Comparing Message Propagation Approaches in Vehicular Ad Hoc Networks

Vehicular Ad-Hoc Networks (VANETs) enable direct vehicle-to-vehicle communication for Intelligent Transportation Systems (ITS), supporting safety applications such as collision avoidance and emergency message dissemination. Ensuring fast and reliable multi-hop broadcasting in urban environments remains challenging due to interference, dynamic topology, and variable connectivity. This thesis presents the design and evaluation of the Deterministic Fast-Broadcast (DFB) protocol, conceived as an evolution of Fast-Broadcast (FB), which was originally developed for 1D environments and later extended to 2D and 3D. FB employs adaptive transmission range estimation and probabilistic contention to prioritize distant vehicles while maintaining robustness under diverse conditions. DFB modifies this logic by introducing deterministic distance-based backoff times, ensuring that the farthest nodes always transmit first. The objective is to improve dissemination speed and consistency while retaining FB’s reliability and minimal network overhead. DFB was implemented in the NS-3 simulator and evaluated under four timing configurations, combining different slot lengths and granularities. Its performance was compared with both FB and the deterministic Robust and Fast Forwarding (ROFF) protocol. Simulations were primarily conducted in a realistic two-dimensional urban environment modeled on the city of Los Angeles, capturing real propagation constraints such as signal shadowing by buildings. Additionally, FB was validated in a three-dimensional Cube scenario to assess its scalability under ideal propagation conditions. Results show that DFB achieves excellent delivery ratios—often exceeding FB and ROFF—but at the cost of increased collisions, higher slot delays, and greater sensitivity to timing parameters and range estimation errors. FB remains the most robust scheme with minimal overhead, while ROFF provides the fastest dissemination under favorable conditions but requires frequent topology updates and high communication overhead, making it vulnerable to VANETs dynamics. These findings show that deterministic forwarding without complete neighbor awareness cannot reliably outperform well-designed probabilistic schemes in realistic vehicular scenarios.