Making traffic-related decisions in FRIEND: A Cyber-Physical System for traffic flow related information aggrEgatioN and Dissemination

The main purpose of this paper is to present the various mechanisms whereby FRIEND - which is a Cyber-Physical System for traffic Flow Related Information aggrEgatioN and Dissemination, that was introduced in previous work - makes decisions about the state of the traffic and also about the possible occurrence of traffic-related incidents. Recall that FRIEND bases most of the inferences it makes about the status of traffic on two perceived parameters of the traffic flow: speed and density. In turn, the instantaneous density of the flow is deduced by sampling the headway distance of the most recently passing cars. A fundamental theoretical question that we address is the extent to which the sample mean of the collected headway distance data is a good approximation of the overall headway distance. In the paper, we also explain our incident detection algorithm and how we classify various types of incident. Our results show that the traffic parameters discussed in this paper along with historical data collected over a reasonable time period (duly adjusted for diurnal and seasonal variations), and with the aggregated traffic information firmly in hand, FRIEND is ready to disseminate to the traveling public a color-coded traffic status report.