Towards Request Arbitration in Edge-Assisted Smart Intersections Under Timing Constraints

Infrastructure-assisted autonomous driving has the potential to improve the decision-making process of autonomous driving. The car's perception can be augmented with information collected by infrastructure, such as smart intersections, that otherwise cannot be perceived by the car itself. This additional information enables more informed decisions, improving road safety. At the roadside unit, functionality is realized by a chain of tasks. Requests sent by autonomous cars must be handled within a specific time, while it is additionally crucial that the data age of returned data is within a defined bound. In this work, we discuss data age and response time as two competing metrics in data requests from an autonomous car to a smart intersection. A novel policy to arbitrate pending requests of a server is described that utilizes properties of the task chains that follow the Logical Execution Time paradigm to meet both response time and data age constraints of requests. Evaluations demonstrate the importance of the proposed method by improving the acceptance ratio of requests by up to 36 % compared to a server that does not consider data age constraints.