# LightSim: A Lightweight Cell Transmission Model Simulator for Traffic Signal Control Research

arXiv: https://arxiv.org/abs/2602.21852 · HTML: https://mlsanity.com/paper/2602.21852/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in RL for autonomous driving and traffic, one of 15 directions in Deep reinforcement learning methods.
- Builds on 9 of its 27 references that are in the map.
- First posted on arXiv 25 Feb 2026.

## Placed in

- Deep reinforcement learning methods / [RL for autonomous driving and traffic](https://mlsanity.com/deeprl/rl-for-autonomous-driving-and-traffic/)

## Builds on (most cited first, 9 of 9)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://mlsanity.com/paper/2106.01345/) (2021)
- [PettingZoo: Gym for Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2009.14471/) (2020)
- [CoLight: Learning Network-level Cooperation for Traffic Signal Control](https://mlsanity.com/paper/1905.05717/) (2019)
- [Learning Phase Competition for Traffic Signal Control](https://mlsanity.com/paper/1905.04722/) (2019)
- [Flow: A Modular Learning Framework for Mixed Autonomy Traffic](https://mlsanity.com/paper/1710.05465/) (2017)
- [AttendLight: Universal Attention-Based Reinforcement Learning Model for Traffic Signal Control](https://mlsanity.com/paper/2010.05772/) (2020)
- [A Decentralized Reinforcement Learning Framework for Efficient Passage of Emergency Vehicles](https://mlsanity.com/paper/2111.00278/) (2021)
- [Hierarchical GNN-Based Multi-Agent Learning for Dynamic Queue-Jump Lane and Emergency Vehicle Corridor Formation](https://mlsanity.com/paper/2601.04177/) (2026)

## Abstract (arXiv)

Reinforcement learning for traffic signal control is bottlenecked by simulators: training in SUMO takes hours, reproducing results often requires days of platform-specific setup, and the slow iteration cycle discourages the multi-seed experiments that rigorous evaluation demands. Much of this cost is unnecessary, since for signal timing optimization the relevant dynamics are queue formation and discharge, which the Cell Transmission Model (CTM) captures as a macroscopic flow model. We introduce LightSim, a pure Python, pip-installable traffic simulator with Gymnasium and PettingZoo interfaces that runs over 20000 steps per second on a single CPU. Across cross-simulator experiments spanning single intersections, grid networks, arterial corridors, and six real-world city networks, LightSim preserves controller rankings from SUMO for both classical and reinforcement learning strategies while training 3 to 7 times faster. LightSim is released as an open-source benchmark with nineteen built-in scenarios, seven controllers, and full reinforcement learning pipelines, lowering the barrier to signal control research from days to minutes.