To develop miniaturized devices for electrical impedance spectroscopy (EIS), sinusoidal signal generators (SSGs) need to be fully integrated on chip. High spectral purity and wide frequency tunability are required. In this work, a novel approach to SSG design is proposed, leveraging first-order interpolation executed entirely in the analog domain. Compared to standard SSGs based on zero-order interpolation, lower distortion is obtained for the same clock frequency. Experimental testing of a 0.18-$\mu $m CMOS prototype, employing 32 samples for interpolation, yielded average total harmonic distortion (THD) equal to 0.225% across a 100 Hz–300 kHz frequency range, with 130-$\mu $A consumption. Performance is consistent across different amplitude settings and process–voltage–temperature (PVT) conditions. Such findings potentially lay the groundwork for further analog-interpolation-based waveform generation solutions.

Thanks to their simplicity and scalability, R-2R DACs find widespread employment in present-day applications of integrated circuits, including miniaturized sensing devices and other emerging contexts. However, they typically yield limited accuracy due to mismatch errors. To tackle this issue, two dynamic element matching (DEM) techniques, customized for R-2R DACs, are herein proposed, trading off operating frequency with accuracy. Statistical behavioural simulations have been performed to compare the presented approaches, highlighting comparable performances, with reductions of the standard deviation of INL down to a factor of 0.26. Thereby, design guidelines for optimized DEM-compensated R-2R DACs are suggested.

This paper presents the experimental assessment of a recently proposed ultra-low voltage fully-differential inverter-based amplifier. Thanks to the novel common-mode stabilization loop, the amplifier reaches effective differential operations without compromising the output differential performance even at supply voltages as low as 150 mV. Experimental measurements demonstrate the amplifier functionality at supply voltages between 0.15 V and 0.5 V. In particular, a differential dc voltage gain of 22 dB, a gain-bandwidth product of 750 Hz with a capacitive load of 100 pF, and a static current consumption of only 57 nA are measured at Vdd= 0.3 V, with an area consumption of only 800 μm2.

Driven by the ongoing challenge of designing high-accuracy digital-to-analog converters (DACs) at the cost of a relatively small area occupation, optimal combination algorithms (OCAs) recently gained attention within the myriad of possible calibration techniques for DACs. OCAs show appealing properties with respect to traditional approaches such as dynamic element matching (DEM). At start-up or upon request, mismatches affecting DAC elements are measured on-chip, allowing rearrangement in the selection logic of the DAC unit elements. The newly found arrangement is, hence, used during normal operation, achieving superior linearity. As of today, several alternative OCAs have been proposed; however, designers willing to implement OCA-calibrated DACs are faced with unclear tradeoffs and insufficient design guidelines. In this work, we provide a detailed comparison of existing OCAs based on statistical behavioral simulations. Starting from this, we investigate the relationships between OCAs’ performances and circuit-level design aspects. Specifically, OCAs’ effectiveness in improving the static linearity is linked to the number of DAC bits and the accuracy of the auxiliary comparator required by every OCA. Unforeseen trends emerge, and new design considerations are suggested, fostering novel awareness on the subject of high-accuracy DAC designs enabled by OCA-based calibration techniques.

The use of Machine Learning in IoT devices has become the only viable path in today's landscape, where millions of connected devices surround us and increasingly affect our lives. These resource-limited devices interact with the surrounding world via actuators and sensors. Many of these devices use Machine Learning techniques to be able to interpret the world and choose the appropriate action to take. Therefore the purpose of this work is to create a system that allows the application of Machine Learning algorithms directly to the ends of the network, where sensors and actuators reside. The system is designed to rely on the SENSIPLUS smart-sensor as a data acquisition device, and consists of an automatic code generation and compilation system, which through the use of a Toolchain, allows to run artificial intelligence algorithms directly on microcontroller devices.

Water monitoring systems continuously working ensure real–time pollutant detection capabilities according to their sensitivity and specificity. It is necessary to balance such features because, although being able to sense several substances is a desired feature, the reduction of false positives is a primary goal a classification system should have. High false positive makes the system unusable. The current solution enables a 24/7 service with a sampling rate equal to 0.6 Hz. Our goal is to limit false positives to 1 per day, thus achieving 99.99% accuracy at least. In this paper, we add a false positive reduction module to our pre-existent system, aiming to manage false positive boosters as sensor drift and signal oscillations. Obtained results, using a Multi Layer Perceptron classifier, confirm the false positive reduction while keeping high true positive rates.

Indoor air monitoring represents one of the most challenging global aims for the protection of people health and safety. Lots of efforts, either in the academic or industrial field, are addressed to the development and integration of sensing technologies and Artificial Intelligence techniques for the realization of a smart system capable to detect and recognize gases. In this work, we propose a first prototype of an integrated system involving both sensing and Artificial Intelligence technologies, developed as a two layer architecture. The Hardware Layer is the SENSIPLUS® microchip, a smart sensor IoT ready node endowed with on board sensors and implementing novel measuring technique based on current/voltage correlations. The Software Layer is the SENSIPLUS®Deep Machine, a Deep Learning module based on a Long Short-Term Memory neural network, particularly suitable for times series analysis. The paper presents preliminary experiments for the recognition of three distinct gases with respect to air that demonstrates the proposed system effectiveness.