Network Service Configuration via Hidden Markov Models

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Network administrators face challenges in creating effective network service configurations due to broad guidelines and manual processes, leading to user errors and resource wastage.

Innovation Solution

A service management platform utilizing machine learning, specifically Hidden Markov Models, to analyze telemetry data and recommend optimized network service configurations, automating the process and reducing human subjectivity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual processes are used to create network service configurations based on broad guidelines, then network administrators have flexibility in decision-making, but user errors increase and productivity decreases

Engineering Contradiction:
Improveconfiguration accuracyVSAvoidconfiguration creation speed
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system enables self-service by automatically generating network service configurations through machine learning models that analyze telemetry data and service requirements, eliminating the need for manual configuration creation while ensuring accuracy through automated validation processes

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual process of configuration creation with an automated machine learning system that uses Hidden Markov Models and other ML techniques to generate configurations, thereby improving both reliability and productivity simultaneously

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If manual configuration creation processes are used, then network administrators can apply domain knowledge, but resource wastage increases due to errors and rework

Engineering Contradiction:
Improveconfiguration correctnessVSAvoidcomputational resource wastage
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The system performs preliminary action by pre-generating optimized configurations through machine learning models before deployment, validating configurations in advance, and preventing errors before they consume computational resources in production environments

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback mechanisms where telemetry data from live networks continuously trains and improves the machine learning models, creating a closed-loop system that learns from past configurations and outcomes to progressively reduce resource wastage while maintaining correctness

Inventive Principle:
Principle #23Feedback

3Productivity

If automated machine learning systems are used to generate network service configurations, then productivity increases and resource efficiency improves, but system complexity increases

Engineering Contradiction:
Improveconfiguration deployment speedVSAvoidsystem architecture complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system achieves universality by creating a multi-functional platform that handles configuration generation, validation, optimization, and deployment through integrated machine learning models, reducing the need for separate specialized systems while maintaining high productivity

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent uses intermediary components such as feature extraction layers, model training pipelines, and configuration validation intermediaries that bridge the gap between complex machine learning algorithms and simple configuration outputs, managing system complexity while preserving productivity benefits

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3668007B1System for identifying and assisting in the creation and implementation of a network service configuration using hidden markov models (HMMS)
Publication Date: 2022.05.04 JUNIPER NETWORKS INC
  • EP3668007B1 patent drawingFigure 1A
  • EP3668007B1 patent drawingFigure 1B
  • EP3668007B1 patent drawingFigure 1C

AI summary

A device may receive a request for a network service configuration (NSC) that is to be used to configure network devices. The device may select a graphical data model that has been trained via machine learning to analyze a dataset that includes information relating to a set of network configuration services, where aspects of a subset of the set of network configuration services have been created over time. The device may determine, by using the graphical data model, a path through a set of states of the graphical data model, where the path corresponds to a particular NSC. The device may select the particular NSC based on the path determined. The device may perform a first group of actions to provide data identifying the particular NSC for display, and/or a second group of actions to implement the particular NSC on the network devices.