Analogue Intelligent chip for short and middle range Radar signal processing – AIR

The AIR project targets at a reconfigurable and programmable integrated circuit based on analogue neural network for analysis of the short and middle range radar (24GHz/60GHz) system for new applications in remote and privacy-preserving monitoring of vital signs (heartbeat, respiration). The project will also develop and publicly release the evaluation kit (scenarios, metrics, protocols, datasets) to objectively evaluate the performance of the developed analogue and digital AI systems featuring movement detection methods. The project constitutes a holistic approach to assessment of analogue computing in a novel application of artificial intelligence (AI) to radar based monitoring, which includes:
1) development of AI algorithm suitable for analogue hardware implementation as well as its software counterparts optimized to run on various digital platforms (CPU, GPU, FPGA) for comprehensive comparison;
2) design of an analogue neural network-based circuit to efficiently perform the analysis and classification of the radar signal, and its implementation in CMOS chip;
3) evaluation and benchmarking of the chip performance versus the software/digital realizations of the algorithm.

The project will be conducted by a consortium of three partners with complementary expertise. The team at the Institute of Applied Computer Science of the Lodz University of Technology, Poland has an extensive experience in development of AI and machine learning algorithms, and their applications to signal processing and classification in real-life scenarios. The research group at VTT Technical Research Centre of Finland has experience in development of millimetre wave radar systems and its applications to vital signs monitoring, as well as in development of ultra-low power circuits for bio-inspired signal processing and their implementations with both standard CMOS and emerging nanotechnologies. The team at Laboratoire national de métrologie et d’essais has an unique and unparalleled experience in evaluation of AI-based ICT systems and specializes in the design of experimental plans and evaluation metrics, and in the specification and qualification of reference data.