A5084914367- Advanced Memory and Neural Computing
- Error Correcting Code Techniques
- Neural Networks and Reservoir Computing
- Analog and Mixed-Signal Circuit Design
- Neural Networks and Applications
- Advanced Radiotherapy Techniques
- Stochastic Gradient Optimization Techniques
- Integrated Circuits and Semiconductor Failure Analysis
- VLSI and Analog Circuit Testing
- Advancements in Semiconductor Devices and Circuit Design
- CCD and CMOS Imaging Sensors
- Neural dynamics and brain function
- Breast Cancer Treatment Studies
- Soil Moisture and Remote Sensing
- Colorectal Cancer Surgical Treatments
- Semiconductor materials and devices
- Radiation Effects in Electronics
- Advanced X-ray and CT Imaging
- Radiation Therapy and Dosimetry
- Effects of Radiation Exposure
- Digital Filter Design and Implementation
- Geophysics and Gravity Measurements
- Advanced Electrical Measurement Techniques
- Lung Cancer Diagnosis and Treatment
- Planetary Science and Exploration
Universitat de les Illes Balears
2011-2024
Gemini South Observatory
2024
Health Research Institute of the Balearic Islands
2021-2023
Hospital Universitario Son Espases
2013-2020
Fundació Universitat-Empresa de les Illes Balears
2014
Hospital Universitario Fundación Jiménez Díaz
2013
Demos
2002-2003
European Space Research and Technology Centre
1997-2001
This paper presents the passive reflectometry and interferometry system (PARIS) concept how it originated in European Space Agency (ESA), Noordwijk, The Netherlands, 1993 as a novel method to perform mesoscale ocean altimetry. PARIS uses signals of opportunity such from global navigation satellite systems (GNSS), which are reflected off surface Essentially, relative delay between direct received Low Earth Orbit provides information about sea height. describes an original experiment on...
Edge artificial intelligence (AI) is receiving a tremendous amount of interest from the machine learning community due to ever-increasing popularization Internet Things (IoT). Unfortunately, incorporation AI characteristics edge computing devices presents drawbacks being power and area hungry for typical deep techniques such as convolutional neural networks (CNNs). In this work, we propose power-and-area efficient architecture based on exploitation correlation phenomenon in stochastic (SC)...
Abstract The Gemini South telescope is now equipped with a new high-resolution spectrograph called the High-resolution Optical SpecTrograph (GHOST). This instrument provides high-efficiency, spectra covering 347–1060 nm in single exposure of either one or two targets simultaneously, along precision radial velocity spectroscopy utilizing an internal calibration source. It can operate at spectral element resolving power 76,000 56,000, and reach signal-to-noise ratio ∼5 1 hr on V ∼ 20.8 mag...
This paper presents a gamma radiation dosimeter based on floating gate sensor. The sensor is coupled with signal processing circuitry, which furnishes square wave output signal, the frequency of depends total dose. Like any other dosimeter, it exhibits zero bias operation and reprogramming capabilities. has been designed in standard 0.6 m CMOS technology. whole occupies silicon area 450 m250 m. initial sensitivity to dose Hz/rad, temperature supply voltage kHz/°C 0.067 kHz/mV, respectively....
This paper presents an improved version of our previous gamma radiation sensor based on a floating gate MOSFET whose output current is changed by the total ionizing dose. Both versions exhibit zero bias operation and reprogramming capabilities. They have been designed in standard CMOS technology, require little silicon area, low power consumption. Sensitivity to dose -11.4 μA/krad, range over 3.6 krad, lowest detectable lower than 2 rad. The new features much higher linearity supply voltage...
Stochastic Computing (SC) has the potential to dramatically improve important nanoscale circuit metrics, including area and power dissipation, for implementing complex digital computing systems, such as large neural networks, filters, or decoders, among others. This paper reviews state-of-the-art design of SC building blocks covering both arithmetic circuits, multipliers, adders, dividers, finite state machines (FSMs) that are needed numerical integration, accumulation, activation functions...
Nowadays Internet of Things (IoT) and mobile systems use more Machine Learning based solutions, which implies a high computation cost with low energy consumption. This is causing revival interest in unconventional hardware computing methods capable implementing both linear nonlinear functions less overhead than conventional fixed point floating alternatives. Particularly, this work proposes novel Radial Basis Function Neural Network (RBF-NN) implementation on Stochastic Computing (SC),...
Spiking neural networks (SNN) are able to emulate real behavior with high confidence due their bio-inspired nature. Many designs have been proposed for the implementation of SNN in hardware, although realization high-density and biologically-inspired is currently a complex challenge scientific technical interest. In this work, we propose compact digital design high-volume that considers intrinsic stochastic processes present biological neurons enables hardware implementation. The model...
This work aimed to enhance a previous neural network hardware implementation based on an efficient combination of Stochastic Computing (SC) and Morphological Neural Networks (MNN). enhancement focused exploiting the natural ease morphological neurons be pruned in order drastically shrink resources increase compactness our network. That is why we extended original hybrid two-layer classify MNIST problem, much more demanding benchmark with about 160,000 trainable parameters. The 92% weights...
This paper presents a comparison of two gamma radiation sensors intended to be embedded in CMOS integrated circuits. Both are based on current source, whose output depends upon the cumulated dose, followed by current-frequency converter. The differ sensing elements: one uses conventional transistors and other floating gate transistor. Results discussed terms sensitivity supply voltage temperature dependence.
A new digital BIST for OAs embedded in mixed-signal circuits is proposed this paper. During test mode, the transient response of OA under shall be measured order to detect any deviation overshoot with respect fault-free circuit. The a very sensitive parameter and can easily obtained by sampling at particular time. analog stimuli signal generated means current sink made single PMOS transistor. decision block purely digital, only two TDM comparators, flip-flops some logical circuitry just...
An ionizing radiation sensor based on a floating-gate capacitor with output in current mode has been tested as X-ray dose monitor for radiotherapy applications. The had initially designed gamma rays detection space missions proving its usefulness. In this new application the 6 MeV X at constant rate up to 3.6 krad of total dose. measured sensitivity is -11.4 μA/krad. Once corrected ambient temperature variations lowest detectable found be 2 rad. Additionally, and supply voltage measured; we...
In recent years Reservoir Computing has arisen as an emerging machine-learning technique that is highly suitable for time-series processing. Nevertheless, due to the high cost in terms of hardware resources, implementation these systems one single chip complex. this brief, we propose a reservoir computing system with morphological neurons allows us reduce considerably area associated neural synapses. The main consequence using tropical algebra input multipliers are substituted by adders,...
A new and general bandwidth (BW) tuning strategy for biquad OTA-C filters is presented. We have taken benefit of the advantages classical Phase-Locked Loop (PLL) scheme to on-line tune central frequency fo we applied it also BW filters. The proposed design procedure uses only two different sizes OTAs. This methodology has been five band-pass filters, allowing fo-tuning BW-tuning in an independent way. Furthermore, influence OTA non-idealities discussed by including compact expressions all...
In this work we propose a new methodology for neural network hardware implementation based on an efficient combination of Stochastic Computing and Morphological Neural Networks (MNN). The main reasons behind the use morphological neurons instead conventional ones are that activation functions not necessary emerges as natural compact way to implement MNN. proposed design has been verified by implementing classical pattern recognition problems such Fisher's IRIS dataset or handwritten digit...
Deploying modern neural networks on resource-constrained edge devices necessitates a series of optimizations to ready them for production. These typically involve pruning, quantization, and fixed-point conversion compress the model size enhance energy efficiency. While these are generally adequate most devices, there exists potential further improving efficiency by leveraging special-purpose hardware unconventional computing paradigms. In this study, we explore stochastic their impact...
We present a Built-In-Current-Sensor (BICS) based on monitoring the supply current peak of CMOS opamps using oscillation-test-strategy. The BICS takes weighed sample through each opamp branch and monitors value under oscillation. An envelope detector additional digital circuitry is used to provide pass/fail flag. Simulation results demonstrate high defect coverage with very small impact nominal operation.
We present the design and experimental characterization of a magnetic field sensor. This sensor is based on an array Hall resistors uses lateral BJTs as active elements to read effect it. Its output in current mode, instead more frequent voltage mode. Results show that raw sensibility 5mA/T can be achieved, with no measurable hysteresis.