Discrete Event Traffic Simulation Framework using Object-Driven Cellular Automata

Cellular automata have run on discrete-event engines since Cell-DEVS, but the cell remains active and the moving object a value in its state. This paper inverts that assignment for traffic microsimulation. In the Object-Driven Cellular Automata (ODCA) framework, each vehicle is an asynchronous process and each cell is a resource of capacity one, so behaviour attaches to the vehicle; the cell holds state but no transition rules. A vehicle acquires cells through a request–wait–seize–delay–release protocol, congestion emerges from resource contention rather than from cell rules or explicit gap checks, and speed is a real number rather than a discrete cell state, removing the integer-speed artifact of cell-based models, event-driven ones included. The delayed release produces headways consistent with Newell's car-following model, yielding a triangular fundamental diagram without calibration. Drivers re-evaluate speed and direction when neighbours move rather than at fixed intervals, so computation concentrates where vehicles interact. We derive closed-form expressions linking the protocol parameters to capacity, critical density and backward wave speed, and extend them to mixed autonomous and human-driven traffic. In 20 replications on a four-lane freeway offered 7,000 veh/h, raising the autonomous share from 0% to 70% increases served throughput by 18.7% (from 5,885 ± 24 to 6,983 ± 35 veh/h, 95% CI) and cuts average delay by 96% (from 215 ± 5 to 8.6 ± 0.3 s). Throughput understates the capacity effect once the network serves 99.8% of the demand offered, after which the benefit appears as delay reduction, at a lane-drop bottleneck as well.