On Probabilistic Data Collection in the NOTICE Architecture

NOTICE is a secure, privacy-preserving architecture for the automatic detection of traffic incidents and the dissemination of related traffic advisories. NOTICE uses belts of piezo-electric elements embedded in the highways to detect variations in the attributes of traffic flow. By using their piezo-electric elements the belts detect passing vehicles and initiate meaningful interaction with them. NOTICE learns about traffic conditions by collecting data from passing vehicles and, as a rule, this data is highly correlated. Thus, there is no need to collect data from all vehicles, especially in dense traffic, and one of the important issues we investigate in this work is probabilistic data collection where instead of collecting traffic-related data from all vehicle, the belt flips a coin and only collects data from a vehicle with some application-dependent probability p. Vehicular Networks and Traffic-related systems are just like mission-oriented sensor networks, both are time-varying systems consist of both humans and sensors that collect, communicate and coordinate to accomplish a near real-time solution or mission. The main contribution of this work is to analyze various parameters of probabilistic data collection in the NOTICE architecture and to reveal how they impact data aggregation and the time to detect traffic-related events. Extensive simulations of various traffic regimens have confirmed the accuracy of our analytical predictions.