The construction and maintenance of roads is expensive. Metrics such as traffic speeds, volumes, patterns and trends heavily influence how roads should be built and maintained. Improving the data quality of your metrics means you can plan and execute roadwork more efficiently.

We have built an infrastructure consisting of distributed sensors and signal processing devices, a data ingestion application, and a NoSQL database optimized for analytics. The system registers information about vehicles passing measurement stations located along major roads in Norway and provides continuously updated reports and low-latency traffic event information.

While building the system and integrating it with an increasing number of measurement stations, we have encountered a wide variety of architectural challenges. We believe the main reasons are: a very diverse technology stack, a need for continuously cost-effective scalability, and a desire to keep the architecture simple and agile.

The presentation will go through a selection of challenges we have met and provide our thoughts and reasoning behind the choices we made underway. Outcomes, both positive and negative, intended and unintended, will be discussed.