Predictive resource allocation in an edge computing network utilizing machine learning

Patent metadata

Field	Value
Publication number	US-2025181416-A1
Application number	US-202519046919-A
Country	US
Kind code	A1
Filing date	Feb 6, 2025
Priority date	Nov 6, 2019
Publication date	Jun 5, 2025
Grant date	—

Abstract

The present technology relates to improving computing services in a distributed network of remote computing resources, such as edge nodes in an edge compute network. In an aspect, the technology relates to a method that includes aggregating historical request data for a plurality of requests, wherein the aggregated historical request data a time of the request, a location of a device from which the request originated, and/or a type of service being requested. The method also incudes training a machine learning model based on the aggregated historical request data; generating, from the trained machine learning model, a prediction for a type of service to be request; identifying an edge node, from a plurality of edge nodes, based on a physical location of the edge node; and based on predicted service, allocating computing resources for the computing service on the identified edge node.

First claim

What is claimed is:

1 . A computer-implemented method for reducing latency in providing a computing service, the method comprising:
aggregating historical request data for a plurality of requests; training a machine learning model based on the aggregated historical request data; generating, from the trained machine learning model, a prediction for a type of computing service to be requested at a predicted time and a predicted location; based on the generated predicted location, identifying an edge node, from a plurality of edge nodes, based on a physical location of the edge node; and based on the generated predicted type of computing service and the predicted time, allocating computing resources for the computing service on the identified edge node.

2 . The computer-implemented method of claim 1 , wherein the machine learning model is at least one of a decision tree, a random forest, a neural network, a continual learning model, or a deep learning model.

3 . The computer-implemented method of claim 1 , wherein allocating the computing resources comprises performing at least one of:
deploying a virtualized software; deploying a virtualized instance; deploying a virtualized machine; deploying virtualized infrastructure; deploying a virtualized container; loading a database into memory of the identified edge node; caching content for the computing service; or allocating storage resources in memory of the identified edge node.

4 . The computer-implemented method of claim 1 , wherein identified edge node includes at least one of a server, a graphics processing unit (GPU), a central processing unit (CPU), or a field-programmable gate array (FPGA).

5 . The computer-implemented method of claim 1 , further comprising:
receiving, from a computing device, a request for the predicted service type at the predicted time; and performing, by the identified edge node, the requested service with the allocated computing resources.

6 . The computer-implemented method of claim 5 , wherein the computing device is one of a smart phone, laptop, vehicle, drone, a mobile computer, or a plane.

7 . A system for reducing latency in providing a computing service, the system comprising:
a plurality of edge nodes having different physical locations; at least one processor; and memory, operatively connected to the at least one processor and storing instructions that, when executed by the at least one processor, cause the at least one processor to perform a set of operations comprising:
training a machine learning model based on historical request data for a plurality of requests;
generating, from the trained machine learning model, a prediction for a type of computing service to be requested at a predicted time and a predicted location;
based on the generated predicted location, identifying an edge node, from a plurality of edge nodes, based on a physical location of the edge node; and
based on the generated predicted type of computing service and the predicted time, allocating computing resources for the computing service on the identified edge node prior to the predicted time.

8 . The system of claim 7 , wherein the historical request data includes at least the following data for a plurality of requests: a time of the request, a location of the device from where the request originated, and a type of service being requested.

9 . The system of claim 7 , wherein the machine learning model is at least one of a decision tree, a random forest, a neural network, a continual learning model, or a deep learning model.

10 . The system of claim 7 , wherein allocating the computing resources comprises performing at least one of:
deploying a virtualized software; deploying a virtualized instance; deploying a virtualized machine; deploying virtualized infrastructure; deploying a virtualized container; loading a database into memory of the identified edge node; caching content for the computing service; or allocating storage resources in memory of the identified edge node.

11 . The system of claim 7 , wherein identified edge node includes at least one of a server, a graphics processing unit (GPU), a central processing unit (CPU), or a field-programmable gate array (FPGA).

12 . The system of claim 7 , wherein the operation further comprise:
receiving, from a mobile computing device, a request for the predicted service type at the predicted time; and performing, by the identified edge node, the requested service with the allocated computing resources.

13 . The system of claim 12 wherein the mobile computing device is one of a smart phone, laptop, vehicle, drone, a mobile computer, or a plane.

Assignees

• Centurylink Ip Llc

Inventors

• Casey Steven M
• Castro Felipe
• Opferman Stephen
• Mcbride Kevin M

Classifications

• G06N3/092

  Reinforcement learning · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0895

  Weakly supervised learning, e.g. semi-supervised or self-supervised learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• H04L47/83

  based on usage prediction · CPC title