The combination of customizable computer hardware and efficient algorithms for processing complex-data is the basis for reconfigurable computer systems that are able to change their structure and their function in response to external and/or internal stimuli. Reconfigurable structures provide the means to develop advanced computer systems that can function, to a large extent, autonomously without human intervention and have the ability to correct data, as well as to adapt and repair themselves. They are distributed, scalable, resilient, predictive and intelligent. They can handle data-intensive requirements, can process complex massive data, and have low-latency in data processing. To be able to do all this they require increased performance and lower power consumption. The scientific background of the Computer Structures and Systems research programme addresses both these issues and is based on advanced algorithm engineering and adaptive computing hardware.
The research Programme is designed to align with European and national roadmaps and strategic papers: the HiPEAC Vision 2017, the ARTEMIS strategic research agenda, and Slovenia's Smart Specialisation Strategy. These documents foresee relevant research and development in areas strongly related to reconfigurable systems: dependability, architectures for data-intensive systems, hardware/software co-design, resource planning and scheduling to allow for energy efficiency, code scalability, adaptive and learning control methodologies, dynamic adaptation to changing contexts, decision and control in uncertain and changing contexts.
The existence of complex massive data in real-life processing means that reconfigurable computer structures require new and innovative approaches to run and manage the processes. As a consequence, such (usually embedded) structures must be customizable and adaptable to changing operational contexts, environments or system characteristics, while ensuring resilience, energy efficiency and recoverability. The interdisciplinary state-of-the-art research challenges combine fields from computer science and mathematics: Reconfigurable optimisation algorithms (to efficiently deal with massive data in dynamic and uncertain environments), based on context-awareness (to decide when to reconfigure), with the support of applied statistical analysis and network topology (to determine how to reconfigure). They are implemented by reconfigurable hardware platforms (based on intrinsically parallelised FPGAs), that ensure self-correction (for structure reliability) and allow for approximate computing (for energy efficiency).