A High Performance Data Acquisition Framework for Decentralized Scientific Experiments

Detectors are an essential component of most scientific experiments like particle accelerators. Yet, as the scientific process continues to evolve, these detectors are continuously upgraded to fulfil high-performance requirements e.g. cope with high-speed readout and data rates in the Giga- or Terahertz range.

The increasing demand for these fast detectors raises two technical challenges:

(1) How can data from the detector with such high data rates be collected efficiently?

(2) How can the access to the detector data be organized especially in decentralized environments?

To overcome the high-speed demands, high-performance networking technologies are utilized. An example is Remote Direct Memory Access (RDMA), which allows collecting detector data quickly and efficiently without involving the kernel of the host operating system (OS). However, these technologies only help in reading the detector data, but not in managing data flows between high-speed detectors, and to the interested data consumers, e.g. scientists who want to read and analyze the scientific data.

In this Master’s Thesis, you will build a novel high-performance data acquisition framework that will help scientists to manage and access measurement data and process information in a decentralized environment. The framework will extend EPICS, a commodity control and data acquisition system commonly used to control large scientific instrumentation. You are expected to gain familiarity with the EPICS protocol, and with the concepts of high-performance RDMA data transfer. You will then extend the base functionality of EPICS by integrating RDMA capabilities into the framework such that it is suited to handle large data streams.

Personal qualification

• Excellent skills in C/C++ programming
• Very good understanding of Distributed Systems and Computer Network