Embedding neural networks as a matrix for network device in wireless network

Patent metadata

Field	Value
Publication number	US-12563418-B2
Application number	US-202318348039-A
Country	US
Kind code	B2
Filing date	Jul 6, 2023
Priority date	Jul 1, 2022
Publication date	Feb 24, 2026
Grant date	Feb 24, 2026

Abstract

Embodiments herein disclose a method and a device for embedding neural networks as a matrix for a network device in wireless networks. The method includes receiving s from the network device. Further, the method also includes determining the KPI among the plurality of KPIs as target KPIs that related to a network anomaly using a ML model. Further, the method also includes determining a correlation of the target KPI with the plurality of KPIs for the network anomaly using the ML model. Further, the method also includes determining the matrix indicating a relation of the target KPI with the plurality of KPIs. Furthermore, the method includes optimizing a resource of the network device by embedding the matrix in the network device.

First claim

What is claimed is:

1 . A method, comprising:
receiving, by a network server, a plurality of Key Performance Indicators (KPIs) from at least one network device in a wireless network; determining, by the network server, from among the plurality of KPIs a target KPIs related to at least one network anomaly using a Machine Learning (ML) model; determining, by the network server, a correlation of the target KPI with the plurality of KPIs for the at least one network anomaly using the ML model; determining, by the network server, at least one matrix as a compressed representation of a neural network trained by the network server and that indicates a relationship of the target KPI with the plurality of KPIs; and providing, by the network server to the at least one network device, the at least one matrix to be embedded in the at least one network device to permit the at least one network device to use the embedded matrix to optimize at least one resource of the at least one network device.

2 . The method as claimed in claim 1 , wherein determining, by the network server, the at least one matrix indicating a relation of the target KPI with the plurality of KPIs, includes:
correlating, by the network server, the plurality of KPIs with the target KPI based on the at least one network anomaly; transmitting, by the network server, the plurality of KPIs to an Artificial Intelligence (AI) server for training; and receiving by the network server, the plurality of KPIs trained from the AI server.

3 . The method as claimed in claim 1 , wherein the at least one matrix comprises at least one of a KPI matrix and a coefficient matrix.

4 . The method as claimed in claim 1 , wherein the at least one resource includes plural resources, and wherein the plural resources comprises one or more of memory requirements of the at least one network device, computation requirements of the at least one network device, flops of the at least one network device, RAM requirements, GPU requirements, CPU cycles, or prediction times of the ML model of the at least one network device.

5 . The method as claimed in claim 4 , wherein the prediction times of the ML model of the at least one network device includes one or more of: a congestion prediction time, a handover prediction time, a MAC scheduling time, or a call mute reduction time.

6 . The method as claimed in claim 1 , wherein optimizing, by the network server, the at least one resource of the at least one network device based on the at least one matrix, comprises:
sending, by the network server, at least one coefficient matrix to the at least one network device; and embedding, by the network server, the at least one coefficient matrix in the at least one network device to optimize the at least one resource of the at least one network device.

7 . The method as claimed in claim 1 , wherein receiving, by the network server, the plurality of KPIs from at least one network device, comprises: receiving a dataset comprises plurality of KPI from the at least one network device; wherein the plurality of KPIs are related to time-series data of the at least one network device.

8 . A method by a network device in a wireless network, comprising:
embedding, by the network device, at least one matrix as a representation of a neural network to optimize at least one resource of the network device, wherein the at least one matrix includes a relationship between a target KPI with a plurality of KPIs of at least one network device; and predicting, by the network device, at least one future network anomaly based on the at least one matrix embedded in the network device.

9 . A network server, comprising:
a memory; at least one processor coupled to the memory; and an optimal resource controller coupled to the memory and the processor, and configured to:
receive a plurality of Key Performance Indicators (KPIs) from at least one network device in a wireless network;
determine from among the plurality of KPIs a target KPIs related to at least one network anomaly using a Machine Learning (ML) model;
determine a correlation of the target KPI with the plurality of KPIs for at least one network anomaly using the ML model;
determine at least one matrix as a compressed representation of a neural network trained by the network server and that indicates a relationship of the target KPI with the plurality of KPIs; and
providing, by the network server to the at least one network device, the at least one matrix to be embedded in the at least one network device to permit the at least one network device to use the embedded matrix to optimize at least one resource of the at least one network device.

10 . The network server as claimed in claim 9 , wherein determining the at least one matrix indicating a relation of the target KPI with the plurality of KPIs comprises:
correlating the plurality of KPIs with the target KPI based on the at least one network anomaly; transmitting the plurality of KPIs to an Artificial Intelligence (AI) server for training; and receiving the plurality of KPIs that is trained from the AI server.

11 . The network server as claimed in claim 9 , wherein the at least one matrix comprises at least one of a KPI matrix and a coefficient matrix.

12 . The network server as claimed in claim 9 , wherein the at least one resource includes plural resources, and wherein the plural resources comprise one or more of memory requirements of the at least one network device, computation requirements of the at least one network device, flops of the at least one network device, RAM requirements, GPU requirements, CPU cycles, or prediction times of the ML model of the at least one network device.

13 . The network server as claimed in claim 12 , wherein the prediction times of the ML model of the at least one network device includes one or more of: a congestion prediction time, a handover prediction time, a MAC scheduling time, or a call mute reduction time.

14 . The network server as claimed in claim 9 , wherein optimizing the at least one resource of the at least one network device based on the at least one matrix, comprises:
sending at least one coefficient matrix to the at least one network device; and embedding the at least one coefficient matrix to the at least one network device to optimize the at least one resource of the at least one network device.

15 . The network server as claimed in claim 9 , wherein receiving the plurality of KPI from at least one network device, comprises: receiving a dataset comprises plurality of KPI from the at least one network device; wherein the plurality of KPI are related to time-series data of the at least one network device.

Assignees

• Samsung Electronics Co Ltd

Inventors

• Murgai Vishal
• Kumar Swaraj
• Singh Sukhdeep

Classifications

• H04L41/0836

  to enhance reliability, e.g. reduce downtime · CPC title

• H04L41/16

  using machine learning or artificial intelligence · CPC title

• H04W24/02 Primary

  Arrangements for optimising operational condition · CPC title