Abstract:

Urban traffic congestion leads to increased travel delays, fuel consumption,and pollution, especially at busy intersections where traditional fixed-time signals fail to adapt to dynamic traffic conditions. TrafficOS is a real-time trafficlight scheduling system designed to intelligently control signal phases basedon live traffic data received from sensors or cameras. The system models intersections using graph structures to identify conflict-free signal phases and applies queue-based analysis to estimate lane-wise waiting vehicles. A real-timescheduling framework, incorporating Rate Monotonic Scheduling (RMS) andEarliest Deadline First (EDF) algorithms, is used to determine optimal greentime durations while ensuring strict safety constraints and deadline guarantees.The system integrates a microcontroller-based actuator module for deterministic switching of traffic lights using hardware timers and interrupt routines. Livedata processing and decision-making are handled at the edge to avoid networklatency and enable decentralized control. Simulation and prototype resultsshow a significant improvement in traffic throughput, reduced average vehiclewaiting time, and zero conflicting signal activations. Compared to fixed-timecontrollers, TrafficOS demonstrates a notable reduction in congestion, proving the effectiveness of real-time scheduling in smart traffic management. Thissolution forms a scalable foundation for future smart city transportation infrastructure.