Performance assessment framework for Vehicle-to-Everything as a Sensor (V2XaaS)

Vehicle-to-Everything (V2X) communication can extend the Field of View (FoV) of a traditional Automated Vehicle (AV) by extending the perception of onboard sensors with shared information. This information can be active speed limits, accident notifications, or perception data shared between vehicles. There are no limits on what can be shared. In fact, the community encourages sharing information about one’s surroundings.

However, when using Vehicle-to-Everything–as–a–Sensor (V2XaaS), the Automated Driving (AD) stack in an AV must be aware of the currently provided quality of the information shared via V2X communication to avoid degradation of overall performance and especially safety. Specifically, oversharing information leads to loaded communication channels, and sharing data with low accuracy may be useless.

Related research provides only in part an overview of the components involved V2XaaS and their performance, let alone a metric to rate the quality of individual influencing factors or the overall quality. Specifically, related research in the AD domain gives the impression of primarily focusing on improving perception results and assuming that the V2XaaS framework does not affect the quality of the data available. In contrast, the V2X domain gives the impression of expecting perception data to be perfect and only having to deal with wireless communication characteristics.

The work in this thesis aims to provide a sound foundation for integrating and evaluating V2XaaS systems. This includes an overview of the components involved, the performance linked to each component, and a metric to rate the Quality of Service (QoS) provided by the surrounding V2X field.

The work defines the components of V2XaaS and their key performance indicators. This analysis led to the setup of a demonstrator system for laboratory and field experiments that was used to perform an extensive systematic analysis of components. This analysis aims to provide well-founded ranges for the performance indicators, focusing on accuracy and delay. Those well-founded areas are then used to create a metric rating the QoS of the surrounding V2X field.

The resulting metric allows the AD system of an AV to adapt its behavior depending on

the quality of the surrounding V2X field, which in turn allows it to keep up or even improve comfort, performance, and safety.

With the systematic evaluation, we aim to ease and broaden the application of V2XaaS. Even more, the contributions we made allow AD systems to have an indicator of the QoS that V2XaaS can provide at a specific moment.