Traffic Matrix Estimation Using Network Topology Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating traffic matrices in communication networks are topology-agnostic, leading to inaccuracies and high computational costs, with current approaches like NetFlow/IPFIX requiring extensive resources and being unsuitable for large networks, and existing estimation techniques relying on flawed statistical assumptions or gravity models that fail to accurately represent traffic demands.
Innovation Solution
Incorporating network graph statistics and edge betweenness centrality into traffic matrix estimation, using routing information and link measurements to determine an initial estimate, and iteratively refining it with additional network data, allowing for faster and more accurate traffic matrix inference that scales to larger networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If NetFlow/IPFIX measurement approaches are used, then traffic matrix measurement capability is improved, but processor demand on routers increases significantly
Solution Approach 1:
The patent introduces an external server as an intermediary that performs traffic matrix estimation. Instead of having routers directly perform complex NetFlow/IPFIX processing, the routers simply collect link measurements and routing information, then send this data to the external server which performs the estimation. This transfers the computational burden from resource-constrained routers to a more powerful external system.
Solution Approach 2:
The patent extracts the traffic matrix estimation function from the router system and places it on an external server. The core estimation algorithm, including gravity model calculations and optimization processes, is removed from the routers and executed externally, leaving routers with only simple data collection and transmission functions.
2Measurement precision
If NetFlow/IPFIX measurement approaches are used, then traffic matrix measurement capability is improved, but device complexity increases
Solution Approach 1:
The patent creates a universal estimation framework that works across heterogeneous router platforms. The external server implements a standardized estimation process that can handle data from different router types and networking equipment vendors, eliminating the need for platform-specific implementations and reducing deployment complexity.
Solution Approach 2:
The external server acts as a mediator that standardizes the interface between diverse router equipment and the traffic matrix estimation process. It handles data collection, processing, and estimation in a unified manner, abstracting away the heterogeneity of underlying networking equipment.
3Ease of operation
If topology-agnostic estimation methods are used, then ease of deployment is improved, but estimation accuracy deteriorates
Solution Approach 1:
The patent performs preliminary actions by collecting and analyzing network topology information and routing data before performing traffic matrix estimation. The gravity model uses pre-collected topology features (such as shortest paths, node degrees, and network structure) to guide the estimation process, improving accuracy without complicating deployment.
Solution Approach 2:
The patent changes the parameters used in estimation from generic topology-agnostic models to topology-aware parameters. It incorporates specific network structure parameters such as edge betweenness centrality, shortest path lengths, and routing information into the gravity model, transforming the estimation process to utilize available topological information for improved accuracy.
4Measurement precision
If gravity model with topology features is used, then estimation accuracy is improved, but computational cost increases
Solution Approach 1:
The patent applies partial action by selectively using topology features that provide the most benefit for accuracy. Instead of computing all possible topological metrics, it focuses on key features such as shortest paths, node degrees, and edge betweenness centrality that have the greatest impact on estimation accuracy, reducing unnecessary computational overhead.
Solution Approach 2:
The patent performs preliminary computation of topology features and stores them for use during estimation. Network topology information, routing data, and pre-computed metrics are calculated in advance and cached, so they do not need to be re-computed for each estimation iteration, significantly reducing computational cost during the actual estimation process.
Data Source
AI summary
Systems and methods include receiving network topology information of a network including a plurality of routers; receiving link measurements defining bandwidth on links in the network; determining routes in the network based on the network topology information; and utilizing the routes and the link measurements to determine an estimate of an initial traffic matrix that includes the bandwidth between origin routers and destination routers.