VANET-based Smart Navigation for Emergency Evacuation and Special Events Ahmed Elbery School of Computing Queen’s University Kingston, Ontario, Canada aelbery@cs.queensu.ca Hossam S. Hassanein School of Computing Queen’s University Kingston,Ontario, Canada hossam@cs.queensu.ca Nizar Zorba EE Department Qatar University Doha, Qatar nizarz@qu.edu.qa Hesham A. Rakha CEE Department Virginia Tech Blacksburg, Virginia, USA hrakha@vt.edu Abstract—In this paper we propose, develop, and analyze the performance of a new system-optimum navigation model that uti- lizes Vehicular Ad-hoc Networks (VANETs), linear programming optimization, and stochastic routing to efficiently and smartly navigate vehicle crowds in case of an emergency evacuation or after special events. The objective of the proposed system is to clear the network in a shorter time by better utilizing the network resources while taking into consideration the road capacities. In this model, road links are weighted based on travel time. Road link capacities and current traffic conditions are used as constraints in the optimization problem. Vehicles are employed as sensors to compute travel times of the links and send this information to the Traffic Management Center (TMC) in real-time. The TMC periodically optimizes the traffic assignment. Subsequently, routes for vehicles are created/updated based on the latest optimized assignments. To test the model, a real network with calibrated traffic is used. The proposed model is compared to the Sub-population Feedback Dynamic Time-dependant Assignment (SFDTA) navigation. Moreover, we analyze its sensitivity to the re-optimization interval at different traffic demand levels. The results show that the proposed system decreases the network-wide travel time and is successful in clearing the network earlier especially in the case high vehicle traffic demands. Index Terms—VANET-based Navigation, Smart Cities, Crowd Management, Stochastic Routing, Constrained Routing. I. I NTRODUCTION Intelligent Transportation Systems (ITSs) employ advanced navigation techniques to improve mobility by reducing travel time [1], [2], fuel consumption, and the environmental impact from the transportation sector [3]. However, navigation sys- tems that perform well in day-to-day traffic conditions are not suited for some non-recurrent events such as large sporting events or emergency evacuation, in case of natural disasters. In such events, there is a large number of vehicles (vehicular crowds) needs to be routed and exit the crowd area. Most of the current navigation techniques will not work efficiently in such situations because most of them utilize best path routing models. The main problem with best path routing is the single cost function. These techniques do not account for other parameters such as road capacity, traffic volume, or underutilized alternative routes. Moreover, at any given time, shortest path navigation techniques provide the same guidance for all vehicles based on their destinations, causing the shortest path to collapse, while other longer paths are underutilized. It also can result in route oscillations and unstable global traffic behavior [4]. The challenge in non-recurrent events is two-fold. First, the sheer volume of traffic that needs to leave the event area can cause serious congestion and may result in network grid-lock. Consequently, adversely affecting the mobility (i.e., increasing travel time, fuel consumption, and emission levels). Second, the road networks are not designed to support such a high traffic demand, basically, because of the required high road capacities [5] and special traffic control techniques [6]. Thus, vehicle crowds combined with road network resource constraints, bring forward the need for efficient and smart management techniques that better utilize the network facilities while taking into consideration the road capacity constraints. Inspired by Vehicular Ad-hoc Networks (VANETs) [7] the advancement in information technology, vehicular navigation based on real-time information shows potential benefits for crowd management [8] to reduce travel time [2] and en- ergy/fuel consumption [9], [10]. Utilizing VANETs in real- time navigation brings new opportunities to address the prob- lem of vehicular crowd navigation. Therefore, this paper focuses on utilizing VANET com- munication, to efficiently and smartly route vehicular crowds to minimize the network-wide travel time and to clear the network faster. The contributions of this paper are: • Proposing and developing a system-optimum navigation model for vehicular crowds, • Using a real network and a calibrated traffic to test the proposed model, and • Performing sensitivity analysis against two system param- eters; traffic load and re-optimization interval. Firstly, the proposed system uses vehicles as network sensors and utilizes VANETs as an infrastructure to collect road state-conditions (traffic volume and travel time on each road segment) in real-time. The collected information is used along with historical information to optimize the network- wide vehicular traffic-assignment using linear programming (LP) [11]. In the optimization problem, the road capacities and current traffic loads are used to constrain the network congestion. Then, a stochastic route construction algorithm is used to create/update the vehicles’ routes. The main idea behind the proposed model is to efficiently utilize all the network resources, allowing vehicles traveling 978-1-7281-1016-5/19/$31.00 ©2019 IEEE