Smart Stream-Based Car Information Systems that Scale: An Experimental Evaluation

Real-time information from embedded sources is hard to process if the number of sources is high. One typical example of this application area are car information systems where the cyber-physical system of a car connects to arbitrary sources in and outside of the car to deliver value-adding information to the driver of the car. In this paper, we propose a new architecture for such a car information system based on a smart data streaming infrastructure. This architecture has been implemented and we have run several experiments to examine the quality of our proposed solution. The current implementation is based on Spark Streaming, Couchbase and written in Scala. We have deployed our implementation on a distributed system using cloud services. This allows us to perform experiments with a high load as typical for the application scenario. First results from our experiments show that our proposed solution for a smart data based car information system is resilient to typical failures and scales for the described use-cases.