Mission and Vision

The Computer Systems Lab (CSL) is a research‑driven laboratory within the University of Thessaly’s Department of Electrical and Computer Engineering. Our mission is to advance computer systems by blending hardware innovation, software optimisation, and machine‑learning techniques to deliver energy‑efficient, high‑performance, and dependable systems from IoT devices to cloud datacentres. We envision computing platforms that can adapt autonomously to changing workloads and environments, using intelligent resource management to provide seamless performance with minimal energy consumption.

Research Areas

• Reconfigurable computing: We map emerging applications on FPGA‑based platforms. Current work focuses on improving the performance and energy efficiency of real‑time visual SLAM algorithms; this includes designing programming models and runtime systems that automatically partition hardware and software and dynamically allocate resources.

• Approximate computing: CSL investigates how to trade off accuracy for efficiency. The team develops programming models and runtime systems that separate applications into critical and non‑critical parts. We also leverage voltage margins in modern CPUs to lower energy use; ML‑based software predicts and applies the right undervolting at runtime, both at node and datacentre scale.

• Power and performance optimisation: Our researchers develop methodologies across the computing stack—including compiler optimisations, algorithmic transformations, undervolting techniques, and accelerator‑based computing—to improve performance and reduce power consumption.

• Heterogeneous computing: We design architectures and programming models for systems combining CPUs, GPUs, FPGAs, and specialised accelerators. Past projects include reconfigurable MPSoCs, hardware‑software co‑design for heterogeneous platforms, and runtime systems that exploit each platform’s characteristics.

• Distributed computing: CSL’s work ranges from wireless sensor networks to edge–cloud systems. We developed middleware for agent‑based applications that adapt to available sensors and actuators and now explore drone swarms and edge computing to coordinate heterogeneous drones and improve mission autonomy.

Research Impact

CSL’s researchers publish widely in leading venues. Our work explores new approaches for flexible deployment of applications across the cloud–edge continuum, where initial deployment is planned declaratively and runtime placement adapts automatically to the available resources and network conditions. Evaluation of these strategies has shown significant reductions in code size and deployment overheads as well as notable improvements in communication efficiency and latency. Beyond this, we have explored accelerated machine‑learning inference, autonomous resource allocation, and hardware–software co‑design to push the boundaries of performance and energy efficiency. These advances underscore our commitment to practical innovations that quietly but meaningfully improve computing systems.

Open‑Source Software

To support reproducible research and technology transfer, CSL releases open‑source software on GitHub. Our codebase ranges from system‑level prototypes to analysis scripts and hardware designs, providing a practical reference for the broader community. Among these, the MLSysOps framework—an autonomic resource‑management stack for the cloud–edge continuum—stands out as our flagship open‑source project. We continually share other repositories that encapsulate research concepts and programming models, inviting practitioners to explore and extend our ideas while fostering collaboration.

People and Opportunities

Our team comprises faculty members, postdoctoral researchers, PhD candidates, research associates, and undergraduate students, reflecting a diverse mix of experience. We collaborate with universities and industry partners across Europe and actively participate in international networks like HIPEAC. CSL welcomes applications from motivated undergraduate students, PhD candidates, and postdoctoral researchers. We offer a supportive environment where students can work on cutting‑edge projects, co‑author publications, and gain hands‑on experience in both hardware and software systems. Recent achievements include undergraduates being selected for the CERN openlab summer programme, where they worked on cutting‑edge computing projects with industry partners, and the organisation of the ML4ECS workshop at HiPEAC, which brought together researchers and industry experts to discuss machine learning for edge computing systems. Participation in concertation meetings on the cloud–edge–IoT continuum demonstrates our engagement with the European research community.