WiBus: A Wi-Fi based Monitoring System for Public Transportation with Dynamic Route Tracking Vitor Borges C. da Silva, Tatiana Sciammarella, Miguel Elias M. Campista, and Lu´ ıs Henrique M. K. Costa Universidade Federal do Rio de Janeiro - PEE/COPPE/GTA - DEL/POLI E-mail:{borges,tatiana,miguel,luish}@gta.ufrj.br Abstract—A great challenge of today’s metropolises is the constant traffic jams, which are a consequence of excessive use of private vehicles. More people would adopt public transportation, e.g. buses, if the time of arrival (ToA) for the next vehicle to each bus stop could be predicted. In this direction, we propose WiBus, which is a system to estimate the buses ToA, based on information from opportunistic IEEE 802.11 contacts. Unlike previous systems, WiBus explicitly takes into account bus route changes, adjusting the routes with an algorithm for dynamic route creation and maintenance. The system is implemented and analyzed via emulation of a real scenario. Experimental results show that WiBus has accumulated error of at most a few minutes. I. I NTRODUCTION Today, big cities suffer with constant traffic jams generated by an increasing number of vehicles. In Brazil, in just one year, from September 2012 to 2013, almost 3 million new vehicles started circulating on the streets, causing a signif- icant deterioration of the traffic system [1]. The preference for private transportation is a consequence of the lack of confidence on public transportation commitment with timeta- bles. To change this trend, the idea of the Advanced Public Transportation Systems (APTS) has emerged using, among other technologies, communication networks to provide more information to users [2]. Communication networks in vehicular scenarios predominantly use cellular (3G/4G) or IEEE 802.11 technologies. This work aims at improving the quality of service provided by public transportation systems. The idea is to offer fresh information regarding the ToA of the next bus of every bus line to each possible bus stop. Nevertheless, estimating the ToA of a bus at a given bus stop is a challenge, because traffic and passengers can influence the time a bus takes in its route. Moreover, situations such as an accident or a roadwork may result in changes in a bus route. If these changes were ignored, the estimates would become useless. Therefore, two requirements to calculate the ToA without significant errors are to track buses position; and to be aware of buses route changes in a dynamic fashion, which is a contribution of our work. To deal with the first requirement, many solutions use GPS, due to its greater accuracy. However, the GPS use requires an additional communication link to inform vehicle position. As an alternative, there are localization techniques that use network technologies [3], which can provide both services of locating a bus and sending the corresponding information to an access point. This potential has been explored even in simple proximity-based localization systems. In such systems, the position of a vehicle is given by the position of the node the vehicle is connected to. Regarding estimates of buses ToA, some works use historical series collected even for months [4]. Then, the estimate is calculated based on averages from the same period of time in the past. Hence, this method does not behave well in case of atypical situations, such as accidents. In opposition to historical-based approaches, in real- time solutions is assumed that the time taken by previous vehicles in a given route is also the time next vehicles will spend [5]. Thus, only information from the same day is used, allowing this method to model unforeseen situations more efficiently. Finally, there are methods that use Kalman filters, which do not perform well when location data are temporally sparse [6]. This paper proposes WiBus, a system to estimate the buses ToA using IEEE 802.11 networks. WiBus tracks the position of buses based on their proximity to IEEE 802.11 access points installed alongside streets. This allows the use of only one type of device to track and communicate, reducing the system complexity. In this work, as localization information is not obtained at the same rate as with a GPS, and because we do not assume the existence of a database with past information, we use a real-time method to estimate ToAs. Based on information from previous buses, WiBus computes the ToA of a bus and dynamically adapts these times to route changes. Therefore, even if a bus deviates from its route, the system can adapt and keep users informed. WiBus estimates incur in errors on the order of a few minutes, as obtained from experiments with real datasets. Our experimental results show the benefits of using WiBus in real environments. This paper is organized as follows. Section II and III present, respectively, the architecture and the implementation of WiBus. In Section IV, the evaluation setup is described and the corresponding results are shown. Finally, Section V concludes this work and presents future directions. II. WI BUS ARCHITECTURE The WiBus system has four types of entities: Central, Roadside Unit (RSU), Bus, and Client. The Central entity plays the most important role in the system. It has all the information needed to compute ToA estimates of buses, their route, location, and the time spent between consecutive stops. The Central also answers customers’ ToA requests. Roadside Units (RSUs) are access points installed at bus stops. They provide wireless access to each router within each bus. These