Radio frequency plan generation for network deployments

Patent metadata

Field	Value
Publication number	US-12452687-B2
Application number	US-202217582241-A
Country	US
Kind code	B2
Filing date	Jan 24, 2022
Priority date	Dec 3, 2021
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

Examples described herein relate to generation of radio frequency (RF) plans for network deployments. Examples described herein may receive an input RF plan with modified set of features of a network deployment area. A first machine learning (ML) model generates an intermediate RF plan indicating candidate AP locations based on the modified set of features and a first set of parameters. A second ML model determines a network optimization score for the intermediate RF plan. Based on the optimization score, the first set of parameters are optimized. The first ML model generates an output RF plan indicating optimized AP locations based on the optimized first set of parameters and the modified set of features.

First claim

We claim:

1. A non-transitory machine-readable medium comprising instructions that, when executed by a processing resource of a computing device, cause the computing device to:
receive a reference radio frequency (RF) plan and an input RF plan of a network deployment area, wherein the reference RF plan indicates original access point (AP) locations and an original set of features of the network deployment area, wherein the input RF plan indicates a modified set of features of the network deployment area, wherein the original set of features and the modified set of features comprise at least spatial features and network features associated with the network deployment area, and wherein the network features comprise AP metrics of throuqhput or bandwidth;
extract, using an image segmentation process implemented by a first machine learning (ML) model, a group of pixels from the input RF plan;
label, using the first ML model, a feature of each group of pixels from the input RF plan;
generate, using the first ML model, an intermediate RF plan based on the modified set of features, the label of the feature of each group of pixels, and a first set of parameters of the first ML model, wherein the intermediate RF plan indicates candidate AP locations and the modified set of features;
provide, by the first ML model to a second ML model, the intermediate RF plan;
determine, using the second ML model, a network optimization score for the intermediate RF plan based on the candidate AP locations and the modified set of features;
determine, using the second ML model, an error value indicating a difference in the network optimization score for the intermediate RF plan and the network optimization score for the reference RF plan;
provide, by the second ML model to the first ML model, the error value;
in response to a determination that the error value is greater than a threshold, determine an optimized first set of parameters of the first ML model based on the error value; and
generate, using the first ML model, an output RF plan indicating optimized AP locations and the modified set of features based on the optimized first set of parameters.

2. The non-transitory machine-readable medium of claim 1 , further comprising instructions to:
receive a new input RF plan indicating a new modified set of features of the network deployment area; and
generate a new output RF plan based on the new modified set of features and the optimized first set of parameters, wherein the new output RF plan indicates the optimized AP locations for the network deployment area.

3. The non-transitory machine-readable medium of claim 1 , wherein the instructions to generate the intermediate RF plan includes instructions to:
extract the modified set of features from the input RF plan based on a segmentation of the input RF plan;
predict, using the first ML model, the candidate AP locations based on the modified set of features extracted from the input RF plan;
encode the modified set of features extracted from the input RF plan into a latent space dataset based on the first set of parameters of the first ML model; and
generate the intermediate RF plan indicating the candidate AP locations by decoding the latent space dataset based on the first set of parameters of the first ML model.

4. The non-transitory machine-readable medium of claim 3 , wherein the instructions to extract the modified set of features from the input RF plan based on the segmentation of the input RF plan includes instructions to:
segment the input RF plan into a plurality of groups, wherein each group of the plurality of groups is associated with a property; and
classify each group of the plurality of groups into a feature of the modified set of features.

5. The non-transitory machine-readable medium of claim 1 , wherein the instructions to determine the network optimization score includes instructions to:
predict, using the second ML model, network metrics values based on the modified set of features and the candidate AP locations; and
compute the network optimization score based on the network metrics values.

6. The non-transitory machine-readable medium of claim 1 , wherein the spatial features include one or more of walls, rooms, open areas, partitions, or ceiling, and the network features include one or more of user density or AP metrics.

7. The non-transitory machine-readable medium of claim 1 , wherein the output RF plan is associated with one or more of a higher bandwidth, a higher throughput, a lower interference, a lower latency, a higher availability, a lower packet loss, or a higher connectivity, relative to a random RF plan for the network deployment area.

8. The non-transitory machine-readable medium of claim 1 , wherein the first ML model is a generative model and the second ML model is a discriminative model.

9. A device comprising:
a processing resource;
a machine-readable medium storing instructions that, when executed by the processing resource, cause the processing resource to:
receive a reference radio frequency (RF) plan and an input RF plan of a network deployment area, wherein the reference RF plan indicates original access point (AP) locations and an original set of features of the network deployment area, wherein the input RF plan indicates a modified set of features of the network deployment area, wherein the original set of features and the modified set of features comprise at least spatial features and network features associated with the network deployment area, and wherein the network features comprise AP metrics of throuqhput or bandwidth;

extract, using an image segmentation process implemented by a first machine learning (ML) model, a group of pixels from the input RF plan;
label, using the first ML model, a feature of each group of pixels from the input RF plan;

generate, using the first ML model, an intermediate RF plan based on the modified set of features, the label of the feature of each group of pixels, and a first set of parameters of the first ML model, wherein the intermediate RF plan indicates candidate AP locations and the modified set of features;
provide, by the first ML model to a second ML model, the intermediate RF plan;

determine, using the second ML model, a network optimization score for the intermediate RF plan based on the candidate AP locations and the modified set of features;
determine, using the second ML model, an error value indicating a difference in the network optimization score for the intermediate RF plan and the network optimization score for the reference RF plan;
provide, by the second ML model to the first ML model, the error value;

in response to a determination that the error value is greater than a threshold, determine an optimized first set of parameters of the first ML model based on the error value; and
generate, using the first ML model, an output RF plan indicating optimized AP locations and the modified set of features based on the optimized first set of parameters.

10. The device of claim 9 , wherein the processing resource executes one or more of the instructions to:
receive a new input RF plan indicating a new modified set of features of the network deployment area; and
generate a new output RF plan based on the new modified set of features and the optimized first set of parameters, wherein the new output RF plan indicates the optimized AP locations for the network deployment area.

11. The device of claim 9 , wherein the processing resource executes one or more of the instructions to:
extract the modified set of features from the input RF plan based on a segmentation of the input RF plan;
predict, using the first ML model, the candidate AP locatio

Assignees

• Hewlett Packard Entpr Dev Lp

Inventors

• Halli Siddharood
• Gupta Gopal
• Bhande Ajay Vishwanath
• Malyala Charan

Classifications

• H04W24/02 Primary

  Arrangements for optimising operational condition · CPC title

• G06N3/0475

  Generative networks · CPC title

• G06N3/088

  Non-supervised learning, e.g. competitive learning · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N20/10

  using kernel methods, e.g. support vector machines [SVM] · CPC title