Erlang-based Desynchronised Urban Traffic Simulation (...)
summary
Implements a micro‑scale urban traffic simulator that accepts controlled desynchronisation to avoid full barrier sync and still respect the cellular‑automata model, using Erlang distribution with hidden nodes and a compact codebase based on maps. Experiments report near linear scalability to 19,200 cores on a supercomputer with 11M cars and 240k spaces. Practical tweaks include blocking lane‑changes at borders and limiting desync to a distance threshold; the prototype is ~400 lines of Erlang and runs successfully.
AI-written summary · about
chapters (10)
- 0:00Research motivation
- 0:34Simulation objectives
- 0:53Importance of detail
- 3:50Erlang in HPC
- 4:09Traffic model design
- 5:31Implementation and scaling
- 14:06Supercomputer challenges
- 15:00Performance results
- 19:51Future research
- 21:24Audience Q&A
related talks
48mSynchronizing Game Components
36mSimulating the Chaos of Real-World Networks for Resilient Applications
44mSeeing Real-Time Traffic Problems
17mSignal priority for improving fluidity and decreasing fuel consumption
01h 00mExploring Traffic Jams in Your Data Flows
14mAnalysis and Modelling of Road Traffic Using SUMO to Optimize the Arrival Time of Emergency Vehicles
19mSensor-based Flow Optimization on connected real-world intersections via a SUMO Feature Gap Analysis
43mRealtime performance at scale (...)