Passive Traffic Shaping Detection via Delay Jitter Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for traffic shaping detection and bandwidth estimation in networks face challenges such as generating additional network traffic overhead, difficulty in separating delays caused by traffic shaping from other factors, and fluctuations in bandwidth estimation due to traffic bursts.
Innovation Solution
A method that uses passive traffic shaping detection and bandwidth estimation, selecting bandwidth samples with delay jitters greater than a jitter threshold, updating a probability distribution of sampled bandwidth, determining if traffic shaping is occurring based on the density of the probability distribution, and adjusting the upper limit of estimated bandwidth accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If active traffic shaping detection methods are used, then traffic shaping can be detected, but additional network traffic overhead is generated
Solution Approach 1:
The system uses existing network packets for bandwidth sampling and traffic shaping detection without generating additional probe packets. The detection mechanism leverages naturally occurring traffic to identify traffic shaping patterns, eliminating the need for extra detection traffic while maintaining detection accuracy
Solution Approach 2:
The patent extracts traffic shaping detection capabilities from active probing methods and implements passive detection using existing network traffic. By taking out the detection function from the data plane and implementing it through analysis of normal traffic patterns, no additional overhead is generated
2Measurement precision
If all sampled packets are used for bandwidth estimation, then more data is available for estimation, but delays caused by traffic shaping cannot be separated from other delay factors
Solution Approach 1:
The system applies different processing rules to different packets based on their delay characteristics. Packets with delay jitter above a threshold are identified as potentially shaped and processed differently from normal packets. This local differentiation allows accurate bandwidth estimation while isolating traffic shaping delays from other delay factors
Solution Approach 2:
The patent changes the parameter selection criterion from using all packets to using only packets with delay jitter below a threshold. This parameter change filters out packets affected by traffic shaping, allowing bandwidth estimation to focus on unshaped traffic and accurately separate shaping delays from other delay factors
3Quantity of substance
If bandwidth estimation includes traffic bursts, then more bandwidth samples are available, but bandwidth estimation fluctuates
Solution Approach 1:
The system uses a partial action approach by selecting only a subset of packets (those with delay jitter below threshold) for bandwidth sampling. This partial selection excludes traffic burst packets from the sampling set, providing sufficient bandwidth samples while eliminating the instability caused by including excessive action data from bursts
Data Source
AI summary
Apparatus, method and non-transitory computer readable medium for bandwidth estimation for a network using traffic shaping detection are provided, including selecting periodically sampled packets from the network, bandwidth samples each having a delay jitter greater than a jitter threshold, the delay jitter of a packet determined as the difference between a delay for the packet in the network and a minimum delay for all packets sampled within a delay statistics window for the packet; updating a probability distribution of sampled bandwidth based on the selected bandwidth samples; determining whether traffic shaping is occurring in the network based on a density of the updated probability distribution of sampled bandwidth; and based on a determination that traffic shaping is occurring in the network, adjusting an upper limit of estimated bandwidth for the network based on an estimated traffic shaping rate, the estimated traffic shaping rate determined based on the updated probability distribution.