Granted Patent B2

US 12,609,869 · App. 18/857,581 · Granted Apr 21, 2026

Inventors: Carla Di Cairano-Gilfedder (London, GB); Anthony Conway (London, GB)

Assignee: BRITISH TELECOMMUNICATIONS PUBLIC LIMITED COMPANY

H04L41/145H04L41/0816H04L41/0823H04L41/16

Abstract

Claims (38)

1 . A digital twin system for controlling a computing system including multiple hardware and software entities, the digital twin system comprising:

a digital twin comprising a virtual model of the computing system including the multiple hardware and software entities, the virtual model incorporating three-dimensional physical dimensions of the hardware entities and their environment, the virtual model being configured to emulate the behaviour of the computing system;

a communications interface for performing bi-directional communications with the computing system, the communications including receipt of real-time information regarding performance of the computing system and sensed conditions of the computing system and its environment;

a machine learning system trained on (i) historical and real-time performance data; and (ii) historical and real-time sensed data of the computing system and its environment;

an orchestrator configured to use the digital twin and the trained machine learning system in conjunction with the received real-time performance data and sensed data for the computing system and its environment to determine at least one action for the computing system and its environment that supports one or more selected system management objectives, and to transmit a command over the bi-directional link to the computing system to perform the at least one action.

2 . The digital twin system of claim 1 , wherein the one or more system management objectives are selected from:

maximisation of quality of service

maximisation of resources utilisation

minimisation of failure risk

minimisation of operational cost

or a combination of them using a weighted function.

3 . The digital twin system of claim 1 , wherein the machine learning system is configured to receive adaptation training during production use of the digital twin system for controlling the computing system.

4 . The digital twin system of claim 1 , wherein the virtual model determines a granularity of hardware and associated software entities for incorporating into the virtual model.

5 . The digital twin system of claim 4 , wherein the granularity is determined subject to a machine learning process, and optionally wherein the granularity is dynamically updated during production use of the digital twin system for controlling the computing system.

6 . The digital twin system of claim 1 , wherein the virtual model determines a clustering of hardware entities for incorporating into the virtual model.

7 . The digital twin system of claim 6 , wherein the clustering is determined subject to a machine learning process, and optionally wherein the clustering is dynamically updated during production use of the digital twin system for controlling the computing system.

8 . The digital twin system of claim 1 , wherein the historical and real-time performance data comprises at least one of utilisation, loss, delay and service failure.

9 . The digital twin system of claim 1 , wherein the historical and real-time sensed environmental conditions of the computing system comprise at least temperature of one or more of the hardware entities.

10 . The digital twin system of claim 1 , wherein the ML system is further trained on static information relating to the computing system and/or the environment.

11 . The digital twin system of claim 1 , wherein the orchestrator is configured to simulate potential interventions on the digital twin to assess such interventions in relation to the system management objectives.

12 . The digital twin system of claim 1 , wherein the orchestrator is configured to redeploy services from a first hardware or software entity in a cluster to a second hardware or software entity in the cluster, the orchestrator being further configured to use a spatial three-dimensional view of the cluster, maintained by the digital twin, to assist in the choice of the second hardware or software entity.

13 . The digital twin system of claim 1 , wherein the computing system is a software-defined network and/or wherein the computing system includes a virtualisation layer above the hardware entities, and wherein the orchestrator is configured to instantiate and deploy virtual resources using the virtualisation layer.

14 . A method for using a digital twin system to control a computing system including multiple hardware and software entities, the digital twin system including a digital twin comprising a virtual model of the computing system including the multiple hardware and software entities, the virtual model incorporating three-dimensional physical dimensions of the hardware entities and their environment, the virtual model being configured to emulate the behaviour of the computing system, wherein the method comprises:

performing bi-directional communications between the digital twin system and the computing system, the communications including receipt of real-time information regarding performance of the computing system and sensed conditions of the computing system and its environment;

determining, by an orchestrator, at least one action for the computing system that supports one or more selected system management objectives, wherein the at least one action is determined using the digital twin and a machine learning system, in conjunction with the received real-time performance data and sensed conditions of the computer system and its environment, wherein the machine learning system is trained on: (i) historical and real-time performance data; and (ii) historical and real-time sensed data for the computing system and its environment;

and transmitting a command over the bi-directional link to the computing system to perform the at least one action.

15 . A non-transitory computer-readable storage medium storing a computer program comprising software instructions that when executed on a processor of a computing device cause the computing device to perform the method of claim 14 .

16 . The method of claim 14 , further comprising receiving adaptation training during production use of the digital twin system for controlling the computing system.

17 . The method of claim 14 , further comprising:

determining a granularity of hardware and associated software entities for incorporating into the virtual model; and

wherein the granularity is determined subject to a machine learning process, and the granularity is dynamically updated during production use of the digital twin system for controlling the computing system.

18 . The method of claim 14 , further comprising:

determining a clustering of hardware entities for incorporating into the virtual model; and

wherein the clustering is determined subject to a machine learning process, and the clustering is dynamically updated during production use of the digital twin system for controlling the computing system.

19 . The method of claim 14 , wherein the historical and real-time sensed environmental conditions of the computing system comprise at least temperature of one or more of the hardware entities.

20 . The method of claim 14 , further comprising:

redeploying services from a first hardware or software entity in a cluster to a second hardware or software entity in the cluster, and

using a spatial three-dimensional view of the cluster, maintained by the digital twin, to assist in the choice of the second hardware or software entity.

Assignments (1)

Priority Claims (2)

Continuity (1)

References Cited (17)