Together with the world-class expertise in Optic technology, Rayfos specializes in advanced signal processing system integration for a variety of measurement applications. In today’s complex computing systems, multidisciplinary cross domain knowledge is required: systems are composed of diverse components of different specifications whose integration in a real time system is not a straightforward task. Furthermore there are no tools able to use unanimously for system set up, measurements, operation control or verification.
At Rayfos we have developed a systematic and methodological procedure in system integration that can be used in various environments and applications, including industrial manufacturing, health monitoring, experimental set up simulation in research etc.
For any system integration process we define the following building sub-systems:
Main computing platform –system host
Physical interfaces capture physical values from the natural environment in which the customer application operates. Sensors, actuators, audio/video capture devices etc. can be devices that act as physical interfaces in the edge of the system. The device interfaces should comply with industry standards in order to make the selection a simple off-the-shelf process. Additionally, for larger devices like cameras we select them based on their configurability and open driver support for usage in measurement and control software packages.
In this process the values captured from the physical interfaces must be collected, transformed and transferred to appropriate modules for further processing. We invoke data acquisition hardware to receive signals from the physical interfaces and perform some basic pre-processing: typically A/D converters transform the data in the digital domain.
Data acquisition hardware is often an off-the-shelf selection, however in our customer’s complex systems requirements, several additional processes are required: from custom set-up, to specialized calibration and electrical signal analysis up to original board design might be needed in order to efficiently operate sensors while preserving that acquisition is implemented in a modular manner so it can be scalable when systems must upgrade to achieve higher performance goals.
System integration’s main challenge is the integrated controller: an advanced controller should perform accurate, reliable real-time control of all peripherals while implementing additional management functions possibly providing data verification, security, error detection etc. In addition the controller must interface with the system host in order to provide seamless coherent interface between the peripherals and the actual processing modules.
Real-time control requires thorough understanding of the building components, the applications attributes and the runtime behavior of the system. At Rayfos we develop full system-level modeling at one level ‘higher’ than conventional component simulation with a commercial tool. In this way we are able to have an exhaustive view of all aspects of the system under control and management.
Our custom developments include appropriate high-speed components such as reprogrammable hardware logic together with carefully designed scheduling, synchronization and flow control software/firmware modules tailored uniquely to each application’s requirements. Rayfos implements controller logic supporting scalable processing architectures with efficient on-the-fly configuration and high-speed real time performance. We can support arbitrary number and type of sensing modules plus variable size, heterogeneous, scalable processing elements. This forms a powerful system prototype ready to be interfaced with the main computing platform-the host. For communication with system host our controllers support standard interfaces (i.e. PCI) typically exploited in PC environments.
Main computing platform-system host
The main platform of our system design is usually a PC running the application software. Implemented there is the application algorithm including but not constrained to biomedical imaging data processing, industrial quality control, scientific research data manipulation etc. The application is also able to initiate, stop, restart any of the controller’s and the sensors’ functionality. Normally our customers’ system requirements feature very high speed data computation. For performance efficiency we provide computing platforms enhanced with dedicated co-processing capabilities, for example invoking GPUs. Our expertise includes also effectively partitioning the computing algorithm and rapidly developing/porting the demanding modules to co-processors. Furthermore we are able to support customers to develop/update their application API in order to exploit the added values features that the new integrated system offers with a focus on user friendly monitoring and control of the application’s algorithm, progress evaluation, online results assessment etc. In this context we work closely with the customer to upgrade their -proprietary- software with all new functionality (libraries, modules) needed in order to effectively work with the new system as originally desired.