Granted Patent B2

US 12,699,641 · App. 17/502,601 · Granted Aug 4, 2026

Inventors: John Coster (Sammamish, WA); Anand Kumar Subramaniam (Sammamish, WA); Sean Seemann (Seattle, WA); Shalini Selvaraj (Snoqualmie, WA); Burhan Nurdin (Bothell, WA); Mohammad Bari (Woodinville, WA); Karthik Kandukoori (Georgetown, TX)

Assignee: T-Mobile USA, Inc.

G06F11/3457G06F30/20G06N20/00G06F8/65

Abstract

Claims (84)

1 . A non-transitory, computer-readable storage medium carrying instructions, which, when executed by at least one data processor of a system, cause the system to:

extract datasets including sensor data from respective data streams of multiple systems coupled to a telecommunications network,

wherein the sensor data is extracted in parallel as data batches from the multiple data streams,

wherein the sensor data includes temperature and power consumption readings of sensors disposed on components of the multiple systems,

wherein the sensor data is periodically received from the sensors disposed on the components of the multiple systems;

standardize the sensor data into a common format and into a classification for a performance metric,

wherein the performance metric is related to a capability of the multiple systems based on the temperature and power consumption readings of the sensors,

wherein the sensor data is aggregated across a temporal dimension to classify the sensor data to one or more timeframes when the sensor data was generated or captured,

wherein the sensor data is further aggregated across a spatial dimension to classify the sensor data based on whether the sensors are disposed on circuitry, devices, or structures of the multiple systems, and

wherein the standardized sensor data includes the aggregate sensor data and the classification is based on the temporal dimension and the spatial dimension;

adjust a model of a target system of the multiple systems based on the standardized sensor data and in accordance with the classification for the performance metric;

simulate, based on the model, a change to the target system and a resulting impact on the performance metric for the telecommunications network,

wherein the change to the target system includes an update to software in the target system or a hardware expansion or contraction in the target system;

determine, based on simulating the change to the target system and the resulting impact on the performance metric, a procedure for mitigating the resulting impact on the performance metric for the telecommunications network;

generate a software network interface including a visualization of the simulation including the resulting impact on the performance metric for the telecommunications network,

wherein the software network interface enables an end user to interact with the visualization, and

wherein the software network interface includes an indication of a procedure to mitigate the resulting impact on the performance metric; and

perform the procedure for mitigating the resulting impact on the performance metric for the telecommunications network when updating the software in the target system or performing the hardware expansion or contraction in the target system.

2 . The non-transitory computer-readable storage medium of claim 1 , wherein the telecommunications network includes 5G network components, and wherein to simulate the change and the resulting impact comprises causing the system to:

predict a first impact on the telecommunications network in response to a simulated expansion in a cellular-network infrastructure.

3 . The non-transitory computer-readable storage medium of claim 1 , wherein to simulate the change and the resulting impact comprises causing the system to:

predict a second impact on network access to the telecommunications network in response to a simulated deployment of a mobile application on multiple mobile devices on the telecommunications network.

4 . The non-transitory computer-readable storage medium of claim 1 :

wherein the target system corresponds to a datacenter of the telecommunications network, and

wherein the software network interface includes runnable code or presents text related to the simulation of the resulting impact on the performance metric for the telecommunications network.

5 . The non-transitory computer-readable storage medium of claim 1 , wherein the system is further caused to:

predict, based on the model, a third impact on a cost metric in response to the simulated change to the target system,

wherein the software network interface includes the third impact on the cost metric and a procedure for reducing the cost metric.

6 . The non-transitory computer-readable storage medium of claim 1 , wherein to generate the software network interface comprises causing the system to:

generate an interactive decision-making tool that enables real-time analytics with respect to a cost metric and a risk to the telecommunications network based on the simulated change.

7 . The non-transitory computer-readable storage medium of claim 1 :

wherein the multiple systems include multiple storage systems of the telecommunications network,

wherein the target system includes a particular storage system of the multiple storage systems, and

wherein the performance metric includes a power consumption of the storage system.

8 . The non-transitory computer-readable storage medium of claim 1 , wherein to standardize the sensor data comprises causing the system to:

aggregate datasets of the sensor data across the multiple systems, which are located at different geographic regions, and

wherein the standardized sensor data includes the aggregate sensor data and the classification is based on the different geographic regions.

9 . The non-transitory computer-readable storage medium of claim 1 , wherein to standardize the sensor data comprises causing the system to:

aggregate datasets of the sensor data across layers of abstraction for the multiple systems,

wherein the layers include a physical hardware, system software on the physical hardware, an application on the system software, and a service of the application, and

wherein the standardized sensor data includes the aggregate sensor data and the classification is based on levels of abstraction.

10 . The non-transitory computer-readable storage medium of claim 1 , wherein to simulate the resulting impact on the performance metric comprises causing the system to:

constrain the simulation to a threshold value for the performance metric.

11 . A system comprising:

at least one hardware processor; and

at least one non-transitory memory storing instructions, which, when executed by the at least one hardware processor, cause the system to:

extract sensor data from multiple systems of a network,

wherein the sensor data includes measurements taken by sensors disposed at the multiple systems,

wherein the sensor data includes temperature and power consumption readings of sensors disposed on components of the multiple systems, and

wherein the sensor data is periodically received from the sensors disposed on the components of the multiple systems;

standardize the sensor data into a common format and into a classification for a performance metric,

wherein the performance metric includes a capability of the multiple systems based on the temperature and power consumption readings of the sensors,

wherein the sensor data is aggregated across a temporal dimension to classify the sensor data to one or more timeframes when the sensor data was generated or captured,

wherein the sensor data is further aggregated across a spatial dimension to classify the sensor data based on whether the sensors are disposed on circuitry, devices, or structures of the multiple systems, and

wherein the standardized sensor data includes the aggregate sensor data and the classification is based on the temporal dimension and the spatial dimension;

adjust a machine learning (ML) model for the performance metric based on the standardized sensor data and in accordance with the classification;

simulate, based on the ML model, an impact on the performance metric in response to a simulated change of a target system;

determine, based on simulating the change to the target system and the resulting impact on the performance metric, a procedure for mitigating the impact on the performance metric when performing the change to the target system;

perform the procedure for mitigating the impact on the performance metric when performing the change to the target system; and

administer a network portal that enables interactive analytics for engaging with the simulation and the impact on the performance metric.

12 . The system of claim 11 further caused to:

predict an impact on the network in response to a simulated change including a simulated update to a software application or an expansion of the network.

13 . The system of claim 11 further caused to, prior to the impact on the performance metric being simulated:

classify the sensor data based a hardware or software components of the multiple networks,

wherein the ML model is optimized in accordance with the classification of the hardware or software components.

14 . A method for performing a change to a storage system and mitigating an impact on a performance metric for the storage system, comprising:

receiving sensor data generated by sensors disposed on components of the storage system,

wherein the sensor data is indicative of a temperature and power consumption of the components of the storage system,

wherein the sensor data is periodically received from the sensors disposed on the components of the storage system;

classifying the sensor data based on a performance metric for the storage system,

wherein the performance metric is related to a capability of the storage system based on the temperature and power consumption readings of the sensors,

wherein the sensor data is aggregated across a temporal dimension to classify the sensor data to one or more timeframes when the sensor data was generated or captured,

wherein the sensor data is further aggregated across a spatial dimension to classify the sensor data based on whether the sensors are disposed on circuitry, devices, or structures of the multiple systems, and

wherein the standardized sensor data includes the aggregate sensor data and the classification is based on the temporal dimension and the spatial dimension;

simulating, based on the classified sensor data, the impact on the performance metric in response to a simulated change to the storage system;

determining, based on said simulating, a procedure for mitigating the impact on the performance metric when performing the change to the storage system;

performing the procedure for mitigating the impact on the performance metric when performing the change to the storage system; and

generating a network portal that enables end users to perform analytics to mitigate the impact on the performance metric for the storage system.

15 . The method of claim 14 further comprising, prior to simulating the impact on the performance metric:

training a machine learning (ML) model to optimize for the performance metric based on the classified sensor data,

wherein the simulation of the impact on the performance metric is based on the ML model.

16 . The method of claim 14 , wherein the change to the storage system includes an update to a software running on the storage system or a hardware expansion of the storage system.

17 . The method of claim 14 further comprising:

predicting, based on the classified sensor data, a temperature of the storage system or surrounding the storage system and power consumption by the storage system.

Assignments (1)

Continuity (1)

References Cited (150)