The research area focuses on the design and implementation of advanced networked systems, addressing both architectural and systems-level challenges in modern datacenter, edge, and cloud environments. Core activities include the development of high-performance packet processing pipelines, the co-design of hardware and software abstractions, and the exploration of novel network stack architectures. Particular attention is devoted to full-stack innovation, where new hardware primitives for fast and programmable packet processing are tightly integrated with operating system mechanisms and runtime support. The group investigates how emerging technologies (such as SmartNICs, FPGA-accelerated network interfaces, and in-network acceleration) can be leveraged to improve performance, efficiency, and programmability across the entire system stack.
A significant research thrust concerns operating system and runtime support for next-generation networking. This includes the design of kernel extensions using eBPF for safe and dynamic programmability, as well as kernel-bypass techniques based on DPDK and AF_XDP to achieve low latency and high throughput. The group develops also advanced compiler techniques and programming abstractions that expose hardware acceleration capabilities to applications while preserving safety, portability, and ease of use. By bridging hardware capabilities with system software and application-level interfaces, the research aims to enable scalable, efficient, and flexible network services suitable for demanding workloads such as distributed systems, AI infrastructures, and high-performance cloud applications.
Methodologically, the research combines system design, prototyping, and experimental evaluation with analytical modeling and simulation. Experimental platforms include FPGA-accelerated NICs donated by AMD/Xilinx and SmartNICs featuring System-on-Chip accelerators from NVIDIA, which allow rapid prototyping and hardware/software co-design. Solutions are evaluated through detailed measurement campaigns and benchmarking under realistic workloads, complemented by simulation studies using established tools such as NS-3. This integrated experimental and analytical approach enables rigorous validation of new architectures and provides practical insights into the performance, scalability, and deployability of next-generation networked systems.
