Mobility in pedestrian communication simulations: Impact of microscopic models and solutions for integration

Abstract

Modeling Pedestrian-to-X (P2X) communication plays an important role for use-cases such as crowd networking, wireless contact tracing or the evaluation of Intelligent Transportation Systems (ITS) for the protection of Vulnerable Road Users (VRUs). Since in general the mobile communication depends on the position of the pedestrians, their mobility needs to be modeled. Often simplified mobility models such as the random-waypoint or cellular automata based models are used. However, for ad hoc networks and Inter-Vehicular Communication (IVC), it is well-known that a detailed model for the microscopic mobility has a strong influence — which is why state-of-the-art simulation frameworks for IVC often combine vehicular mobility and network simulators. Therefore, this paper investigates to what extent a detailed modeling of the pedestrian mobility on an operational level influences the results of P2X communication and its applications.We model P2X scenarios within the open-source coupled simulation environment CrowNet. It enables us to simulate the identical P2X scenario while varying the pedestrian mobility simulator as well as the used model. In a first step, two communication scenarios (pedestrian to server via 5G New Radio, pedestrian to pedestrian via PC5 Sidelink) are investigated in different mobility scenarios. Initial results demonstrate that time- and location-dependent factors represented by detailed microscopic mobility models can have a significant influence on the results of wireless communication simulations, indicating a need for detailed pedestrian mobility models in particular for scenarios with pedestrian crowds.While for scenarios with a focus on pedestrian communications this can be achieved by applying locomotion models available in pedestrian dynamics simulators, modeling vehicle-to-pedestrian communication is challenging: Commonly used vehicular mobility simulators apply very simplified pedestrian mobility models; pedestrian dynamics simulators often have no or very limited support for modeling vehicular traffic. Therefore, in a second step, we present a simple approach to simultaneously couple two microscopic mobility simulators (one for modeling pedestrian and one for modeling vehicular mobility) with a pedestrian communication simulation. We implement the concept and apply it in an example scenario modeling vehicle-to-pedestrian communication in a part of the city center of Munich, where we evaluate the influence on application metrics such as the accuracy of the position information received for a pedestrian.