CRII: NeTS: Denoising Internet Delay Measurements using Weak Supervision
Funding source: NSF CNS-1850297. Period of performance: 06/15/2019 -- 05/31/2022.Project Overview
Understanding the delay characterisitcs of the Internet is one of the key goals of Internet measurement researchers, service providers, and content delivery networks. To this end, a myriad of measurement tools and techniques were proposed by the researchers in academia and industry, and datasets from such measurement tools are curated to facilitate analyses at a later time.
Despite the benefits of the proposed tools to measure the delay characterisitics of the Internet and the availability of datasets from measurement efforts, what is critically lacking is a systematic framework to interpret the results from the tools and datasets. The key hinderance to creating this framework is measurement noise, which we define as the presence of non-representative and erroneous values in the delay measurements. Noise confounds all types of end-to-end delay measurements and can lead to performance issues and unnecessary operational decisions. State-of-the-art denoising techniques are (1) time consuming and labor intensive: they are done manually due to the lack of ground truth data required to classify and discern noise from the actual delay behaviors of the network, and (2) ineffective: they are too naïve with simple heuristics and filters, which are impractical and which can lead to unnecessary operational decisions.
To tackle these challenges, this research will develop a systematic weak supervision-based framework and culminate in NoMoNoise for denoising Internet delay measurements in an automated and rapid fashion.
People
- PI: Ram Durairajan
- B.S. Students: Jared Knofcynzski, Juno Mayer, Nolan Rudolph, Nick Hendersen
- Ph.D. Students: Chris Misa, Yukhe Lavinia
- Collaborators: Daniel Lowd, Reza Rejaie, Sudarsun Kannan, Walter Willinger
Publications
- ARISE: A Multi-Task Weak Supervision Framework for Network Measurements
Jared Knofczynski, Ramakrishnan Durairajan and Walter Willinger
In IEEE JSAC Series on Machine Learning in Communications and Networks, July 2022.
[PAPER] [CODE]
- Dynamic Scheduling of Approximate Telemetry Queries
Chris Misa, Walt O'Connor, Ramakrishnan Durairajan, Reza Rejaie and Walter Willinger
In Proceedings of USENIX NSDI'22, Renton, WA, April 2022.
[PAPER]
- Revisiting Network Telemetry in COIN: A Case for Runtime Programmability
Chris Misa, Ramakrishnan Durairajan, Reza Rejaie and Walter Willinger
In IEEE Network (In-Network Computing: Emerging Trends for the Edge-Cloud Continuum), September 2021.
[PAPER] [PROJECT WEBSITE]
- ELF: High-Performance In-band Network Measurement
Joel Sommers and Ramakrishnan Durairajan
In Proceedings of TMA'21, Virtual, September 2021.
[PAPER] [CODE]
- Challenges in Using ML for Networking Research: How to Label If You Must
Yukhe Lavinia, Ramakrishnan Durairajan, Reza Rejaie and Walter Willinger
In Proceedings of Workshop on Network Meets AI & ML (NetAI'20)
co-located with ACM SIGCOMM'20, New York, USA, August 2020.
[PAPER]
- MicroMon: A Monitoring Framework for Tackling Distributed Heterogeneity
Babar Khalid*, Nolan Rudolph*, Ramakrishnan Durairajan and Sudarsun Kannan
In Proceedings of 12th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage'20)
co-located with Usenix ATC'20, Massachusetts, USA, July 2020.
[PAPER] (* co-primary authors)
- On the Practicality of Learning Models for Network Telemetry
Soheil Jamshidi, Zayd Hammoudeh, Ramakrishnan Durairajan, Daniel Lowd, Reza Rejaie and Walter Willinger
In Proceedings of TMA'20, Berlin, Germany, June 2020.
[PAPER]
- Denoising Internet Delay Measurements using Weak Supervision
Anirudh Muthukumar and Ramakrishnan Durairajan
In Proceedings of IEEE ICMLA'19, Florida, USA, December 2019.
[PAPER]
- Can We Containerize Internet Measurements?
Christopher Misa, Sudarsun Kannan and Ramakrishnan Durairajan
In Proceedings of ACM/IRTF/ISOC Applied Networking Research Workshop (ANRW'19)
co-located with IETF 105, Montreal, Canada, July 2019.
[PAPER]