Interfacing Detectors and Collecting Data for Large-Scale Experiments in High Energy Physics Using COTS Technology

Data-acquisition systems for high-energy physics experiments like the ATLAS experiment at the European particle-physics research institute CERN are used to record experimental physics data and are essential for the effective operation of an experiment. Located in underground facilities with limited space, power, cooling, and exposed to ionizing radiation and strong magnetic fields, data-acquisition systems have unique requirements and are challenging to design and build. Traditionally, these systems have been composed of custom-designed electronic components to be able to cope with the large data volumes that high-energy physics experiments generate and at the same time meet technological and environmental requirements. Custom-designed electronics is costly to develop, effortful to maintain and typically not very flexible. This thesis explores an alternative architecture for data-acquisition systems based on commercial off-the-shelf (COTS) components. A COTS-based data distribution device called FELIX that will be integrated in ATLAS is presented. The hardware and software implementation of this device is discussed, with a specific focus on performance, heterogenity of systems and traffic patterns. The COTS-based readout approach is evaluated in the context of the future requirements of the ATLAS experiment. The main contributions of the thesis are an analysis of the ATLAS data-acquisition system with a focus on the readout system, a software architecture for the main application on FELIX hosts, a performance analysis and tuning based on computer science methods for central FELIX software components with respect to the requirements of the ATLAS experiment, a network communication library with a high-level software interface to utilize high-performance computing network technology for the purpose of data-acquisition systems, and an evaluation and discussion of ATLAS data-acquisition using FELIX systems as a case study for COTS-based data-acquisition in high-energy physics.