# US 2016/0366565 A1 · Grouping wireless devices in a communications network

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2016/0366565 A1
- Document ID: US20160366565A1
- Status: Published application
- Kind: A1
- Application number: 14/855,197
- Filed: 2015-09-15
- Published: 2016-12-15
- Earliest claimed: 2015-06-12
- Assignees: Telefonaktiebolaget L M Ericsson (publ)
- Applicants: Telefonaktiebolaget L M Ericsson (Publ)
- Inventors: Marianne FJELBERG; Richard Cöster
- Web page: https://patentgrep.com/patent/US20160366565A1

## CPC

- H04W 4/08: User group management
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/20: Ensemble learning
- H04W 8/186: {Processing of subscriber group data}
- H04W 24/02: Arrangements for optimising operational condition
- G06N 20/00: Machine learning
- H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

There is provided mechanisms for grouping wireless devices of a cell of a communications network. A method is performed by a system in the communications network. The method comprises acquiring user network data for wireless devices in a cell of the communications network and cell network data for the cell. The method comprises determining groups for the wireless devices and assigning each one of the wireless devices to one of the groups using unsupervised machine learning with the user network data and the cell network data as input. The method comprises assigning a wireless device entering the cell to one of the groups using supervised machine learning with user network data of said wireless device and the cell network data as input.

## Claims

### Claim 1

A method for grouping wireless devices of a cell of a communications network, the method being performed by a system in the communications network, the method comprising at least one of the following steps:
acquiring user network data for wireless devices in a cell of the communications network and cell network data for the cell;
determining groups for the wireless devices and assigning each one of the wireless devices to one of the groups using unsupervised machine learning with the user network data and the cell network data as input; and
assigning a wireless device entering the cell to one of the groups using supervised machine learning with user network data of said wireless device and the cell network data as input.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the unsupervised machine learning comprises finding candidate groups describable by a weighted combination of multivariate Gaussian distributions of the user network data and the cell network data.

### Claim 3 (depends on 1)

The method according to claim 1, wherein the unsupervised machine learning comprises at least one of the following steps:
selecting a number K of groups;
assigning repeatedly until convergence or for a fixed number of times, each one of the wireless devices to one of the groups by one or more of the following steps:
determining an assignment score for each wireless device for each group; and
determining adjusted properties of the groups based on the assignment score.

### Claim 4 (depends on 1)

The method according to claim 1, wherein gap statistics is used to find which number of groups K to use for classifying the wireless devices.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the wireless devices are clustered and/or classified into five groups.

### Claim 6 (depends on 1)

The method according to claim 1, wherein a Gaussian Mixture Model, GMM, is used during the unsupervised learning.

### Claim 7 (depends on 1)

The method according to claim 1, wherein the supervised machine learning comprises at least one of the following steps:
learning a machine learning multi-class classification model using a training set of data to map the user network data to the correct group;
tuning parameters of the machine learning multi-class classification model using a validation set of data; and
determining a final correctness of the machine learning multi-class classification model using a testing set of data.

### Claim 8 (depends on 1)

The method according to claim 1, wherein the supervised machine learning is based on constructing at least one decision tree.

### Claim 9 (depends on 1)

The method according to claim 1, wherein Random Forest classification is used during the supervised learning.

### Claim 10 (depends on 1)

The method according to claim 1, wherein each group represents a unique set of radio characteristics values.

### Claim 11 (depends on 1)

The method according to claim 1, wherein the user network data comprises current user throughput, previous user throughput, channel quality indicator, CQI, rank, handover information, time to handover, or any combination thereof.

### Claim 12 (depends on 1)

The method according to claim 1, wherein the cell network data comprises cell throughput, number of active wireless devices in the cell, number of wireless devices in queue of the cell, or any combination thereof.

### Claim 13 (depends on 11)

The method according to claim 11, wherein the user network data comprises current user throughput, CQI, time to handover, and wherein the cell network data comprises cell throughput, and number of active users.

### Claim 14 (depends on 1)

The method according to claim 1, wherein determining the groups comprises one or more of the following steps:
determining which user network data and which cell network data to use for classifying the wireless devices and assigning said each one of the wireless devices to one of the groups.

### Claim 15 (depends on 1)

The method according to claim 1, further comprising:
acquiring auxiliary data such as time indication, wireless device identity information, wireless device group information, wireless device velocity information, speed information, radio access network node information, cell information, position information, or any combination thereof.

### Claim 16 (depends on 15)

The method according to claim 15, further comprising:
scaling the auxiliary data so as to normalize all values of the auxiliary data.

### Claim 17 (depends on 1)

The method according to claim 1, further comprising:
scaling the user network data and the cell network data so as to normalize all values of the user network data and the cell network data.

### Claim 18 (depends on 16)

The method according to claim 16, further comprising:
adding an offset value to all normalized values.

### Claim 19 (depends on 1)

The method according to claim 1, further comprising:
predicting, for wireless devices of one of the groups, into which other group said wireless devices of said one of the groups are most likely to enter.

### Claim 20 (depends on 19)

The method according to claim 19, wherein Markov Chains are used during said predicting.

### Claim 21 (depends on 1)

The method according to claim 1, wherein current user throughput is excluded from the user network data, the method further comprising:
predicting current user throughput for wireless devices of one of the groups.

### Claim 22 (depends on 1)

The method according to claim 1, further comprising:
determining a load balancing action for wireless devices of one of the groups with the user network data as input.

### Claim 23 (depends on 1)

The method according to claim 1, further comprising:
determining whether to perform carrier aggregation for wireless devices of one of the groups with the user network data as input.

### Claim 24

A system for grouping wireless devices of a cell in a communications network, the system comprising processing circuitry, the processing circuitry being configured to cause the system to perform a set of operations comprising:
acquire user network data for wireless devices in a cell of the communications network and cell network data for the cell;
determine groups of the wireless devices and assign each one of the wireless devices to one of the groups using unsupervised machine learning with the user network data and the cell network data as input; and
assign a wireless device entering the cell to one of the groups using supervised machine learning with user network data of said wireless device and the cell network data as input.

### Claim 25 (depends on 24)

The system according to claim 24, further comprising a storage medium storing said set of operations, and wherein the processing circuitry configured to retrieve said set of operations from the storage medium to cause the system to perform said set of operations.

### Claim 26 (depends on 24)

The system according to claim 24, wherein said set of operations is provided as a set of executable instructions.

### Claim 27

A system for grouping wireless devices of a cell in a communications network, the system comprising:
processing circuitry; and
a computer program product storing instructions that, when executed by the processing circuitry, causes the system to:
cause acquiring of user network data for wireless devices in a cell of the communications network and cell network data for the cell;
cause determination of groups of the wireless devices and cause assignment of each one of the wireless devices to one of the groups using unsupervised machine learning with the user network data and the cell network data as input; and
cause assignment of a wireless device entering the cell to one of the groups using supervised machine learning with user network data of said wireless device and the cell network data as input.

### Claim 28

A system for grouping wireless devices of a cell in a communications network, the system comprising:
an acquire module for acquiring user network data for wireless devices in a cell of the communications network and cell network data for the cell;
a determine module for determining groups of the wireless devices and assigning each one of the wireless devices to one of the groups using unsupervised machine learning with the user network data and the cell network data as input; and
an assign module for assigning a wireless device entering the cell to one of the groups using supervised machine learning with user network data of said wireless device and the cell network data as input.

### Claim 29

A computer program for grouping wireless devices of a cell in a communications network, the computer program comprising computer code which, when run on processing circuitry of a system, causes the system to:
acquire user network data for wireless devices in a cell of the communications network and cell network data for the cell;
determine groups for the wireless devices and assign each one of the wireless devices to one of the groups using unsupervised machine learning with the user network data and the cell network data as input; and
assign a wireless device entering the cell to one of the groups using supervised machine learning with user network data of said wireless device and the cell network data as input.

### Claim 30 (depends on 27)

A computer program product comprising a computer program according to claim 27, and a non-transitory computer readable storage medium on which the computer program is stored.
