feasibility check of a new radio access network (RAN) slice performed by a network node in a telecommunications network, the method comprising:

computing a first estimate of occupied resources for each resource within at least one of the network node and a cell of the network node, the first estimate performed for each resource in each interval of time in a plurality of intervals of time using historical data of a measurement of utilization of each resource;

determining a second estimate of resource demand for each resource in each interval of time in the plurality of intervals of time for the new RAN slice using a trained supervised machine learning model;

determining whether to admit the new RAN slice based on a comparison of a sum of the first estimate and the second estimate for each interval of time and each resource to a corresponding defined threshold value per resource; and

admitting the new RAN slice when, for each resource and for each interval in the plurality of intervals, the sum of the first estimate and the second estimate has a value that does not exceed the corresponding defined threshold value.

2 . The method of claim 1 , further comprising:

rejecting the new RAN slice based on the sum of the first estimate and the second estimate for an interval of time in the plurality of intervals of time having a value greater than the corresponding defined threshold value.

3 . The method of claim 1 , wherein each resource comprises at least one of:

a hardware resource;

a processing resource; and

a radio resource.

4 . The method of claim 1 , wherein the historical data comprises at least one of:

configuration management information for a plurality of network nodes in the telecommunications network;

performance management information for the plurality of network nodes in the telecommunications network; and

traffic recordings for the plurality of network nodes in the telecommunications network.

5 . The method of claim 1 , further comprising:

extracting the historical data from:

at least one of:

the network node; and

at least one cell of the network node; and

data for the new RAN slice, the data including a RAN slice profile.

6 . The method of claim 5 , wherein the RAN slice profile for the new RAN slice comprises data including a performance requirement for a communication device per at least one of:

a single network slice selection assistance indicator (S-NSSAI);

a specified quality of service class identifier (QCI); and

a specified 5G quality of service class identifier (5QI).

7 . The method of claim 1 , further comprising:

collecting historical data for each of a plurality of network nodes and each of a corresponding cell for each of the plurality of network nodes in a live telecommunications network.

8 . The method of claim 7 , further comprising:

identifying the historical data providing a value for each resource and each interval of time in the plurality of intervals of time; and

merging, into merged data, the identified historical data with a plurality of RAN slice profiles, each historical data being associated with a historical RAN slice and a corresponding RAN slice profile.

9 . The method of claim 8 , further comprising:

dividing the merged data into two portions of divided data comprising a first portion of data from the merged data prior to deployment of the historical RAN slice, and a second portion of data from the merged data subsequent to deployment of the historical RAN slice.

10 . The method of claim 9 , further comprising:

storing the divided data in a centralized location.

11 . The method of claim 1 , further comprising:

training the supervised machine learning model to estimate resource demand for a new RAN slice for the telecommunications network across a plurality of network nodes and cells within the plurality of network nodes, the estimate performed for each resource in each interval of time in the plurality of intervals of time using a historical measurement of characteristics from at least one RAN slice across a plurality of network nodes and cells within the plurality of network nodes, each historical measurement related to a historical RAN slice profile, the cell, and a RAN slice profile, the measurement having two data sources including a first data measurement recorded before a historical RAN slice is deployed and a second data measurement recorded after the historical RAN slice is deployed.

12 . The method of claim 11 , wherein training the supervised machine learning model comprises:

inputting data to the supervised machine learning model, the data comprising input features including:

first data describing a property and a characteristic of the at least one RAN slice profile;

second data describing a property and a characteristic of the network node; and

third data describing a property and a characteristic of a cell of the network node;

estimating the resource demand for the new RAN slice for the telecommunications network for the cell of the network node and the network node using the input features; and

outputting a variable, the variable comprising the estimated resource demand related to the new RAN slice executing on the network node and the cell.

13 . The method of claim 1 , wherein the telecommunications network comprises a 5th generation RAN,

wherein the network node is a gNodeB network node,

wherein the cell is a radio cell of the gNodeB, and

wherein the RAN slice is a network slice subnet for the RAN.

14 . A network node in a telecommunications network, the network node comprising:

processing circuitry; and

memory coupled with the processing circuitry, wherein the memory includes instructions that when executed by the processing circuitry causes the network node to perform operations for a feasibility check of a new radio access network (RAN) slice, the operations comprising:

computing a first estimate of occupied resources for each resource within at least one of the network node and a cell of the network node, the first estimate performed for each resource in each interval of time in a plurality of intervals of time using historical data of a measurement of utilization of each resource;

determining a second estimate of resource demand for each resource in each interval of time in the plurality of intervals of time for the new RAN slice using a trained supervised machine learning model;

determining whether to admit the new RAN slice based on a comparison of a sum of the first estimate and the second estimate for each interval of time and each resource to a corresponding defined threshold value per resource; and

admitting the new RAN slice when, for each resource and for each interval in the plurality of intervals, the sum of the first estimate and the second estimate has a value that does not exceed the corresponding defined threshold value.

15 . The network node of claim 14 , the operations further comprising:

rejecting the new RAN slice based on the sum for an interval of time in the plurality of intervals of time having a value greater than the defined threshold value.

16 . A non-transitory computer-readable medium having instructions stored therein that are executable by processing circuitry of a network node for a telecommunications network, whereby execution of the instructions cause the network node to perform operations for a feasibility check of a new radio access network (RAN) slice comprising:

computing a first estimate of occupied resources for each resource within at least one of the network node and a cell of the network node, the first estimate performed for each resource in each interval of time in a plurality of intervals of time using historical data of a measurement of utilization of each resource;

determining a second estimate of resource demand for each resource in each interval of time in the plurality of intervals of time for the new RAN slice using a trained supervised machine learning model;

determining whether to admit the new RAN slice based on a comparison of a sum of the first estimate and the second estimate for each interval of time and each resource to a corresponding defined threshold value per resource; and

admitting the new RAN slice when, for each resource and for each interval in the plurality of intervals, the sum of the first estimate and the second estimate has a value that does not exceed the corresponding defined threshold value.

17 . The non-transitory computer-readable medium of claim 16 , the operations further comprising:

rejecting the new RAN slice based on the sum for an interval of time in the plurality of intervals of time having a value greater than the corresponding defined threshold value.

18 . The network node of claim 14 , wherein each resource comprises at least one of:

a hardware resource;

a processing resource; and

a radio resource.

19 . The network node of claim 14 , wherein the historical data comprises at least one of:

configuration management information for a plurality of network nodes in the telecommunications network;

performance management information for the plurality of network nodes in the telecommunications network; and

traffic recordings for the plurality of network nodes in the telecommunications network.

20 . The network node of claim 14 , the operations further comprising:

extracting the historical data from:

at least one of:

the network node; and

at least one cell of the network node; and

data for the new RAN slice, the data including a RAN slice profile.

Assignments (1)

Continuity (1)

References Cited (18)