Approximate computing along with low-resolution and low-dimensional signal processing are emerging as key enabling technologies for future energy-efficient sensing and communication systems. However, research efforts in this area are historically fragmented across different implementation levels including hardware design, embedded systems, signal processing, and application domains. The proposed initiative addresses this gap by building a cross-disciplinary 4TU community connecting researchers from Electrical Engineering and Computer Science working on complementary aspects of the problem. The collaboration combines expertise in approximate computing architectures (TU/e), embedded systems (UT-CAES), signal processing (TU Delft), and wireless systems (UT-RS). The proposed activities will strengthen the Dutch ICT research ecosystem by establishing new collaborations across 4TU institutes and creating shared design pipelines for representative benchmark tasks. This proposal aims to establish a cross-disciplinary 4TU research community on energy-efficient signal processing and computing for wireless communication and sensing. A particular emphasis will be placed on developing shared benchmark pipelines and evaluation methodologies that can serve as a common platform for future cross-institutional collaboration.
The project will support:
- organization of a cross-disciplinary 4TU-ESPC workshop bringing together researchers and industrial stakeholders working in approximate computing and energy efficient wireless systems;
- short-term exchange visits between participating groups, including for joint student project definition and supervision, invited lectures in MSc courses;
- dissemination activities including development of shared benchmark pipelines.
Team: Anastasia Lavrenko (UT), Ghayoor Gillani (UT), Geethu Joseph (TU Delft), Chang Meng (TU/e)