How Frequency Injection Locking Can Train Oscillatory Neural Networks to Compute in Phase
A. Todri-Sanial, S. Carapezzi, C. Delacour, M. Abernot, T. Gil, Elisabetta Corti, S.F. Karg, J. Nüñez, M. Jiménez, M.J. Avedillo and B. Linares-Barranco
Journal Paper · IEEE Transactions on Neural Networks and Learning Systems, first online, 2021
IEEE ISSN: 2162-237X
Brain-inspired computing employs devices and architectures that emulate biological functions for more adaptive and energy-efficient systems. Oscillatory neural networks (ONNs) are an alternative approach in emulating biological functions of the human brain and are suitable for solving large and complex associative problems. In this work, we investigate the dynamics of coupled oscillators to implement such ONNs. By harnessing the complex dynamics of coupled oscillatory systems, we forge a novel computation model--information is encoded in the phase of oscillations. Coupled interconnected oscillators can exhibit various behaviors due to the strength of the coupling. In this article, we present a novel method based on subharmonic injection locking (SHIL) for controlling the oscillatory states of coupled oscillators that allow them to lock in frequency with distinct phase differences. Circuit-level simulation results indicate SHIL effectiveness and its applicability to large-scale oscillatory networks for pattern recognition.
A Plethysmography Capacitive Sensor for Real-Time Monitoring of Volume Changes in Acute Heart Failure
E. Rando, P. Perez, S. Fernandez-Scagliusi, F.J. Medrano, G. Huertas and A. Yufera
Journal Paper · IEEE Transactions on Instrumentation and Measurement, vol. 70, article 4005912, 2021
IEEE ISSN: 0018-9456
A small, wearable, low-weight, and low-power-consumption device for plethysmography capacitive sensing is proposed herein. The device allows carrying out real-time monitoring of leg volume changes in patients suffering from heart failure (HF) conditions. The dynamic of fluid overload in patients with acute HF serves as a prognosis marker for this type of severe disease and, consequently, these patients can benefit from a wearable monitoring system to measure their body volume evolution during and after hospitalization. Our approach is based on contactless capacitive wearable structures implemented by two different sensor realizations located in the ankle: 180°-parallel capacitor plates (two modes of operations are compared, with the patient’s body connected to ground and to the average voltage between plates) and planar-parallel capacitor plates whose overlapped surface varies with the volume of the patient’s leg. Both realizations exhibit good sensitivity to leg volume changes. The acquisition of capacitance values is performed via a simple circuit that achieves notable performance in simulated volume analysis. A preliminary pilot clinical prototype is described as well.
Hierarchical Yield-Aware Synthesis Methodology Covering Device-, Circuit-, and System-Level for Radiofrequency ICs
A. Canelas, F. Passos, N. Lourenço, R. Martins, E. Roca, R. Castro-Lopez, N. Horta and F.V. Fernandez
Journal Paper · IEEE Access, vol. 9, pp 124152-124164, 2021
IEEE ISSN: 2169-3536
This paper presents an innovative yield-aware synthesis strategy based on a hierarchical bottom-up methodology that uses a multiobjective evolutionary optimization algorithm to design a complete radiofrequency integrated circuit from the passive component level up to the system level. Within it, performances’ calculation aims for the highest possible accuracy. A surrogate model calculates the performances for the inductive devices, with accuracy comparable to full electromagnetic simulation; and, an electrical simulator calculates circuit- and system-level performances. Yield is calculated using Monte-Carlo (MC) analysis with the foundry-provided models without any model approximation. The computation of the circuit yield throughout the hierarchy is estimated employing parallelism and reducing the number of simulations by performing MC analysis only to a reduced number of candidate solutions, alleviating the computational requirements during the optimization. The yield of the elements not accurately evaluated is assigned using their degree of similitude to the simulated solutions. The result is a novel synthesis methodology that reduces the total optimization time compared to a complete MC yield-aware optimization. Ultimately, the methodology proposed in this work is compared against other methodologies that do not consider yield throughout the system’s complete hierarchy, demonstrating that it is necessary to consider it over the entire hierarchy to achieve robust optimal designs.
A Configurable RO-PUF for Securing Embedded Systems Implemented on Programmable Devices
M.C. Martínez-Rodríguez, E. Camacho-Ruiz, P. Brox and S. Sánchez-Solano
Journal Paper · Electronics, vol. 10, no. 16, article 1957, 2021
MDPI ISSN: 2079-9292
Improving the security of electronic devices that support innovative critical services (digital administrative services, e-health, e-shopping, and on-line banking) is essential to lay the foundations of a secure digital society. Security schemes based on Physical Unclonable Functions (PUFs) take advantage of intrinsic characteristics of the hardware for the online generation of unique digital identifiers and cryptographic keys that allow to ensure the protection of the devices against counterfeiting and to preserve data privacy. This paper tackles the design of a configurable Ring Oscillator (RO) PUF that encompasses several strategies to provide an efficient solution in terms of area, timing response, and performance. RO-PUF implementation on programmable logic devices is conceived to minimize the use of available resources, while operating speed can be optimized by properly selecting the size of the elements used to obtain the PUF response. The work also describes the interface added to the PUF to facilitate its incorporation as hardware Intellectual Property (IP)-modules into embedded systems. The performance of the RO-PUF is proven with an extensive battery of tests, which are executed to analyze the influence of different test strategies on the PUF quality indexes. The configurability of the proposed RO-PUF allows establishing the most suitable ‘cost/performance/security-level’ trade-off for a certain application.
Digital Implementation of Oscillatory Neural Network for Image Recognition Applications
M. Abernot, T. Gil, M. Jiménez, J. Núñez, M.J. Avellido, B. Linares-Barranco, T. Gonos, T. Hardelin and A. Todri-Sanial
Journal Paper · Frontiers in Neuroscience, vol. 15, article 713054, 2021
FRONTIERS ISSN: 1662-453X
Computing paradigm based on von Neuman architectures cannot keep up with the ever-increasing data growth (also called ‘data deluge gap ’). This has resulted in investigating novel computing paradigms and design approaches at all levels from materials to system-level implementations and applications. An alternative computing approach based on artificial neural networks uses oscillators to compute or Oscillatory Neural Networks (ONNs). ONNs can perform computations efficiently and can be used to build a more extensive neuromorphic system. Here, we address a fundamental problem: can we efficiently perform artificial intelligence applications with ONNs? We present a digital ONN implementation to show a proof-of-concept of the ONN approach of ‘computing-in-phase’ for pattern recognition applications. To the best of our knowledge, this is the first attempt to implement an FPGA-based fully-digital ONN. We report ONN accuracy, training, inference, memory capacity, operating frequency, hardware resources based on simulations and implementations of 5 × 3 and 10 × 6 ONNs. We present the digital ONN implementation on FPGA for pattern recognition applications such as performing digits recognition from a camera stream. We discuss practical challenges and future directions in implementing digital ONN.
On- and Off-centre Pathways in a Retino-Geniculate Spiking Neural Network on SpiNNaker
B.S. Bhattacharya and T. Serrano-Gotarredona
Conference · IEEE EMBS Conference on Neural Engineering NER 2021
We present a neural circuit inspired by the brain visual pathway simulated with neuromorphic hardware. The main recipient of the visual information from retinal ganglion cells is the dorsal part of the Lateral Geniculate Nucleus (LGN). Previously, we have implemented a Spiking Neural Network of the LGN on SpiNNaker receiving inputs from an electronic retina (e-retina) chip comprising a Dynamic Vision Sensor. Here, we incorporate a retinotopic structure in the existing LGN network emulating the on- and off-centre receptive fields of the retinal spiking neurons, which are well simulated by the e-retina output. We have parameterised the model to mimic the `push' (excitation) and `pull' (inhibition) dynamics observed in the LGN due to the centre-surround organisation of its receptive fields and synaptic connectivities. The model presented here lay the groundwork for research on building a biologically plausible spiking neural network of visual cognition.
Recording Strategies for High Channel Count, Densely Spaced Microelectrode Arrays
N. Pérez-Prieto and M. Delgado-Restituto
Journal Paper · Frontiers in Neuroscience, vol. 15, article 681085, 2021
FRONTIERS ISSN: 1662-453X
Neuroscience research into how complex brain functions are implemented at an extra-cellular level requires in vivo neural recording interfaces, including microelectrodes and read-out circuitry, with increased observability and spatial resolution. The trend in neural recording interfaces toward employing high-channel-count probes or 2D microelectrodes arrays with densely spaced recording sites for recording large neuronal populations makes it harder to save on resources. The low-noise, low-power requirement specifications of the analog front-end usually requires large silicon occupation, making the problem even more challenging. One common approach to alleviating this consumption area burden relies on time-division multiplexing techniques in which read-out electronics are shared, either partially or totally, between channels while preserving the spatial and temporal resolution of the recordings. In this approach, shared elements have to operate over a shorter time slot per channel and active area is thus traded off against larger operating frequencies and signal bandwidths. As a result, power consumption is only mildly affected, although other performance metrics such as in-band noise or crosstalk may be degraded, particularly if the whole read-out circuit is multiplexed at the analog front-end input. In this article, we review the different implementation alternatives reported for time-division multiplexing neural recording systems, analyze their advantages and drawbacks, and suggest strategies for improving performance.
Neutron-Induced, Single-Event Effects on Neuromorphic Event-Based Vision Sensor: A First Step and Tools to Space Applications
S. Roffe, H. Akolkar, A.D. George, B. Linares-Barranco and R.B. Benosman
Journal Paper · IEEE Access, vol. 9, pp 85748-85763, 2021
IEEE ISSN: 2169-3536
This paper studies the suitability of neuromorphic event-based vision cameras for spaceflight and the effects of neutron radiation on their performance. Neuromorphic event-based vision cameras are novel sensors that implement asynchronous, clockless data acquisition, providing information about the change in illuminance ≥120dB with sub-millisecond temporal precision. These sensors have huge potential for space applications as they provide an extremely sparse representation of visual dynamics while removing redundant information, thereby conforming to low-resource requirements. An event-based sensor was irradiated under wide-spectrum neutrons at Los Alamos Neutron Science Center and its effects were classified. Radiation-induced damage of the sensor under wide-spectrum neutrons was tested, as was the radiative effect on the signal-to-noise ratio of the output at different angles of incidence from the beam source. We found that the sensor had very fast recovery during radiation, showing high correlation of noise event bursts with respect to source macro-pulses. No statistically significant differences were observed between the number of events induced at different angles of incidence but significant differences were found in the spatial structure of noise events at different angles. The results show that event-based cameras are capable of functioning in a space-like, radiative environment with a signal-to-noise ratio of 3.355. They also show that radiation-induced noise does not affect event-level computation. Finally, we introduce the Event-based Radiation-Induced Noise Simulation Environment (Event-RINSE), a simulation environment based on the noise-modelling we conducted and capable of injecting the effects of radiation-induced noise from the collected data to any stream of events in order to ensure that developed code can operate in a radiative environment. To the best of our knowledge, this is the first time such analysis of neutron-induced noise has been performed on a neuromorphic vision sensor, and this study shows the advantage of using such sensors for space applications.
SL-Animals-DVS: event-driven sign language animals dataset
A. Vasudevan, P. Negri, C. di Ielsi, B. Linares-Barranco and T. Serrano-Gotarredona
Journal Paper · Pattern Analysis And Applications, vol. 24, no. 2, 2021
SPRINGER ISSN: 1433-7541
Non-intrusive visual-based applications supporting the communication of people employing sign language for communication are always an open and attractive research field for the human action recognition community. Automatic sign language interpretation is a complex visual recognition task where motion across time distinguishes the sign being performed. In recent years, the development of robust and successful deep-learning techniques has been accompanied by the creation of a large number of databases. The availability of challenging datasets of Sign Language (SL) terms and phrases helps to push the research to develop new algorithms and methods to tackle their automatic recognition. This paper presents ‘SL-Animals-DV’, an event-based action dataset captured by a Dynamic Vision Sensor (DVS). The DVS records non-fluent signers performing a small set of isolated words derived from SL signs of various animals as a continuous spike flow at very low latency. This is especially suited for SL signs which are usually made at very high speeds. We benchmark the recognition performance on this data using three state-of-the-art Spiking Neural Networks (SNN) recognition systems. SNNs are naturally compatible to make use of the temporal information that is provided by the DVS where the information is encoded in the spike times. The dataset has about 1100 samples of 59 subjects performing 19 sign language signs in isolation at different scenarios, providing a challenging evaluation platform for this emerging technology.
Digital Non-Linearity Calibration for ADCs with Redundancy using a new LUT Approach
A. Gines, G. Leger and E. Peralias
Journal Paper · IEEE Transactions on Circuits and Systems I-Regular Papers, vol. 68, no. 8, pp 3197-3210, 2021
IEEE ISSN: 1549-8328
This paper presents a novel Look-up Table (LUT) calibration technique for static non-linearity compensation in analog-to-digital converters (ADCs) with digital redundancy, such as Successive Approximation Register (SAR), Algorithmic, Sub-ranging or Pipeline ADCs. The method compensates the performance limitations of the conventional LUT approach in presence of comparison noise and/or non-monotonicity. In these circumstances, the input-output transfer function of a redundant ADC becomes significantly multivalued - that is, different output codes can be achieved for the same input level at different time instants. This behavior is motivated because from sample to sample, in a design with redundancy, the processing signal path is not unique, causing that the error under calibration becomes time-dependent, something which is not contemplated in the conventional calibration model. To deal with this effect, this work proposes a digital low-cost post-processing of the standardized Integral-Non-linearity (INL), which resolves multivalued situations using a direct access to the internal redundant codes. The method improvements are validated by realistic SAR and Pipeline ADC case studies at behavioral level, and by experimental data from an 11-bit 60Msps Pipeline ADC implemented in a 130nm CMOS process. These experimental results show that the proposed calibration achieves an improvement of approximately 1.6 effective bits at full-scale input amplitude.