National
Memristívna senzorika pre post-digitálnu elektroniku | |
Memristive sensorics for post-digital electronics | |
Program: | Plán obnovy EÚ |
Project leader: | Dehghan Mohammad |
Annotation: | In-sensor computing is a new paradigm for 21st century electronics inspired by nature. In present-day electronics, all the noisy, unstructured data output by sensors needs to be digitised first for further processing. This may soon become a showstopper given the exponential rise in the amount of sensing devices and data they produce, both in consumer electronics like self-driving vehicles and in the Industrie 4.0 framework. On the other hand, in bio-inspired systems, the sensing and processing are not separate; instead, the sensing nodes directly form synaptic connections in the hardware neural network, where the external stimuli being sensed directly alter the synaptic weight matrix, allowing simple algorithms encoded in the neural network to process the signals into reasonable output in real-time. Under this thesis, the student will learn and understand how to build such a prototypical smart sensing system from scratch, i.e. by depositing, patterning and stacking ultra-thin (~nm) oxide and metal films into the simple sensor and memristor devices and arranging these building blocks into functional sensing neural network matrices on a chip. The expertise acquired will cover nano-fabrication methods with a focus on atomic layer deposition (ALD), material analyses and electrical characterisation techniques, and an understanding of the hardware neural networks based on emerging devices. The thesis will be a part of a wider project, and the student will become a part of our research team. We are looking for creative and dedicated team players, prior experience in related areas is a plus. |
Duration: | 1.9.2023 – 30.6.2026 |