Models and Performance of VANET based Emergency Braking

The network research community is working in the field of automotive to provide VANET based safety applications to reduce the number of accidents, deaths, injuries and loss of money. Several approaches are proposed and investigated in VANET literature, but in a completely network-oriented fashion. Most of them do not take into account application requirements and no one considers the dynamics of the vehicles. Moreover, message repropagation schemes are widely proposed without investigating their benefits and using very complicated approaches. This technical report, which is derived from the Master Thesis of Michele Segata, focuses on the Emergency Electronic Brake Lights (EEBL) safety application, meant to send warning messages in the case of an emergency brake, in particular performing a joint analysis of network requirements and provided application level benefits. The EEBL application is integrated within a Collaborative Adaptive Cruise Control (CACC) which uses network-provided information to automatically brake the car if the driver does not react to the warning. Moreover, an information aggregation scheme is proposed to analyze the benefits of repropagation together with the consequent increase of network load. This protocol is compared to a protocol without repropagation and to a rebroadcast protocol found in the literature (namely the weighted p-persistent rebroadcast). The scenario is a highway stretch in which a platoon of vehicles brake down to a complete stop. Simulations are performed using the NS_3 network simulation in which two mobility models have been embedded. The first one, which is called Intelligent Driver Model (IDM) emulates the behavior of a driver trying to reach a desired speed and braking when approaching vehicles in front. The second one (Minimizing Overall Braking Induced by Lane change (MOBIL)), instead, decides when a vehicle has to change lane in order to perform an overtake or optimize its path. The original simulator has been modified by - introducing real physical limits to naturally reproduce real crashes; - implementing a CACC; - implementing the driver reaction when a warning is received; - implementing different network protocols. The tests are performed in different situations, such as different number of lanes (one to five), different average speeds, different network protocols and different market penetration rates and they show that: - the adoption of this technology considerably decreases car accidents since the overall average maximum deceleration is reduced; - network load depends on application-level details, such as the implementation of the CACC; - VANET safety application can improve safety even with a partial market penetration rate; - message repropagation is important to reduce the risk of accidents when not all vehicles are equipped; - benefits are gained not only by equipped vehicles but also by unequipped ones