The Software Campus project ADAM addresses the connection between approximated analysis of data streams and the advantages of modern hardware architectures.
In the current era of the internet of things and industry 4.0, the quantity and velocity of data sources available in form of continuous data streams increases drastically. Keeping data analysis up to this increase is essential for current and future data-driven services like fraud detection or predictive maintenance.
However, data stream analysis is usually performed in clusters consisting of standard server hardware. Analyzing more data in these systems means increasing the size of the cluster which, in turn, increases cost for hardware and power consumption. Furthermore, increasing the size of the cluster demands highest scalability from stream processing systems and used algorithms. Thus, next-generation stream processing systems have to be designed for maximum cost-efficiency and scalability.
The goal of the ADAM project is to develop concepts for cost-efficient and scalable stream processing system based on stream summaries and specialized hardware. Using stream summaries allows for trading off accuracy for more scalable and efficient processing. Evaluating and analyzing these summaries using targeted compute architectures (e.g., FPGAs, GPUs) reduces the amount of required general-purpose hardware and can significantly decrease power consumption.
Project Duration: 04/2018 - 12/2020
Supervisor: Prof. Dr. Volker Markl