TrafficView Prototype Moves Outdoors

A research team of faculty and students from the Rutgers University Computer Science Department led by Professor Liviu Iftode has successfully demonstrated a prototype of the TrafficView system, which uses vehicle-to-vehicle communication to prevent accidents and avoid traffic congestion. TrafficView extends drivers' awareness to aid in hazardous conditions, for example, when making a sharp turn, when a large vehicle obstructs driver's view of the road ahead, or during foggy weather. The demonstration was performed with four vehicles, which continuously exchanged speed and location information over wireless networking technology, as they navigated across the Rutgers University campus. Each vehicle was equipped with an LCD screen, a GPS device, and an embedded computer. The screens displayed a first-person perspective view of vehicles in front, flashing a warning whenever a vehicle ahead suddenly pressed the brakes. Here is a short video clip with the outdoor demonstration.

The TrafficView system was developed by the Distributed Computing Laboratory (DisCo Lab) research group at Rutgers University as part of the TrafficView Project . DiscoLab is directed by Dr. Liviu Iftode, Associate Professor in the Department of Computer Science. The TrafficView project started in 2003, while Dr. Iftode was on the faculty of the University of Maryland, and continued at Rutgers University. The first papers on the project were published in 2004, [1, 2]. The project has been funded by the National Science Foundation through research grants NSF ANI 0520123 (2005-2008) and ITR ANI 0121416 (2001-2006).

The framework defined by the TrafficView system uses vehicle-to-vehicle communication to disseminate and gather information about vehicles on the road. The initial prototype of the system was developed for PDAs running Linux. The current version is Java-based and platform-independent. Each vehicle is equipped with an embedded computer, a GPS receiver, and a wireless networking device. Maps are constructed using the publicly available Tiger/Line database. The wireless interface uses the IEEE 802.11a PHY/MAC protocol augmented with an omni-directional antenna. TrafficView does not require any road-side infrastructure.

The picture on the left shows a demonstration of the TrafficView prototype installed on four vehicles driven around the Rutgers University campus. The LCD screen displays the road ahead with the three vehicles in front. The car immediately ahead is driving at normal speed and is colored green on the screen. The driver in the car in front has just started braking, which is indicated by the red colored car on the screen. Apart from displaying the positions of vehicles, the system employs a brake detection algorithm that continuously monitors the speed of vehicles on the road ahead. Whenever the system detects that a driver in a vehicle in front has pressed the brakes, a warning flashes on the screen and, if the vehicles get too close to each other, an audible alert is emitted. Alerts are essentially hints to the driver to take preventive measures. In many modern vehicles, traditional mechanical-brake systems have been replaced with "drive by wire" electronic brakes, which enables a system, such as TrafficView, to automatically apply brakes, under certain conditions.

One long-term trend in safety design for future high-tech vehicles is active accident-prevention. Some modern vehicles already have adaptive cruise control systems. These use either radar or laser setups to allow a vehicle to slow when approaching another vehicle and accelerate again, to the preset speed, once the vehicle is clear of obstructions. A fundamental limitation of these systems is that they can only sense the immediate environment around the vehicle. Unlike a laser or radar based system, which requires a clear "line-of-sight" between vehicles, TrafficView relies upon wireless networking technology to disseminate inter-vehicular information. This type of system makes it possible to avert chain collisions involving three or more vehicles, where "line-of-sight" information is unavailable. The only limiting factor is the range of wireless communication, which extends to well over a hundred meters.

In addition to generating real-time safety alerts, TrafficView enables drivers to find an optimal route to his or her destination, both in the city and on the highway. The system displays a map that shows the driver's position along with a dynamic view of the traffic ahead, based exclusively on data provided by other cars. In collaboration with the University of Cyprus, the Rutgers team developed the Vehicular Information Transfer Protocol (VITP) , which allows cars to become ad-hoc information service providers for other cars on the road. Using VITP, drivers can issue queries about the traffic status on any road ahead.

Apart from designing protocols for communication among vehicles, ensuring the security of these systems is a challenging problem. Conflicting goals, such as privacy and authentication, must often be taken into account. One of the main problems in vehicular network security is the detection of false data sent by a malicious vehicle. For example, drivers in a hurry may try to divert traffic by falsely reporting a traffic jam on the road ahead. Or a vehicle might report false location information about itself and other vehicles, so as to deny involvement in an accident. Dr. Iftode's team has designed a solution that probabilistically detects malicious vehicles that generate false information.

The TrafficView system is a part of the E-Road project of Dr. Iftode's Discolab, which also includes the Active Highways with Reservation Lanes (in collaboration with UCLA and PATH UC Berkeley), the Adaptive Traffic Lights (in collaboration with UPB Romania), and the EZCab, a cab booking application using short-range wireless communication. This vision is to design and deploy intelligent inter-vehicle systems in order to help in traffic safety, traffic management, and multimedia delivery to vehicles, by designing and implementing an inter-vehicular network architecture on top of an embedded vehicular computing platform. More about the TrafficView project can be found at http://discolab.rutgers.edu/traffic/index.htm