Monitoring a communication network

Patent metadata

Field	Value
Publication number	US-2022014471-A1
Application number	US-202117485946-A
Country	US
Kind code	A1
Filing date	Sep 27, 2021
Priority date	Oct 26, 2018
Publication date	Jan 13, 2022
Grant date	—

Abstract

Methods and systems for monitoring a communication network using machine-learning techniques are disclosed. In some implementations, a forecasted amount of traffic for a communication network is determined using one or more network traffic forecasting models being configured to generate the forecasted amount of traffic based on data indicating one or more previous amounts of traffic for the communication network. A measure of network health is generated based on a measured amount of traffic and the forecasted amount of traffic. Data indicating one or more characteristics of the communication network is processed using one or more machine learning models to generate a predicted measure of network health for a future time period. An indication of the predicted measure of network health for the future time period is provided.

First claim

What is claimed is:

1 . A method performed by one or more computers, the method comprising:
determining, by the one or more computers, a predicted traffic amount for a communication network for each time period in a series of multiple time periods, the predicted traffic amount being determined based on a time series of historical measured traffic for the communication network for multiple time periods before the particular time period; measuring, by the one or more computers, a traffic amount for the communication network for each time period in the series of multiple time periods; generating, by the one or more computers, a network health score for the communication network for each time period in the series of multiple time periods, each network health score being generated based on (i) the predicted traffic amount for the corresponding time period and (ii) the measured traffic amount for the corresponding time period; generating, by the one or more computers, a feature vector of status information for the communication network for each time period in the series of multiple time periods, wherein each of the feature vectors is indicative of a status of network elements of the communication network during the corresponding time period; and training, by the one or more computers, a machine learning model based on the feature vectors and network health scores for the multiple time periods, the trained machine learning model being configured to (i) receive, as input to the trained machine learning model, a feature vector of status information for a first time period and (ii) output, in response to the received feature vector, a network health score indicating a level of health of the communication network predicted to occur at a time period after the first time period corresponding to the received feature vector.

2 . The method of claim 1 , wherein training the machine learning model comprises using the network health scores for the multiple time periods as training targets for the output of the machine learning model, wherein each network health score serves as a training target for output provided in response to input of a feature vector indicating status at a time period prior to the time period the network health score describes.

3 . The method of claim 2 , wherein training the machine learning model involves using a predetermined time offset between the input to the machine learning model and the training target for output of the machine learning model, such that each network health score serves as a training target corresponding to an input feature vector for a time period that precedes the time period of the network health score by the predetermined time offset.

4 . The method of claim 1 , comprising determining a current feature vector indicative of current status of network elements of the communication network; and
providing the current feature vector as in put to the trained machine learning model to obtain a predicted network health score indicating predicted level of health of the communication network at a future time period.

5 . The method of claim 1 , wherein generating the network health score for the communication network for each time period in the series of multiple time periods comprises generating each network health score as a ratio of (i) the predicted traffic amount for the corresponding time period and (ii) the measured traffic amount for the corresponding time period.

6 . The method of claim 1 , wherein the network health scores are generated on a periodic basis as each new time period in the series of multiple time periods elapses.

7 . The method of claim 1 , further comprising, after training the machine learning model, using the machine learning model to generate a predicted network health score for the communication network; and
altering a configuration one or more components of the communication network based on the predicted network health score.

8 . The method of claim 1 , wherein determining the predicted traffic amount comprises determining the predicted traffic amount based on output of one or more traffic forecasting machine learning models trained, based on time series data for historical traffic of the communication network, to output data indicating a predicted traffic amount for a time indicated through input to the one or more traffic forecasting machine learning models.

9 . The method of claim 5 , wherein determining the predicted traffic amount comprises determining, as the predicted traffic amount, a weighted average of predicted traffic amounts from multiple machine learning models that each have been trained based on the time series data for historical traffic of the communication network, wherein the multiple machine learning models include models of at least two different types from among the set of model types consisting of a linear regression model, a neural network model, and a random forest regressor.

10 . The method of claim 1 , wherein the communication network is a satellite communication network.

11 . One or more non-transitory computer-readable media storing instructions that are operable, when executed by one or more computers, to cause the one or more computers to perform operations comprising:
determining, by the one or more computers, a predicted traffic amount for a communication network for each time period in a series of multiple time periods, the predicted traffic amount being determined based on a time series of historical measured traffic for the communication network for multiple time periods before the particular time period; measuring, by the one or more computers, a traffic amount for the communication network for each time period in the series of multiple time periods; generating, by the one or more computers, a network health score for the communication network for each time period in the series of multiple time periods, each network health score being generated based on (i) the predicted traffic amount for the corresponding time period and (ii) the measured traffic amount for the corresponding time period; generating, by the one or more computers, a feature vector of status information for the communication network for each time period in the series of multiple time periods, wherein each of the feature vectors is indicative of a status of network elements of the communication network during the corresponding time period; and training, by the one or more computers, a machine learning model based on the feature vectors and network health scores for the multiple time periods, the trained machine learning model being configured to (i) receive, as input to the trained machine learning model, a feature vector of status information for a first time period and (ii) output, in response to the received feature vector, a network health score indicating a level of health of the communication network predicted to occur at a time period after the first time period corresponding to the received feature vector.

12 . The one or more non-transitory computer-readable media of claim 11 , wherein training the machine learning model comprises using the network health scores for the multiple time periods as training targets for the output of the machine learning model, wherein each network health score serves as a training target for output provided in response to input of a feature vector indicating status at a time period prior to the time period the network health score describes.

13 . The one or more non-transitory computer-readable media of claim 12 , wherein training the machine learning model involves using a predetermined time offset between the input to the machine learning mod

Assignees

• Hughes Network Systems Llc

Inventors

• Arora Amit
• Gharpuray Archana
• Kenyon John

Classifications

• H04L41/147 Primary

  for predicting network behaviour · CPC title

• H04L47/127 Primary

  by using congestion prediction · CPC title

• H04L43/16

  Threshold monitoring · CPC title

• H04L43/045

  for graphical visualisation of monitoring data · CPC title

• H04L41/145

  involving simulating, designing, planning or modelling of a network · CPC title