Hardware Platform Research Articles

Training an Ising machine with equilibrium propagation

Ising machines, which are hardware implementations of the Ising model of coupled spins, have been influential in the development of unsupervised learning algorithms at the origins of Artificial Intelligence (AI). However, their application to AI has been limited due to the complexities in matching supervised training methods with Ising machine physics, even though these methods are essential for achieving high accuracy. In this study, we demonstrate an efficient approach to train Ising machines in a supervised way through the Equilibrium Propagation algorithm, achieving comparable results to software-based implementations. We employ the quantum annealing procedure of the D-Wave Ising machine to train a fully-connected neural network on the MNIST dataset. Furthermore, we demonstrate that the machine’s connectivity supports convolution operations, enabling the training of a compact convolutional network with minimal spins per neuron. Our findings establish Ising machines as a promising trainable hardware platform for AI, with the potential to enhance machine learning applications.

Review of Hardware Platforms for Designing of Internet of Things (IoT) system

Microcontrollers are pivotal components in the realm of the Internet of Things (IoT), providing necessary computational power, memory, and connectivity for a wide range of applications. This review explores the key characteristics of microcontrollers that make suitable for IoT, considering their wide use and new emerging models. Study delves into the fundamental features required for IoT applications, such as low power consumption, compact form factors, diverse connectivity options, and flexible I/O (input-output) capabilities. The review covers a spectrum of microcontroller families from leading manufacturers, assessing their performance, energy efficiency, scalability, and ease of integration with various external devices for automation purpose.

Determining the Quality of Measuring the Level of Lower Extremity Joint Movement in Inclusive Physical Education Using Electronic IT Resources

Objectives. The study aimed to investigate the quality characteristics of the tool developed based on electronic IT resources for measuring the level of movements in the joints of the lower limbs of students with disabilities caused by injuries. Material and methods. The experimental study involved 32 first-year students who had sustained lower limb injuries as a result of the war and were in remission. The methods used included analysis, synthesis, systematization, generalization, technical modeling, pedagogical experiments, and mathematical statistics. Results. A means of measuring movements in the joints of the lower limbs has been developed using electronic IT resources. The basis of the tool is a measuring line consisting of a printed circuit board on which is placed a design of capacitive sensors, a switching line, and a signal converter that measures the signal received by the sensors. The measurement results are transmitted to the controller and then to the PC via an interface implemented based on Bluetooth wireless technology. To implement the measuring tool, a controller is used, which has a board built on the synthesis of the Arduino electronic hardware platform and the Raspberry Pi minicomputer. The test results are displayed on the PC monitor screen. The determination of the qualitative characteristics of the test, in the case of recording the results of a tool developed on the basis of electronic IT resources for measuring the level of movements in the joints of the lower limbs of students with disabilities, established that the level of reliability of the test is above 0.90 (“excellent”) validity – from 0, 6 (“high”). Empirical data collected directly from the experiment have a low level of reliability: qualitative indicators of reliability in the range are below 0.70 (“may have limited applicability”), validity – < 0.3 (“low validity”). Conclusions. Developed based on electronic IT resources, the tool for measuring the level of movements in the joints of the lower limbs of students with disabilities due to limb injuries has significant advantages, such as the availability of functions, ease of use and efficiency. Ensuring high efficiency and objectivity of control contributes to performing control operations in real-time. By using assessment tools with a high level of reliability and validity, we ensure the detection of reliable changes in the state of the joints of the lower limbs of students with disabilities, thus eliminating the influence of errors in making managerial decisions in the planning process of their PE.

Reduced partial shading effect and enhancement of performance metrics using a Fibonacci based algorithm

AbstractPartial shading within arrays diminishes power output, induces hotspots, and compromises module integrity, thereby impacting system performance. The presence of bypass diodes further exacerbates these issues by introducing non‐convexities in power curves, leading to additional power losses. To solve this problem, a new reconfiguration technique named Fibonacci Random Number Generator is proposed in this work which minimizes the effects of shading on the panels. The proposed methodology swiftly reduces current discrepancies between PV array rows by reshuffles the panels in an array to disperse the shade better using a mathematical formula resulting in increased power output and smoother power curves during partial shading events. The effectiveness of the proposed method is measured in terms of GMPP, row current calculations, power loss (PL), mismatch losses (ML), execution ratio (ER), fill factor (FF), and capacity factor (CF) for four distinctive shading conditions. Validation of results in software and hardware platforms showcase the applicability of proposed approach in real‐time environments. Results indicate significant average power improvements of 25.49%, 15.47%, and 9.29% compared to existing popular reconfigurations like Skyscraper, Ken‐Ken, and Chaotic baker map. The proposed method stands out as a potent tool for optimizing PV arrays within real‐world systems grappling with partial shading issues.

Superconducting in-memory computing architecture coupling with memristor synapses for binarized neural networks

In-memory computing electronic components offer a promising non-von Neumann strategy to develop energy-efficient and high-speed hardware systems for artificial intelligence (AI). However, the implementation of conventional electronic hardware demands a huge computational and power budget, thereby limiting their wider application. In this work, we propose a novel superconducting in-memory computing architecture by coupling the memristor device. Leveraging the phase transition of the superconductor induced by external applied Joule power, we can modulate the state of the bottom superconductor based on memristor resistive states and applied voltages, enabling the execution of in-memory computing operations. We then successfully implement vector-matrix multiplication of input and output signals within the designed array, facilitating its integration into AI systems. Constructing a binarized neural network with superconductor-memristor arrays achieves a high level of accuracy, approximately 97%, in handwritten number classification. Through an evaluation of power consumption in our proposed architecture, we find a remarkable ∼48 400× advantage in power efficiency compared to typical memristor systems. This marks the inaugural demonstration of a superconducting in-memory computing architecture through memristor coupling, offering a promising hardware platform for various AI systems with superior energy efficiency and computing capacity.

Probing entanglement in a 2D hard-core Bose-Hubbard lattice.

Entanglement and its propagation are central to understanding many physical properties of quantum systems1-3. Notably, within closed quantum many-body systems, entanglement is believed to yield emergent thermodynamic behaviour4-7. However, a universal understanding remains challenging owing to the non-integrability and computational intractability of most large-scale quantum systems. Quantum hardware platforms provide a means to study the formation and scaling of entanglement in interacting many-body systems8-14. Here we use a controllable 4 × 4 array of superconducting qubits to emulate a 2D hard-core Bose-Hubbard (HCBH) lattice. We generate superposition states by simultaneously driving all lattice sites and extract correlation lengths and entanglement entropy across its many-body energy spectrum. We observe volume-law entanglement scaling for states at the centre of the spectrum and a crossover to the onset of area-law scaling near its edges.

Harnessing Deep Learning for Underwater plastic Trash Identification

Abstract—Machine and deep learning (DL) offer significant opportunities for exploring and monitoring oceans and for tackling important problems ranging from litter and oil spill detection to marine biodiversity estimation. Reasonably priced hardware platforms, in the form of autonomous (AUV) and remote operated (ROV) underwater vehicles, are also becoming available, fuelling the growth of data and offering new types of ap- plication areas. This article presents a research vision for DL in the oceans, collating applications and use cases, identifying opportunities, constraints, and open research challenges. We conduct experiments on underwater marine litter detection to demon- strate the benefits DL can bring to underwater envi- ronments. Our results show that integrating DL in underwater explorations can automate and scale-up monitoring, and highlight practical challenges in enabling underwater operations. This project introduces a refined YOLOv8-based algorithm tailored for the en- hanced detection of small-scale underwater debris, to mitigate the prevalent challenges of high miss and false detection rates . The research presents the YOLOv8 algorithm, which optimizes the backbone, neck layer, and C2f module for underwater characteristics and incorporates an effective attention mech- anism. This algorithm improves the accuracy of underwater trash detection while simplifying the computational model. Empirical evidence underscores the superiority of this method over the other conventional network, manifesting in a significant uplift in detection performance. Notably, the proposed method realized a 63% mean average precision (mAP50), a 60% surge in recall (R). Transcending its foundational utility in marine conservation, this methodology harbors potential for subsequent integra- tion into remote sensing ventures. Such an adaptation could substantially enhance the precision of detection models, particularly in the realm of localized surveillance, thereby broad- ening the scope of its applicability and impact.

A unified model and method for forecasting energy consumption in distributed computing systems based on stationary and mobile devices

The subject of research in this article is the forecasting of energy consumption when computing distributed tasks on computer networks built on the basis of server solutions and distributed systems based on personal smartphones. The goal of this study was to create a universal computing energy cost prediction model that can be applied to both traditional and mobile cloud systems. Tasks: conduct an analysis of energy-saving approaches and technologies used to calculate data; consider computer system models and actions with them, namely: model of distributed job, model of distributed computing system, model of distribution strategy; develop a common and uniform dynamic method of forecasting spent energy with a focus on heterogeneous systems; conduct a study of the proposed approach on stationary and mobile devices. The obtained results include. The results of the experimental measurement of the energy consumption of mobile digital systems and stationary ones are presented. The energy efficiency of computing on GPUs of a stationary device based on CUDA technology and GPUs on mobile devices based on Apple Metal technology was determined. Computation during the calculation of 600 frames on a distributed system from mobile devices with failure settings showed a consumption of 15320 joules of energy. Simulation of computing on a distributed system with stationary devices showed a consumption of 52806 joules of energy. This gives us 3,45 times the consumption benefit from computing on mobile devices. Forecasted consumption is also very accurate. Conclusions. The energy consumption assessment model proved to be quite effective. The results of the experiments show that the energy consumption estimation model takes into account the features of the hardware platform where data processing is performed. Computation of data on the GPU of stationary devices loses energy efficiency to a similar implementation on the GPU of Apple Metal from mobile devices. Therefore, the presented results demonstrate the rationality of using mobile graphics processors for energy-efficient information processing.

An ultra energy-efficient hardware platform for neuromorphic computing enabled by 2D-TMD tunnel-FETs

Brain-like energy-efficient computing has remained elusive for neuromorphic (NM) circuits and hardware platform implementations despite decades of research. In this work we reveal the opportunity to significantly improve the energy efficiency of digital neuromorphic hardware by introducing NM circuits employing two-dimensional (2D) transition metal dichalcogenide (TMD) layered channel material-based tunnel-field-effect transistors (TFETs). Our novel leaky-integrate-fire (LIF) based digital NM circuit along with its Hebbian learning circuitry operates at a wide range of supply voltages, frequencies, and activity factors, enabling two orders of magnitude higher energy-efficient computing that is difficult to achieve with conventional material and/or device platforms, specifically the silicon-based 7 nm low-standby-power FinFET technology. Our innovative 2D-TFET based NM circuit paves the way toward brain-like energy-efficient computing that can unleash major transformations in future AI and data analytics platforms.

Qubit Efficient Quantum Algorithms for the Vehicle Routing Problem on Noisy Intermediate‐Scale Quantum Processors

AbstractThe vehicle routing problem with time windows (VRPTW) is a common optimization problem faced within the logistics industry. In this work, the use of a previously‐introduced qubit encoding scheme is explored to reduce the number of qubits, to evaluate the effectiveness of Noisy Intermediate‐Scale Quantum (NISQ) devices when applied to industry relevant optimization problems. A quantum variational approach is applied to a testbed of multiple VRPTW instances ranging from 11 to 3964 routes. These intances are formulated as quadratic unconstrained binary optimization (QUBO) problems based on realistic shipping scenarios. The results are compared with standard binary‐to‐qubit mappings after executing on simulators as well as various quantum hardware platforms, including IBMQ, AWS (Rigetti), and IonQ. These results are benchmarked against the classical solver, Gurobi. The approach can find approximate solutions to the VRPTW comparable to those obtained from quantum algorithms using the full encoding, despite the reduction in qubits required. These results suggest that using the encoding scheme to fit larger problem sizes into fewer qubits is a promising step in using NISQ devices to find approximate solutions for industry‐based optimization problems, although additional resources are still required to eke out the performance from larger problem sizes.

Wind Power Forecasting with Machine Learning Algorithms in Low-Cost Devices

The urgent imperative to mitigate carbon dioxide (CO2) emissions from power generation poses a pressing challenge for contemporary society. In response, there is a critical need to intensify efforts to improve the efficiency of clean energy sources and expand their use, including wind energy. Within this field, it is necessary to address the variability inherent to the wind resource with the application of prediction methodologies that allow production to be managed. At the same time, to extend its use, this clean energy should be made accessible to everyone, including on a small scale, boosting devices that are affordable for individuals, such as Raspberry and other low-cost hardware platforms. This study is designed to evaluate the effectiveness of various machine learning (ML) algorithms, with special emphasis on deep learning models, in accurately forecasting the power output of wind turbines. Specifically, this research deals with convolutional neural networks (CNN), fully connected networks (FC), gated recurrent unit cells (GRU), and transformer-based models. However, the main objective of this work is to analyze the feasibility of deploying these architectures on various computing platforms, comparing their performance both on conventional computing systems and on other lower-cost alternatives, such as Raspberry Pi 3, in order to make them more accessible for the management of this energy generation. Through training and a rigorous benchmarking process, considering accuracy, real-time performance, and energy consumption, this study identifies the optimal technique to accurately model such real-time series data related to wind energy production, and evaluates the hardware implementation of the studied models. Importantly, our findings demonstrate that effective wind power forecasting can be achieved on low-cost hardware platforms, highlighting the potential for widespread adoption and the personal management of wind power generation, thus representing a fundamental step towards the democratization of clean energy technologies.

Enhancing spectrum utilization and primary user detection in cognitive radio networks through greedy cooperative spectrum sensing

The utilization of the spectrum is optimized through which primary users of modern wireless communication technologies might obtain a higher chance of detection. The research aims to study how the NI-USRP hardware platform can be used to set up greedy cooperative spectrum sensing for cognitive radio networks. Research primarily deals with energy detection and eigenvalue-based detection approaches, both of which are highly recognized for their capacity to sense the spectrum without having prior knowledge of the primary user signals. In the hardware arrangement, there is one transmitter and two cognitive radio receivers. LABVIEW makes it simple to deploy and maximizes the detection probability across a large sample. Here, it was demonstrated that cooperative spectrum sensing is superior to non-cooperative spectrum sensing, which results in a reduction in the risk of errors occurring during detection. The research discovered that the OR combination rule has a higher detection probability than the AND rule at the same time. The research emphasizes the significance of expanding cooperative spectrum sensing to improve overall detection capabilities. SNRs that are more than 10 dB allow the energy detector to operate, and the eigenvalue detector continues to work when the SNR drops to –9 dB.

Energy efficiency and performance analysis of a legacy atomic scale materials modeling simulator (VASP)

This work tackles the performance and energy consumption analysis of a legacy scientific application, the VASP (Vienna Ab-initio Simulation Package), an application commonly used by physicists and chemists for modeling materials at the atomic scale. Many of these scientific applications have been implemented in Fortran, where energy metrics instrumentation is not straightforward. We obtained performance figures (execution time and energy consumption) by instrumenting the source code using EML. This energy measurement library has been modified to introduce Fortran interfaces for these metrics. The analysis was carried out using different matrix algebra libraries, parallelization techniques, and hardware platforms, emphasizing on the MPI, OpenMP, and CUDA parallel implementations of the algorithms used in VASP. We employ various material specifications (atomic structures) and molecular sizes of a silicon-based crystal to create a set of benchmarks for these specifications, leading to some recommendations for final users regarding performance improvements. The proposed benchmarking technique assists the user in selecting the right combination of problem size, compilers, and parallelization options available in VASP. For a given system platform, the user will be able to determine not only the architecture to use (GPU or multicore processors), but also the appropriate library and parallelization according to the atomic structure and molecular size.

Automatic Self Driving Car

This project deals with building an autonomous car that can travel safely and intelligently avoiding the risk of human errors. This raspberry Pi based project can detect the obstacles & traffic light. It can compare the data processed with the data provided to it and is able to take an intelligent decision whether to stop or continue its present path. Important components involved in this project are - the hardware platform which includes raspberry pi board, all the hardware like pi camera and the ultrasonic sensor for improved efficiency & the camera used along with an ultrasonic sensor to provide necessary data from the world for real time processing and application. Second being the cloud platform which will be basically used to train our raspberry pi board for real time applications. Cloud helps us to test as well as train better tracking and decision models & helps in providing the offline computing and storage capabilities for vehicle. Basically, it will be used to train the processor to differentiate between positive (green signal) and negative (red signal) images using various thousands of such signal images as an example. The third and most important part includes the algorithms for perception, control, localization, and recognition. Key Words: Raspberry pi, L293D Driver, Machine Learning, Open CV, Ultrasonic Sensor & Pi Camera.

A shallow 2D-CNN network for crack detection in concrete structures

PurposeThis study aims to obtain methods to identify and find the place of damage, which is one of the topics that has always been discussed in structural engineering. The cost of repairing and rehabilitating massive bridges and buildings is very high, highlighting the need to monitor the structures continuously. One way to track the structure's health is to check the cracks in the concrete. Meanwhile, the current methods of concrete crack detection have complex and heavy calculations.Design/methodology/approachThis paper presents a new lightweight architecture based on deep learning for crack classification in concrete structures. The proposed architecture was identified and classified in less time and with higher accuracy than other traditional and valid architectures in crack detection. This paper used a standard dataset to detect two-class and multi-class cracks.FindingsResults show that two images were recognized with 99.53% accuracy based on the proposed method, and multi-class images were classified with 91% accuracy. The low execution time of the proposed architecture compared to other valid architectures in deep learning on the same hardware platform. The use of Adam's optimizer in this research had better performance than other optimizers.Originality/valueThis paper presents a framework based on a lightweight convolutional neural network for nondestructive monitoring of structural health to optimize the calculation costs and reduce execution time in processing.

Energy-aware fault-tolerant scheduling for imprecise mixed-criticality systems with semi-clairvoyance

Integrating tasks with different functions and criticality levels onto the same hardware platform forms mixed-criticality systems (MCSs). Semi-clairvoyant scheduling, in which the mode transition is revealed at the arrival of high-criticality tasks, has attracted researchers’ attention. Most studies focus on the schedulability test of MCS and do not consider energy consumption. In this paper, we simultaneously consider fault tolerance, energy management, and semi-clairvoyant scheduling in MCS. Firstly, we propose a schedulability test based on compensating adaptive mixed-criticality called CAMC-sem-cp. Secondly, we propose an energy-efficient scheduling algorithm based on the CAMC-sem-cp, called EE-CAMC-sem-cp. Finally, we conduct experiments to evaluate the performance of the proposed algorithms. The experimental results show that EE-CAMC-sem-cp can save 47.84% of the energy consumption compared to CAMC-sem-cp.

Complexity enhancement and grid basin of attraction in a locally active memristor-based multi-cavity map

The complexity of memristive chaotic systems determines whether it is suitable for applications in different subjects. To enhance complexity both in performance and diversity, this article first proposes a discrete model of locally active memristor (LAM) and conceives a three-dimensional memristive multi-cavity map by coupling LAM with the existing sine and cosine modulation (SCM) map. This map is named LAM-SCM map and has numerous independent fixed points with different stabilities. Numerical simulations reveal the memristive parameters-relied lossless displacement and self-shift of multi-cavity attractors, indicating the dynamical effects of LAM on the existing SCM map. The initial-relied dynamics distributions are disclosed by grid basins of attraction, showing the emergence of initial-boosting coexistence. Besides, the performance comparisons verify the superiority of LAM-SCM map over the existing SCM map. Finally, a hardware platform has been developed on FPGA to implement the LAM-SCM map and the captured attractors validate the numerical results. On this basis, we devise a 32-bit pseudorandom number generator (PRNG) based on chaos and implement it on the FPGA-based hardware platform to obtain high-speed and reconfigurable pseudorandom numbers. The results show that the proposed map has enhanced chaos complexity and rich dynamics diversity, which ensures the availability of hardware PRNG.

A Fuzzy-PI Clock Servo with Window Filter for Compensating Queue-Induced Delay Asymmetry in IEEE 1588 Networks.

Clock synchronization is one of the popular research topics in Distributed Measurement and Control Systems (DMCSs). In most industrial fields, such as Smart Grid and Flight Test, the highest requirement for synchronization accuracy is 1 μs. IEEE 1588 Precision Time Protocol-2008 (PTPv2) can theoretically achieve sub-microsecond accuracy, but it relies on the assumption that the forward and backward delays of PTP packets are symmetrical. In practice, PTP packets will experience random queue delays in switches, making the above assumption challenging to satisfy and causing poor synchronization accuracy. Although using switches supporting the Transparent Clock (TC) can improve synchronization accuracy, these dedicated switches are generally expensive. This paper designs a PTP clock servo for compensating Queue-Induced Delay Asymmetry (QIDA), which can be implemented based on ordinary switches. Its main algorithm comprises a minimum window filter with drift compensation and a fuzzy proportional-integral (PI) controller. We construct a low-cost hardware platform (the cost of each node is within USD 10) to test the performance of the clock servo. In a 100 Mbps network with background (BG) traffic of less than 70 Mbps, the maximum absolute time error (max |TE|) does not exceed 0.35 μs, and the convergence time is about half a minute. The accuracy is improved hundreds of times compared with other existing clock servos.

Autonomous mobile robots for outdoor tasks

This paper presents a low-cost hardware platform for outdoor robots, being suitable for education, industrial prototyping and private use. The choice of components isdiscussed, including platform, sensors and controller as well as GPS and image processing hardware. Furthermore, a software approach is proposed, allowing students and researchers to easily implement own algorithms for localization, navigation and the tasks to fulfill. Several robots can be integrated in a framework which connects various different hardware platforms, called the BOSPORUS network. Together with other components, they form an intelligent network for gathering sensor and image data, sensor data fusion, navigation and control of mobile platforms. The architecture of a reference platform on the campus of the Brandenburg University of Applied Sciences is presented and evaluated.

A Mixed-Criticality Traffic Scheduler with Mitigating Congestion for CAN-to-TSN Gateway

The network architecture that Time-Sensitive Networking (TSN) is used as the backbone network and the Controller Area Network (CAN) serves as the intra-domain network is considered as the CAN-TSN interconnection network architecture, which has gained considerable attention within industrial embedded networks, such as spacecraft, intelligent automobiles, and factory automation. The architecture employs the CAN-TSN gateway as a central hub for transmitting and managing a significant volume of communications between the CAN domains and TSN. However, the CAN-TSN gateway faces a high congestion challenge due to the rapid growth in data volume, making it difficult to effectively support different time planning mechanisms provided by TSN. In this paper, we propose a two-stage mixed-criticality traffic scheduler. The scheduler in the first stage adopts a Message Optimization Algorithm (MOA) to aggregate multiple CAN messages into a single TSN message (including the aggregation of critical and non-critical CAN messages), which reduces the number of CAN messages requiring transmission. In the second stage, the scheduler proposes a Message Scheduling Optimization Algorithm (MSOA) to schedule critical TSN messages. This algorithm reassembles all the critical CAN messages (within the un-schedulable TSN messages) to generate new TSN messages for rescheduling. Experimental results show that our proposed scheduler effectively improves the acceptance ratio of critical and non-critical CAN messages and outperforms the state-of-the-art message scheduling method in terms of acceptance ratio while improving the bandwidth utilization and the number of schedule table entries. We further construct a hardware platform to evaluate the performance of MSOA. The consistency between practical results and theoretical results shows the effectiveness of MSOA.