Cooperative Wireless Sensing for Characterization of Congested Traffic States E. del Arco*, A.B. Rodr´ ıguez-Gonz´ alez*, M. Wilby*, J. Ramiro*, J.J. Vinagre*, D.F. Llorca**, M.A. Sotelo**, J. Ramos* and A.J. Caama˜ no* Abstract— In this work we propose the use of vehicles as traffic sensors to quasi-synchronously measure both velocity and position of the probe. Those sensor vehicles wirelessly cooperate to relay that distributed information to a Data Fusion Center. That Data Fusion Center, in turn, calculates the Spatio- Temporal Velocity (STV) Field of the traffic from the gathered data. From the STV Field, it has been previously shown that Congested Traffic States (CTS) can be fully characterized. The actual distribution of stationary probes (induction loops, cameras, ...) used to reconstruct the STV Fields is usually very sparse with relation to the space-time variability of the CTS features of dense traffic. We propose the Wireless Sensor Network System for Early Detection of CTS. As a first result, we present the error incurred in reconstructing the STV Field with an increasing density of sensor vehicles. We show that with a fraction of sensor vehicles sensing communicating their position and speed as low as 10% is enough to stabilize the error. We also show the effect of including finite precision in the positioning system. I. I NTRODUCTION In high density traffic areas surrounding big cities such as belt highways, monitoring variations in the flow of vehicles is critical in many aspects ranging from economic impact to quality of life. The usual method to monitor these beltways is by means of induction coils laid under the tarmac which are able to count the number of cars per unit time, measure velocity and even discriminate between types of vehicles. This data is measured, averaged and communicated with a frequency of tens of seconds to the monitoring centre. However, the installation, operation and maintenance of these coils is costly and cumbersome as it can affect the normal flow of vehicles. Because of this economic aspect, the coils are usually spaced hundreds of metres apart (See Fig. 1). However, it has been recently shown that, when metastable congested traffic states appear (e.g. without the need of bottlenecks) the spatial features of the oscillating traffic flow are but a few car lengths apart [1]. Therefore, it remains to be seen that the present accepted setup for measuring traffic flow in highways, with sensors placed hundreds of car lengths apart capture the essential features of those metastable states that usually precede highly congested traffic. On the other *Authors are with the Department of Signal Theory and Communications, University Rey Juan Carlos (URJC), Camino del Molino, s/n, 28943, Fuenlabrada (Madrid, Spain). **Author is with the Department of Electronics, Alcal´ a de Henares University (UAH), Carretera Madrid-Barcelona, km 33,600, 28871 , Alcal´ a de Henares (Madrid, Spain) Email (Corresponding Author): antonio.caamano@urjc.es. This work was supported by the Spanish Ministry of Industry (MITYC) through the GUIADE Project and by the Spanish Ministry of Science and Innovation (MICINN) through project TEC2009-12098. 10 12 14 16 18 20 3 4 5 6 7 8 20 40 60 80 100 a) b) speed (Km/h) time (hours) space (Km) Fig. 1. a) Position of speed sensors in the north-to-south lanes of the M-30 beltway in Madrid, Spain, near the “Puente de Ventas” congestion point. The third digit in each identification number indicates the position of the coil in the beltway in kilometres. b) Spatio-temporal velocity field measured by the aforementioned sensors on November, 8, 2009. hand, the exact classification of Congested Traffic States (CTS) remains to be analytically described, but a few works have tried to characterize them in terms of microscopic (indi- vidual driver) behaviour of the traffic [2]. The distinction of the Space-Time Velocity (STV) field characteristics of CTS such as Stop-and-Go Waves (SGW), Oscillating Congested Traffic (OCT) or Widening Synchronized Patterns (WSP) is crucial to understand their causes and predict their transition times to Homogeneous Congested Traffic (HCT). In this work we propose the use of the Wireless Sensor Network (WSN) paradigm that make use of a fraction of the vehicles as sensors the communicate via wireless measures of their individual velocities and positions to roadside wireless bridges or clusterheads. These, in turn, forward this space- time distributed information to a Data Fusion Center which accurately reconstructs the STV Field. This enables the accurate classification and characterization of CTS to predict congestions in the future. The structure of the work stands as follows: In Section II we will describe the outline of the system of vehicular WSN, based on the communications system of IEEE 802.11p standard. The limits on the communication channel are described. The problem of the distribution of the sensor vehicles to accurately probe the STV field is addressed. In Section III we present the results of microscopic simulations of traffic and the attempt to reconstruct the STV Field with a varying Fraction of Sensor Vehicles (FSV). We also present the results of the Mean Squared Error (MSE) with an increasing FSV, evaluating the effect of finite precision in positioning systems. Finally in Section IV we present our conclusions and future extensions of the present work. IEEE ITSC2010 Workshop on Emergent Cooperative Technologies in Intelligent Transportation Systems