PRICOM Printed Computing: Enabling Extremely Low Cost Pervasive Near Sensor Computing

Summary

A large number of important domains - such as fast moving consumer goods and personalized medicine - have still not seen the benefits of computing, mainly due to high production costs of rigid silicon technologies. Printed electronics based on additive manufacturing processes holds promise of meeting cost and conformity needs of such applications. However, the realization of traditional digital processor architectures is infeasible due to constraints of low-cost manufacturing, such as form factor, low device count, large feature sizes, and high variations. The fundamental research question, hence, is how to perform accurate, reliable and energy-efficient classification computing to meet target applications’ requirements within the constraints of additive printed manufacturing.
The aim of PRICOM is to make breakthroughs by developing unconventional mixed-signal classifier computing paradigms together with their hardware realization and mapping based on additive printing technologies. This enables to significantly reduce the hardware footprint, and directly process analog sensory inputs while achieving high classification accuracy. Nevertheless, it is a major challenge as analog computing is very sensitive to variations, and at the same time additive manufacturing is inherently prune to printing variations. I aim at closing this gap by 1) utilizing the inherent tolerance of neuromorphic computing to variations with special hardware primitive design and training algorithms, 2) designing novel variation-aware physical design algorithms, and 3) developing an iterative tuning flow exploiting unique features of additive manufacturing. The feasibility of multi-disciplinary research of PRICOM is underpinned by my unique cross-layer expertise and will be tested by fabrication-based demonstration of printed computing systems. PRICOM can enable proliferation of computing in consumer market and personalized medicine, bringing economical gains and improving quality of life.

Resources

Show all and search (13)

Unfold all

Fold all

More information & hyperlinks

Web resources:	https://cordis.europa.eu/project/id/101052764
Start date:	01-10-2022
End date:	30-09-2027
Total budget - Public funding:	2 499 286,25 Euro - 2 499 286,00 Euro

Cordis data

Original description

A large number of important domains - such as fast moving consumer goods and personalized medicine - have still not seen the benefits of computing, mainly due to high production costs of rigid silicon technologies. Printed electronics based on additive manufacturing processes holds promise of meeting cost and conformity needs of such applications. However, the realization of traditional digital processor architectures is infeasible due to constraints of low-cost manufacturing, such as form factor, low device count, large feature sizes, and high variations. The fundamental research question, hence, is how to perform accurate, reliable and energy-efficient classification computing to meet target applications requirements within the constraints of additive printed manufacturing.
The aim of PRICOM is to make breakthroughs by developing unconventional mixed-signal classifier computing paradigms together with their hardware realization and mapping based on additive printing technologies. This enables to significantly reduce the hardware footprint, and directly process analog sensory inputs while achieving high classification accuracy. Nevertheless, it is a major challenge as analog computing is very sensitive to variations, and at the same time additive manufacturing is inherently prune to printing variations. I aim at closing this gap by 1) utilizing the inherent tolerance of neuromorphic computing to variations with special hardware primitive design and training algorithms, 2) designing novel variation-aware physical design algorithms, and 3) developing an iterative tuning flow exploiting unique features of additive manufacturing. The feasibility of multi-disciplinary research of PRICOM is underpinned by my unique cross-layer expertise and will be tested by fabrication-based demonstration of printed computing systems. PRICOM can enable proliferation of computing in consumer market and personalized medicine, bringing economical gains and improving quality of life.