Suche

edacentrum

Projects, Networks, Applications

HoLoDEC: Automated Design Methods for Highly-Efficient Integrated Sensor Modules in Edge-Computing Applications

In HoLoDEC, innovative methods for the design of new analogue and digital circuit and system components for distributed sensor systems and their connection to edge computing systems are being researched. The methodical approach takes into account consistent hardware-software codesign on several hierarchical levels as well as an efficient partitioning of the sensor and electronic systems in terms of higher performance, reliability, energy efficiency and a sustainable use of resources from the individual circuit component to the application of the entire system. To this end, the partners in the project are researching the following topics in more detail, among others:

Requirements for distributed sensor systems (operating parameters/energy demand/lifetime),
Sensor/edge/cloud partitioning for ultra-low power (ULP),
Component design for digital platform (e.g. clock frequency, supply voltage),
Analogue ULP architectures, design knowledge backup and migration,
Automated circuit and layout design using optimisation, generator and AI approaches,
Machine learning, optimisation, automated model generation for higher levels of abstraction,
Energy-saving (AI) algorithms for sensor data processing and data flow analysis.

To this end, the project will develop and apply a continuous design flow, starting from the specification, through the system and circuit design including layout, to the implementation of all system components in silicon and validation using selected demonstrators. The targeted automation of critical design steps will enable faster iteration loops and thus significantly improve the quality of the design results. Machine learning, artificial intelligence and distributed computing methods are also being researched. The project partners cover the entire value chain from methods for circuit, sensor and system design and production to interfaces for edge computing and a wide range of applications (including automotive, industrial and consumer). The research methods will be validated in the project using five silicon demonstrators from the fields of communication, medical technology, automotive and Industry 4.0.