NEURALSENS
Smart gas and temperature sensors with neural-network-based low-level in-sensor data processing capability
NEURALSENS investigates a hardware-oriented approach to smart sensing in which sensing and low-level data processing are integrated directly within the sensor system.
The project combines hydrogen gas sensing elements, temperature sensors, crossbar-array architectures and Sensory Neural Networks (SeNNs). The core concept is to use the electrical and sensitivity characteristics of individual sensing elements as physically implemented network parameters, enabling part of the signal processing to be performed directly in the sensing hardware.
The project is carried out jointly by the Institute of Electrical Engineering SAS (IEE-SAS), Comenius University in Bratislava (CUB), and the Institute of Informatics SAS (II-SAS).
Project objectives
O1 — H₂ sensing elements
Fabrication and characterization of microscopic TiN/TiO₂/Pt-based hydrogen sensing elements suitable for implementation in crossbar-array architectures.
The target is to obtain diode-like electrical characteristics and to demonstrate that the response to 1000 ppm H₂ can be tuned by at least one order of magnitude through controlled variation of electrode size and geometry.
O2 — Sensory Neural Network design
Development of a hardware algorithm design rule for in-sensor computing SeNNs and its validation using a proof-of-concept smart temperature sensor.
The approach links the characteristics of individual sensing elements with the target response of nodes in a crossbar-array Sensory Neural Network.
O3 — Smart hydrogen sensor prototype
Demonstration of a laboratory prototype of a TiN/TiO₂/Pt-based smart hydrogen sensor implementing a Sensory Neural Network with a crossbar array of at least 32 sensing elements and hardware-based in-sensor processing.
The target prototype is intended to reach TRL 3 and provide multiple outputs corresponding to different H₂ concentration ranges.
Project implementation
The project is organised into three closely connected work packages:
KPB1 — H₂ sensing element
Development and characterization of hydrogen sensing elements suitable for subsequent integration into the SeNN architecture.
KPB2 — SeNN algorithm
Development of the Sensory Neural Network simulation and design methodology, first using temperature-sensor characteristics and subsequently extending the approach to hydrogen sensing data. The framework is intended to be generalized for CBA-compatible resistive sensors.
KPB3 — SeNN laboratory prototype
Integration of the developed sensing elements and SeNN design methodology into a laboratory prototype of a smart hydrogen sensing system.
KPB1 is led by CUB, KPB2 by II-SAS and KPB3 by IEE-SAS.
Funding
This project is supported by the European Union – NextGenerationEU through the Recovery and Resilience Plan for Slovakia under project No. 09I05-03-V02-00058 (NEURALSENS).