Application-Aware Network Management via Flow Feature Extraction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Network management systems that rely on static properties of network connections fail to accurately identify applications due to dynamic negotiation of per-session properties and shared properties among applications, and are vulnerable to misinformation from applications, with deep packet inspection being costly and unreliable, especially with encrypted traffic.

Innovation Solution

Implementing an application-aware network management system that uses application classifiers generated by network devices within the control plane, derived from application identifiers, to identify applications and manage network flows, decoupling control and data planes for scalable and consistent application identification across software-defined networks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If deep packet inspection is used to identify applications, then application identification accuracy is improved, but computational cost and processing time increase significantly

Engineering Contradiction:
Improveapplication identification accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent extracts only the essential features needed for application identification from network traffic flows, rather than performing complete deep packet inspection. By selecting specific flow characteristics (packet size, inter-arrival time, protocol type) and excluding unnecessary data, the system achieves accurate application identification while significantly reducing computational overhead and processing time.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system performs preliminary classification of network flows using lightweight feature extraction before applying more complex identification algorithms only to ambiguous cases. This two-stage approach pre-processes the data by filtering out easily identifiable applications, thereby reducing the computational burden on the main identification system and decreasing overall processing time.

Inventive Principle:
Principle #10Preliminary action

2Ease of operation

If static properties of network connections are used for application identification, then processing simplicity is improved, but identification accuracy deteriorates due to dynamic negotiation and shared properties

Engineering Contradiction:
Improveprocessing simplicityVSAvoidapplication identification accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent transitions from static property-based identification to dynamic behavior-based identification. Instead of relying on fixed port numbers or IP addresses that applications may negotiate or share, the system monitors dynamic characteristics of network flows such as packet size variations, inter-arrival time patterns, and sequence behavior, which change over time and are unique to each application's operational pattern.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes the parameters used for identification from static connection properties (port numbers, protocols) to dynamic flow characteristics (packet size distribution, timing patterns, byte rates). This parameter transformation enables the system to distinguish applications even when they use the same static properties, thereby improving identification accuracy while maintaining processing efficiency through focused feature monitoring.

Inventive Principle:
Principle #35Parameter changes

3Speed

If QoS information provided by applications is trusted, then network provisioning speed is improved, but reliability deteriorates due to vulnerability to misinformation

Engineering Contradiction:
Improvenetwork provisioning speedVSAvoidapplication identification reliability
Core Design Contradiction:
SpeedVSReliability

Solution Approach 1:

The system implements a feedback mechanism where QoS information provided by applications is cross-validated against observed network flow characteristics. Instead of blindly trusting application-provided QoS labels, the system continuously monitors actual traffic patterns and compares them with expected patterns for the claimed application type, providing feedback to verify or correct the identification, thereby maintaining fast provisioning while improving reliability.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent introduces an intermediary verification layer between the application and the network provisioning system. This intermediary component analyzes network flow characteristics and acts as a mediator that validates application-provided QoS information before final provisioning decisions are made, preventing misinformation from affecting network resource allocation while maintaining efficient provisioning speeds.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3033687B1Application-aware network management
Publication Date: 2019.07.03 HEWLETT PACKARD ENTERPRISE DEV LP
  • EP3033687B1 patent drawingFigure 1
  • EP3033687B1 patent drawingFigure 2
  • EP3033687B1 patent drawingFigure 3

AI summary

In one implementation, an application-aware network control system determines that a network flow has been instantiated, derives network flow features associated with the network flow in response to determining the network flow has been instantiated, and provides the network flow features to an application identification module. In response to an application identification signal, the application-aware network control system ceases deriving the network flow features, and applies a flow action to the network flow based on an application identification report received from the application identification module.