Filter-Based Bandwidth Estimation Using Active Probing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for estimating available bandwidth in data networks either fail to provide real-time, accurate estimates or require significant data processing and memory resources, and often need access to all network nodes, which is typically not feasible.
Innovation Solution
A filter-based approach using active probing and Kalman filtering to estimate network conditions, specifically available bandwidth, by sampling probe packets and updating estimates in real-time with modest resource requirements, without needing access to intermediate nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional active probing methods are used to estimate available bandwidth, then real-time estimation capability is improved, but data processing resources and memory resources increase significantly
Solution Approach 1:
The patent extracts only the essential information needed for bandwidth estimation by using packet pair timing measurements rather than collecting and processing complete packet capture data. This extraction approach obtains the necessary timing information (inter-arrival times) without the overhead of processing entire packet streams, thereby reducing data processing and memory requirements while maintaining real-time estimation capability
Solution Approach 2:
The patent uses partial action by injecting only packet pairs (or packet trains) at specific rates rather than continuous heavy probing traffic. This partial probing approach provides sufficient information for bandwidth estimation without the excessive resource consumption of conventional methods that use more aggressive probing strategies
2Measurement precision
If conventional active probing methods are used to estimate available bandwidth, then measurement accuracy is improved, but device complexity and operational difficulty increase
Solution Approach 1:
The patent uses an intermediary approach by having the receiver node perform the complex timing analysis and bandwidth estimation calculations, while the sender node only needs to inject packet pairs with timing information. This distributes the computational complexity across network nodes rather than requiring complex equipment at all nodes or centralized access to intermediate nodes, maintaining accuracy while reducing operational difficulty
Solution Approach 2:
The patent enables self-service by allowing the receiver node to autonomously perform the bandwidth estimation using the timing information from packet pairs. The system uses available local resources at the receiver end to conduct the analysis without requiring external access to intermediate network nodes or complex coordinated control across multiple nodes
Data Source
AI summary
If a condition in a data transfer path is modeled appropriately, then a filter-based approach can be used to provide an estimate of the condition. This permits accurate, real-time estimates of the condition with modest requirements for data processing and memory resources.