Dr. Miroslaw Dach (M. Dach Consulting GmbH)
MATLAB is a powerful data processing and modeling tool, which is used in research organizations worldwide. As a rule, MATLAB is deployed as an integrated part of a well-developed computing and research data management environment. In particular, scientists and engineers at the Paul Scherrer Institute (PSI) in Switzerland traditionally implement their numerous MATLAB applications for dealing with experimental data provided by the EPICS based control system. The data are delivered by EPICS Channel Access (CA) servers working on process control computers (IOC), which are distributed around the institutional computer network and communicate directly with a wide variety of measurement sensors and actuators. The applications acting as CA clients are executed on PC consoles running Linux or Windows OS. The main advantage of this approach is that MATLAB applications can easily be modified on demand, which is very important for developments and prototyping. However, the data flows between CA clients and servers experience unwanted time delays introduced by the nature of computer networks. As a result, the whole class of applications, which require fast and reliable time response (e.g. closed loop control tasks where the response time is a key issue) is not covered by this traditional approach. The paper presents an alternative way of the MATLAB code deployment that is based on applications directly bound to EPICS CA servers. Following this way, the performance and reliability of the entire system can be significantly improved, which makes it adequate for time response critical tasks.