SUMO-Based Evaluation of Edge-AI for Intelligent Traffic Control and Emergency Vehicle Preemption

Abstract

Urban traffic congestion can be a critical threat to emergency response times, where delays of even a few seconds can determine survival outcomes. While Intelligent Transportation Systems (ITS) have begun to replace traditional fixed-timer traffic lights, most current solutions depend on centralized Cloud computing. This approach introduces unpredictable network latency, particularly during 4G/5G congestion, which creates "tail latency" risks for autonomous emergency braking and signal preemption in real-time traffic management systems. This paper compares two approaches using a theoretical computer simulation study. One approach processes traffic decisions in the cloud, and the other processes them locally using Edge AI, meaning the intelligence is placed close to the traffic lights. By utilizing the SUMO (Simulation of Urban MObility) traffic simulator integrated with a controlled mathematical network emulation layer, this study showed that Edge AI achieved a mean response time of approximately 56 ms, substantially below the modeled Cloud architecture. This approach helps emergency vehicles move through traffic more reliably and theoretically reduces the instability issues associated with cloud architecture.