Applications in Electronics Pervading Industry, Environment and Society

Table of Contents

1. Short Contributions

   1. Frontmatter

   2. Soluble Mandrel Technology to Produce Parts in Composite Material for Formula 1

      Jacopo Agnelli, David Benedetti, Nicholas Fantuzzi

      Abstract

      The composite production of parts where it is necessary to create hollow structures, such as -for example- air ducts and passages for wiring, does not pass through the traditional model-mold-piece, or direct mold-piece approach, mainly due to the complexity of the shapes to be made and therefore of the equipment to be engineered and produced. On the contrary, thanks to the modern 3D printing technology, or fast prototyping, it is possible to quickly generate the core of the object to be made, on which to proceed with the lamination.

   3. A Reconfigurable 2D-Convolution Accelerator for DNNs Quantized with Mixed-Precision

      Luca Urbinati, Mario R. Casu

      Abstract

      Mixed-precision uses in each layer of a Deep Neural Network the minimum bit-width that preserves accuracy. In this context, our new Reconfigurable 2D-Convolution Module (RCM) computes N = 1, 2 or 4 Multiply-and-Accumulate operations in parallel with configurable precision from 1 to 16/N bits. Our design-space exploration via high-level synthesis obtains the best points in the latency vs area space, varying the size of the tensor tile handled by our RCM and its parallelism. A comparison with a non-configurable module on a 28-nm technology shows many reconfigurable Pareto points for low bit-width configurations, making our RCM a promising mixed-precision accelerator for inference.

   4. Diagnostic Analytics for Pixelated Particle Detectors: A Case Study

      Werner Florian Samayoa, Bruno Valinoti, Romina Molina, Luis G. García, Maria Liz Crespo, Sergio Carrato, Andres Cicuttin, Stefano Levorato

      Abstract

      We present a method for diagnostics analysis for pixelated particle detectors. The method is based on extracting information from the detector in the form of model parameters by using a representative mathematical model. To illustrate the procedure we analyzed real experimental data obtained with the electromagnetic calorimeter ECAL2 of the COMPASS experiment at CERN. Having observed the data, the typical pulses were fitted with a mathematical model. Heat maps were drawn to visualize the distribution of the mean values of each of the fitted parameters. This data visualization technique is useful for highlighting areas with similar behavior and detecting abnormal responses in single cells.

   5. Developing a Toolchain for Synthetic Driving Scenario Datasets

      Marianna Cossu, Riccardo Berta, Alessio Capello, Alessandro De Gloria, Luca Lazzaroni, Francesco Bellotti

      Abstract

      Formalization of driving scenarios is key to define the operational design domain (ODD) of Automated Driving Functions (ADF). Training machine learning (ML) requires huge datasets, that are costly to produce. We propose a toolchain to generate driving scenario video-clip datasets based on the state-of-the-art CarLA driving simulator engine. Scenarios are randomically generated based on a set of parametric features, that are specified by the user. The variability includes both environmental and scenario-specific aspects. As an initial experiment, we have generated a dataset with 200 samples for each one of the 6 implemented classes. The tool is able to achieve a generation rate of about 130 scenarios (7 s. long each) per hour. The tool includes a verification module, which checks the successful completion of each sample.

   6. Ticketing Systems for Smart Public Transportation: Tools at the User Side

      Antoni Martínez-Ballesté, Nicolás Villalobos, Edgar Batista, Pablo López-Aguilar, Agusti Solanas

      Abstract

      Smart transportation systems are an integral part of the smart cities of tomorrow. With the proliferation of miniaturised sensors, IoT devices and 5G communication technologies, plenty of opportunities are yet to be developed to make transport systems more convenient, from the user side, and more cost-efficient and sustainable from the service providers side. Among the many actors involved in this domain, ticketing systems are paramount to access public transportation, such as trains, metros or buses. However, these systems must cope with a number of strong security and privacy requirements. This article overviews the current landscape of tools for a secure deployment of the user side of ticketing systems in public transportation.

   7. Debris Detection and Tracking Through On-Board LiDAR

      Giulio Campiti, Mattia Tagliente, Giuseppe Brunetti, Mario N. Armenise, Caterina Ciminelli

      Abstract

      The uncontrolled growth of space debris around the Earth is forcing satellites to increasingly disrupt their operations in order to prevent potentially catastrophic collisions. Currently, decisions on avoidance maneuvers are made using tracking data mainly obtained through ground-based sensors. As the uncertainties in these data largely affect maneuvering rates, solutions are needed to obtain higher quality observations and therefore decrease the rate of unnecessary maneuvers. This paper studies the possibility of enabling satellites to make autonomous observations of space objects at risk of collision by using onboard LiDAR sensors. As space-based observations do not suffer from diffractions and other problems related to the atmosphere, the proposed solution could be an effective means of obtaining more precise risk estimates. An orbital mechanics analysis of typical conjunction dynamics has been performed to derive the required sensor performance.

   8. Automatic IP Core Generator for FPGA-Based Q-Learning Hardware Accelerators

      Lorenzo Canese, Gian Carlo Cardarilli, Luca Di Nunzio, Rocco Fazzolari, Marco Re, Sergio Spanó

      Abstract

      We introduce a MATLAB-Simulink software able to generate customizable hardware IP cores for the Reinforcement Learning algorithm called Q-Learning. The tool automatically produces the VHDL code and runs both synthesis and implementation for any AMD-Xilinx FPGA using the Vivado software chain. Our automatic generator relies on the “HDL coder” from Mathworks to produce an efficient hardware accelerator based on the state of the art. The model can be customized by the user according to the desired Q-Matrix size and bit-depth for all the algorithm parameters.

   9. Review of Security Vulnerabilities in LoRaWAN

      Junaid Qadir, Ismail Butun, Paolo Gastaldo, Daniele D. Caviglia

      Abstract

      The realm of Low Power Wide Area Network (LPWAN) has a paramount influence on the way we work and live. For instance, real-time applications and rapid packet transiting for long-range have now come into practice that was previously considered mysterious. However, euphoria becomes a problem when it comes to security considerations, as low-power devices possess limited processing units that are unable to elucidate robust security algorithms. In this case, the Low Power Wide Area Network (LoRaWAN) stepped into a technological competition that filled the gap by adopting the end-to-end security feature. Though, LoRaWAN protocol entails fundamental security requirements but the implementation matters. This paper presents security analyses in the LoRaWAN networks. In addition, we provide a bibliometric overview of security considerations in LoRaWAN that helps researchers for thorough insights and implementation.

   10. Experiments on Speeding Up the Recursive Fast Fourier Transform by Using AVX-512 SIMD Instructions

       Giacomo Sansone, Marco Cococcioni

       Abstract

       The Fast Fourier Transform is probably one of the most studied algorithms of all time. New techniques regarding hardware and software are often applied and tested on it, but the interest in FFT is still large because of its applications - signal and image processing, numerical computations, etc. In this paper, we start from a trivial recursive version of the algorithm and we speed it up using AVX-512 Single Instruction Multiple Data (SIMD) instructions on an Intel i7 CPU with native support to AVX-512. In particular, we study the impact of two different storage choices of vector of complex numbers: block interleaving and complex interleaving. Experimental results show that automatic vectorization provides a 10.65% (\(\sim 1.12\times \)) speedup, while with vectorization done by hand the speedup reaches 33.78% (\(\sim 1.51\times \)). We have made our code publicly available, which could be helpful for SIMD instructions teaching purposes.

   11. An Image Processing Algorithm to Optimize the Output Configuration of a Photonic Integrated Circuit

       Luca Gemma, Martino Bernard, Davide Brunelli

       Abstract

       The interest in silicon photonics as a quantum enabling technology is rapidly growing, and Photonic Integrated Chips (PICs) have been proven to be a robust and viable solution in such research fields. As this technology applied to the quantum world is relatively young, some areas of interest remain uninspected, especially the control and output optimization. In this work, we propose an image processing tool to control and optimize a PIC based solely on images captured by a camera and without invasive output detectors. We tested this architecture on a Silicon Oxynitride (SiON) PIC where several Mach-Zehnder interferometers can be voltage driven by Titanium-Titanium Nitride (TiTiN) thermistors. By comparing the results of the image processing algorithm with those retrieved by silicon photodetectors on the same chip, we have proven that our approach can match or even outperform the traditional approach of sensing outputs with silicon photodetectors.

   12. Multi-objective Framework for Training and Hardware Co-optimization in FPGAs

       Mohammad Amir Mansoori, Mario R. Casu

       Abstract

       Although several works have recently addressed the problem of performance co-optimization for hardware and network training for Convolutional Neural Networks, most of them considered either a fixed network or a given hardware architecture. In this work, we propose a new framework for joint optimization of network architecture and hardware configurations based on Bayesian Optimization (BO) on top of High Level Synthesis. The multi-objective nature of this framework allows for the definition of various hardware and network performance goals as well as multiple constraints, and the multi-objective BO allows to easily obtain a set of Pareto points. We evaluate our methodology on a network optimized for an FPGA target and show that the Pareto set obtained by the proposed joint-optimization outperforms other methods based on a separate optimization or random search.

   13. Transiently-Powered Batteryless Device-to-Device Communication Protocol Simulator

       Alessandro Torrisi, Federico Baggio, Davide Brunelli

       Abstract

       Energy harvesting batteryless wireless sensors survive in harsh environments with a minimal energy budget. Indeed, the ambient harvested energy may be sporadic and not constant. Thus, devices operate intermittently experiencing frequent power failures. Communication over these devices must guarantee efficiency and avoid loss of energy. TRAP (TRAnsiently-powered Protocol) can be a solution to avoid packet loss due to receivers’ power failure. TRAP guarantees communication between devices thanks to the node’s energy status awareness. Indeed, effective communication succeeds only if both sides of the communication channel involved in the data packet exchange process have enough energy to complete transmission and reception. TRAP relies on nodes’ energy level information provided through an RF backscatter channel implemented as a self-sustainable and ultra-low-power demanding process. A possible TRAP implementation with RF backscatter has been proven in a laboratory setup for a few devices network. Although results were promising, deep validation and scale-up are necessary. For this purpose, we developed a simulator to verify how the protocol works in different conditions and how performance scales in large sensor networks. The simulation results show that the TRAP protocol reduces waste of energy avoiding failed transmissions due to power failures and increasing communication performances.

   14. Digital Modulation Recognition Method Based on High-Order Cumulant Feature Learning

       Hao Li, Hua Wu, Qinghe Zhen, Yang Liu, Abdussalam Elhanash, Sergio Saponara

       Abstract

       Automatic modulation recognition (AMR) of communication signals is an important research topic in the processing of intercepted signals. In this paper, aiming for the automatic recognition of modulated signals, we propose a feature learning and classification method based on the high-order cumulants, which effectively suppresses Gaussian white noise. Six digital modulation schemes including BPSK, QPSK, 8PSK, 8QAM, 16QAM, and 64QAM can be recognized by comparing the feature with the threshold. During the experiments, we plot the confusion matrix under different conditions. Moreover, it is derived and verified with simulation experiments and actual data acquisition.

   15. Integrated Photonics for NewSpace

       G. Brunetti, N. Saha, G. Campiti, A. di Toma, N. Sasanelli, F. Hassan, M. N. Armenise, C. Ciminelli

       Abstract

       In the last years, an innovative approach in conceiving Space systems has been proposed, known as NewSpace, aiming at developing less expensive satellites in short periods of time, also saving costs, reducing time to access Space and enabling constellation flights. The miniaturization of on-board systems, even preserving reconfigurability and reliability, is required to fulfil the NewSpace goals. These features match with the photonics development trends, increasingly focusing on photonic integrated circuits that show EMI immunity, transparency, low propagation-induced loss, wide bandwidth, and radiation hardness. Here, the recent advances in the field of micro- and nano-photonic devices and systems, with potential applications mainly in the field of satellite technologies, are overviewed, with reference to materials, performance, reliability aspects, and technical bottlenecks, also reporting the development directions and perspectives.

   16. Implementation of Dynamic Acceleration Unit Exchange on a RISC-V Soft-Processor

       Saeid Jamili, Abdallah Cheikh, Antonio Mastrandrea, Marcello Barbirotta, Francesco Menichelli, Marco Angioli, Mauro Olivieri

       Abstract

       Using Artificial Intelligence (AI) techniques has become the best solution in many applications. By the end of Moore's Law, implementing a platform capable of such massive processing for edge-IoT applications has become a significant challenge. However, using static hardware accelerators can be an excellent solution; even so, they typically require a great deal of silicon area and are not optimized for all operation modes. Reconfigurable computing lets parts of the hardware change proportionally to the task during operation, allowing for optimized operation and the use of many hardware accelerators without requiring a large area. In this study, we present a dynamic acceleration unit exchange on a RISC-V soft-processor based on the open-source Klessydra-T13 RISC-V core. We show how reconfiguration can be used to make the hardware accelerator more flexible and improve its performance. As a case study, we show how reconfiguration techniques can be used to speed up AI architectures by reconfiguration of vector accelerator units.

   17. Investigating High-Level Decision Making for Automated Driving

       Alessio Capello, Luca Forneris, Alessandro Pighetti, Francesco Bellotti, Luca Lazzaroni, Marianna Cossu, Alessandro De Gloria, Riccardo Berta

       Abstract

       As the quality of perception systems available for automated driving (AD) increases, we investigate the development of an AD agent based on Reinforcement Learning which exploits underlying systems for longitudinal and lateral control. The goal is addressed by designing high-level actions, trying to imitate the commands of a real driver. The proposed agent is trained in a simulated motorway environment and compared to an agent which outputs low-level actions. Our preliminary results show similar performance results, a more pronounced human-like behaviour and a huge reduction in needed training time because of the higher-level of the available actions.

   18. A Blind Modulation Classification Method Based on Decision Tree and High Order Cumulants

       Yulai He, Hua Wu, Qinghe Zheng, Yang Liu, Abdussalam Elhanashi, Sergio Saponara

       Abstract

       The classification of modulation schemes will be widely used in future communication systems, among which the classification accuracy and speed of modulation classification methods have always been the focus of the research. Our goal is to simplify the complexity of the classification model and improve the modulation recognition accuracy. In this paper, we propose the blind modulation classification method based on the decision tree and high order cumulants. During the experiments, the results are visualized to observe the effect. By comparing the probability of accurate classification at different signal-to-noise ratios at each stage, it is used as an overall measure of the performance of the entire classifier. Moreover, experiments show that the performance of the independent block level and the whole system has excellent accuracy and stability under the optimal threshold, signal-to-noise ratio, and the number of samples.

   19. Design and FPGA Synthesis of BAN Processing Unit for Non-Archimedean Number Crunching

       Federico Rossi, Lorenzo Fiaschi, Marco Cococcioni, Sergio Saponara

       Abstract

       This work presents the design and synthesis of a processing unit for numbers encoded according to the recently introduced BAN format. Such an encoding allows one to represent numbers which are not only finite (as the reals) but also infinitely large or infinitely small, i.e., non-Archimedean. The motivation behind this study is the significant burst the non-Archimedean numerical computations have received in the last 20 years and the applications that have been found. With a hardware support, this operations would significantly increase in speed, enlarging the spectrum of possible applications to industrial and real-time ones.

   20. Prototyping and Preliminary Testing of a Revamped Electric Bus with a Fast Recharge System

       Adriano Alessandrini, Lorenzo Berzi, Fabio Cignini, Tommaso Favilli, Adelmo Niccolai, Fernando Ortenzi, Luca Pugi

       Abstract

       A fast and reliable recharge of the electric public transportation system is a fundamental prerequisite for extended electrification and decarbonization of public transportation systems. As a part of continuous research cooperation, authors have proposed and tested different fast recharge technologies that have been tested on the fleet of electrical Minibus from TecnoBus. In this way, different recharge technologies are compared on identical vehicles with power management systems that are customized and revamped for the testing of a specific technology that must be tested on the vehicle. In this work, the authors focus their attention on the testing of highpower hybrid lithium Titanate cells that can support repeated fast recharge cycles that are performed using a recharge station powered by a flywheel storage system. In this preliminary work, it’s presented the design and integration of the on-board storage system on the benchmark vehicle and its preliminary testing at ENEA Casaccia research center in Rome, Italy.

   21. On the Deployment of Low-Cost Sensors to Enable Context-Aware Smart Classrooms

       Edgar Batista, Oriol Villanova, Joan Rosell-Llompart, F. J. Huera-Huarte, Antoni Martínez-Ballesté, Agusti Solanas

       Abstract

       Smart classrooms augment traditional classrooms with sensors, IoT devices and data networks, and have become a promising arena for context-awareness studies. The collection of contextual data opens the door to numerous applications like environmental sensing, one of the most demanded abilities to face the COVID-19 emergency. To this end, this article presents a flexible and autonomous IoT-based platform for environmental sensing using inexpensive sensors in classrooms. Preliminary results prove the feasibility of the proposed platform and allow us to draw several conclusions on aspects to be considered when deploying these sensors in real settings.

   22. Modulation Recognition Based on BP Neural Network

       Zhiwei Sun, Hua Wu, Qinghe Zheng, Yang Liu, Abdussalam Elhanashi, Sergio Saponara

       Abstract

       In the real wireless communication applications, there are always existing multiple signal modulation schemes. To complete the demodulation of modulated signals, it is necessary to understand the signal modulation method, so it is important for the signal receiver to have the ability to automatically identify the signal modulation schemes. In this paper, for six common digital modulation schemes, each modulation scheme modulates one hundred signals respectively, then extracts five feature parameters from each signal, and finally classifies them using back propagation (BP) neural network. The experimental results illustrate that the recognition accuracy can reach more than 90% when SNR > 10 dB. The model structure is simple and practical, and it can meet the requirements of automatic recognition of modulation schemes.

   23. Towards Efficient Gateways and Servers for Biosensors

       M. D. Grammatikakis, S. Ninidakis, G. Kornaros, D. Bakoyiannis

       Abstract

       Smart E-health biosensors in wearable and mobile devices form an increasing technology trend. In most solutions, patient data is initially transmitted over Bluetooth to a gateway that in turn connects to a remote file server (or cloud). In this work, we focus on enhancing the performance of biosensor data flows across a Bluetooth-to-Ethernet gateway by examining lock-free concurrent queues and traditional lock-based circular FIFOs that embody single-producer single-consumer principles. Our results show that for large biosensor rates, the waiting time of the concurrent queue is smaller. Large rates, above 256 pulses/s, are not supported by the biosensor, but are examined through a self-similar, digital twin process. Finally, by considering a soft real-time analysis and animation application at the server, we explore interesting security vs QoS tradeoffs, leveraging the use of inexpensive crypto ICs.

   24. Improvement of Sodium-Metal Halide Battery Electrical Equivalent Model Including Temperature Dependency

       Gianluca Simonte, Roberto Di Rienzo, Federico Baronti, Roberto Roncella, Roberto Saletti

       Abstract

       Sodium-Metal Halide Batteries are a very promising alternative to the Lithium-ion ones for stationary applications, but their chemical complexity requires an accurate battery model to optimize their use. The electrical equivalent model of the battery is ordinarily used to this aim. The temperature dependency of the model parameters is studied in this work. Three characterization tests are carried-out at 270, 300, and 330 \({}^{\circ }\text {C}\) and analyzed to identify the model parameters. The parameters obtained are then compared with the literature showing that introducing in the model the temperature dependency can improve the accuracy of about six times.

   25. Radio Frequency Drying of Wool Fabrics

       Marco Cocci, Luca Pugi, Enrico Boni, Massimo Delogu, Andrea Rocchetti, Luca Socci, Nicola Andreini

       Abstract

       The drying processes of wool and other textile materials are energyintensive and not very efficient. In this work, the authors investigated an unconventional drying system able to produce a uniform drying preserving the quality of the treated fabric. The proposed process is also very efficient since the proposed radiofrequency technology can optimize the energy transfer to treated fabric minimizing undesired wasted heat fluxes to surrounding air or to other parts of the machine. For the investigation, a simulation model was developed to evaluate tissue heating.

   26. Preliminary Design and Simulation of a Transport and Winding System of an Innovative Radio Frequency Dryer

       Marco Cocci, Massimo Delogu, Lorenzo Berzi, Luca Pugi

       Abstract

       Textile dryers are energy-consuming machines that are widely diffused in the textile industry Authors are currently working on the complete redesign of a sustainable drier which is completely rethought to maximize productivity and sustainability. In this work, the authors focus their attention on modeling and design of winding and transport of the fabric across the machine comparing two different solutions: a conventional transmission and a direct drive version in which every pulley stage is actuated by a different electric motor.

   27. Design and Test of an LSTM-Based Algorithm for Li-Ion Batteries Remaining Useful Life Estimation

       Andrea Begni, Pierpaolo Dini, Sergio Saponara

       Abstract

       The article describes how to extrapolate a useful time-series of features from a raw dataset of complete charging/discharging cycles of Li-ion batteries. The extrapolation of such time-series based on features is helpful to reduce the size of the LSTM (Long Short-Term Memory) as much as possible, differently from classical approaches with LSTM applied to raw data time-series. After a data pre-processing step, this work implements a features-extraction process that allows selecting the best features to describe the performance degradation of the batteries during the time and to estimate the RUL (Remaining Useful Life) during the battery life.

   28. A Clinical Tool for Prognosis and Speech Rehabilitation in Dysarthric Patients: The DESIRE Project

       Massimiliano Donati, Alessio Bechini, Clelia D’Anna, Bruno Fattori, Marco Marini, Martina Olivelli, Susanna Pelagatti, Giulia Ricci, Erika Schirinzi, Gabriele Siciliano, Mirko Tavosanis, Francesca Torri, Nicola Vanello, Luca Fanucci

       Abstract

       Dysarthria is a motor disorder of speech characterized by alteration of articulation and intelligibility of speech. The goal of dysarthria management is to optimize communication effectiveness for as long as possible. To help clinicians in monitoring disease progression and rehabilitation outcomes, the DESIRE tool analyzes several reading sessions in which the patients pronounce predetermined selected words aloud, elaborating a measure of how much the patient’s pronunciation deviates from those of previous sessions and the expected performance. In addition, the electronical record offers a comprehensive view of patient’s status, and the web access allows the care team to remotely monitor progresses, so that they can tailor rehabilitation programs over time. Through the possibility to understand the patient difficulty about specific phonemes, word length, consonant clusters, this innovative tool offers a method to assess and monitoring dysarthria, to address therapeutic strategies, and to provide useful requirements for clinical trials readiness.