FPGA implementations for feed-forward and QKD systems
This project deals with the design and realisation of digital hardware for the evaluation of quantum events. This is necessary in various quantum communication scenarios in order to evaluate and classify photon events, and to react quickly to them.
Specifically, the goal of this project is to design, develop, verify, and test FPGA-based signal processing solutions to digitally decode signals provided by photon detectors, time-stamp them, calibrate them using previously stored values, determine the parameters for the phase modulator and pass them to the digital-to-analogue converters. This is needed to be able to realise modular and expandable quantum-based feed-forward systems in several development stages.
Secondly, an FPGA-based system is to be conceptualised, developed, verified, and tested for a communication scenario that secures communication with the aid of a quantum key distribution (QKD) procedure. This is to realise a classical communication channel with which both time stamps of photon detection events and user data can be transmitted efficiently and securely. The second scenario is partly based on the first and extends it.
Design of FPGA-based partially dynamically reconfigurable systems
The aim of the project is to reduce the amount of sensor data to be transmitted through efficient and sensor-related data (pre)processing and to protect it from unauthorised access. The analysis of the sensor data is to be carried out using AI-based algorithms, e.g., neural networks. These algorithms, supplied by the project partner IMMS, must be adapted for use in an embedded sensor node, optimised and implemented in an energy-efficient and resource-saving manner. In addition, the amount of data can be further reduced by data compression before the transmission. Through these measures, a reduction in the amount of data by several orders of magnitude is achieved. An integrated data encryption is to ensure the privacy of the sensitive data.
This research project focuses on the implementation of AI-based digital data processing for intelligent sensor nodes. The aim is to research solutions that are both energy-efficient and scalable. The focus is on the use of resource-limited, possibly battery-powered systems. In addition, the combination of data from different sensors will be supported. For reasons of efficiency, FPGA-based implementations are favoured.
Ecological Motor Control and Predictive Maintenance with AI
The ECOMAI project offers a basis for Europe to establish a leading role in AI-enhanced electrical motor drive technology – from hardware to applications – through solutions that support the green and digital transitions.
Electric motors are everywhere from laptop fans and dishwashers to industrial machinery, robots, public transport and more. A modern car can alone contain about 40 motors for various functions. But these valuable uses come at a cost. It has been calculated that electric motors account for 40% of worldwide power consumption and 20% of CO2 emissions.
This platform will provide both cost-efficient AI functionality and explore advanced accelerator and approximate computing principles. Furthermore, ECOMAI will deliver an innovative Model-based Design and Automation Framework: a full development toolkit that combines model-based design and an AI compiler for the specialised hardware platform along with a full system modelling and simulation environment. This will make ECOMAI’s technologies easily accessible, particularly for SMEs.