DFG grant for Elisabetta Chicca

Prof. Elisabetta Chicca (RUG) and Prof. Regina Dittmann (Forschungszentrum Jülich, Germany) received a DFG grant for the project "MemTDE". The grant of 600kEuro is awarded for a project in the DFG focus programme "Memristive elements for intelligent technical systems". About 300 kEuro of the grant are dedicated for the work of Prof. Chicca in Groningen.

In the era of the Internet of Things (IoT), an increasing amount of sensory data has to be processed. Since most IoT sensors rely on wireless communication, this results in a steady increase of power consumption which has to be circumvented in future sustainable sensory systems. To address this issue, there is a need for intelligent sensors capable of computing data locally, thereby reducing the burden on communication networks and enabling local decision making. In this project, we will empower a recently proposed computational element, namely the Time Difference Encoder (TDE), suitable for low-latency and low-power sensory information processing, with the advantages provided by a hybrid CMOS-memristive implementation.

To this end, two groups with complementary expertise on memristive device development and analog circuit design which have successfully cooperated in the previous phase will jointly develop a hybrid CMOS-memristive sensory system. The replacement of the capacitors in the TDE by memristive devices opens up the possibility to explore a range of time constants (from milliseconds to seconds) which would be prohibitive in a fully CMOS systems. The access to a wide range of time constant allows the seamless integration of sensory modalities characterized by substantially different temporal dynamics (e.g. vision and olfaction).

In the extension of the project, we will consolidate important advances in the physical understanding and engineering of volatile memristors, as well as the development of efficient event-based sensory processing circuits, algorithms and systems. This consolidation will be a solid basis for several innovations. Specifically, on the material and device science side we will adapt the highly reliable CMOS-compatible devices developed in the first period to the requirements of the MemTDE circuit with respect to decay time, current and voltage level as well as device size. We will make use of the knowledge about the physical origin of the decay time gained in the previous period to engineer device stacks with decay times up to the seconds regime. New, powerful behavioural models for the memristive cells, will be used to design and fabricate a hybrid integrated TDE system.

Regarding the computational side, we will advance the state-of-the-art in event-based sensory processing and investigate novel approaches to sensory motor loops in artificial agents. Finally, the embedding of the novel algorithms into the realized integrated systems paves the way to the development of new technologies for edge applications and artificial agents.