I’m Kyle Simpson Orcid Link, a Postdoctoral Research Assistant and final-year PhD student in Computing Science at the University of Glasgow, supervised by Professor Dimitrios Pezaros. I work in the Glasgow Systems Section on the intersection of programmable dataplane hardware and data-driven networking.
I take interest in research of network programmability, reinforcement learning, computer graphics and combinatorial search, and enjoy hobbyist game development and cooking.
I love solving hard problems in new fields—feel free to email me at any time.
Blog
- (Almost) Lockless Stream Buffering: 27 April 2020.
- Thoughts on Vana'diel: 31 May 2019.
- Felyne Bot, and Initial Thoughts on Rust: 27 February 2018.
- Custom Open vSwitch Actions: 4 December 2017.
Recent Talks/Posters
-
Poster: Online RL in the Programmable Dataplane with OPaL:
2 December 2021,
The 17th International Conference on emerging Networking EXperiments and Technologies, Munich, Germany (Virtual).
Abstract - PhD Skills Workshop Session: Adapting Your Project to Work During COVID: 13 September 2021, SICSA Conference 2021 – Sustainability and Resilience.
-
Revisiting the Classics: Online RL in the Programmable Dataplane:
4 June 2021,
NGN Webinar.
-
Seiðr: Dataplane Assisted Flow Classification Using ML:
10 December 2020,
IEEE Globecom 2020, Taipei, Taiwan (Remote).
Video Recording -
Towards In-Switch Reinforcement Learning:
2 December 2020,
Fifth Annual UK Systems Research Challenges Workshop.
Recent Publications and Conference Papers
-
Seiðr: Dataplane Assisted Flow Classification Using ML:
Kyle A. Simpson, Richard Cziva, Dimitrios P. Pezaros.
IEEE GLOBECOM 2020.
Author-final PDF DOI -
Per-Host DDoS Mitigation by Direct-Control Reinforcement Learning:
Kyle A. Simpson, Simon Rogers, Dimitrios P. Pezaros.
IEEE TNSM, March 2020.
Author-final PDF DOI Source code -
On Maximum Weight Clique Algorithms, and How They Are Evaluated:
Ciaran McCreesh, Patrick Prosser, Kyle Simpson, James Trimble.
CP 2017.
Author-final PDF DOI Benchmark instances