Data Processing Unit Research Articles

Numerical and experimental study of thermal stabilization system for satellite electronics with integrated phase-change capacitor

In this paper experimental and numerical investigations were conducted to analyze the application of phase change material (PCM) for temperature stabilization of small satellite electronics. The reference electronic equipment is a data processing unit generating a nominal power of 9.6 W during its operation. To stabilize its temperature, an integrated PCM module was designed, built, and tested experimentally in a thermal–vacuum chamber. The experimental results were then used to validate a numerical model, developed using commercial software Simcenter FloEFD. The results showed good agreement between measured and predicted temperature evolution in critical locations of the unit. The model predictions were further improved by conducting analysis of uncertain input data, model settings, and by inclusion of the effect of natural convection in the PCM. The improved model showed better agreement between the model predictions and experiments. The comparison showed that the maximum difference between the measured and calculated temperature at the most important location (the heater) was 1.6 °C and difference of the peak values was 0.8 °C. Finally, the model was used to compare the constructed module containing the PCM with a system consisting of an aluminum block of the same volume as PCM. The comparison showed that the use of PCM allows for a reduction of the maximum temperature by 6.2 °C while reducing the mass almost 3.5 times. The results demonstrate that the developed numerical model allows for accurate predictions of temperature distributions in the designed device, and can be effectively used to design PCM modules dedicated to in-space electronics, thereby increasing their peak capabilities and lifetime.

Transport properties of interface-type analog memristors

Interface-type, analog memristors have quite a reputation for real-time applications in edge sensorics, edge computing, and neuromorphic computing. The -type conducting BiFeO3 (BFO) is such an interface-type, analog memristor, which is also nonlinear and can therefore not only store, but also process data in the same memristor cell without data transfer between the data-storage unit and the data-processing unit. Here we present a physical memristor model, which describes the hysteretic current-voltage curves of the BFO memristor in the small and large current-voltage range. Extracted internal state variables are reconfigured by the ion drift in the two write branches and are determining the electron transport in the two read branches. Simulation of electronic circuits with the BFO interface-type, analog memristors was not possible so far because previous physical memristor models have not captured the full range of internal state variables. We show quantitative agreement between modeled and experimental current-voltage curves exemplarily of three different BFO memristors in the small and large current-voltage ranges. Extracted dynamic and static internal state variables in the two full write branches and in the two full read branches, respectively, can be used for simulating electronic circuits with BFO memristors, e.g., in edge sensorics, edge computing, and neuromorphic computing. Published by the American Physical Society 2024

In-line rate encrypted links using pre-shared post-quantum keys and DPUs

The development of quantum computers poses a risk to the cryptographic algorithms utilized by current public key infrastructure (PKI). High performance computing (HPC), inter- and intra-datacenter communications, and governmental communications are particularly vulnerable to security attacks by quantum computers. In response, there are on-going efforts on implementing post-quantum cryptography (PQC) despite the considerable overhead involved in processing PQC encrypted data packets in comparison with their classical counterparts. This work aims to help with the transition to quantum resistant cryptographic schemes by means of providing a thorough experimental performance comparison of the four algorithms selected by the National Institute of Standards and Technology (NIST) for standardization when implemented in a data processing unit (DPU). The experiments are conducted at line-rate, demonstrating for the first time operation at almost 100 Gbit/s and hence suitability for high-capacity data center environments.

Real-Time Intrusion Detection in Power Grids Using Deep Learning: Ensuring DPU Data Security

Deep learning technologies have revolutionized the management of energy, energy consumption, and data security within smart grids through non-intrusive load monitoring (NILM). This paper explores the use of deep learning for real-time intrusion detection in power grids with a primary focus on safeguarding the integrity and security of Data Processing Units (DPUs). An evaluation of various machine learning models, including Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), Decision Trees, and Random Forests, is conducted to detect various types of intrusions, including Fault, Injection, Masquerade, Normal, and Replay. Random Forest produced AUC values of 1.00 for all classes and an overall F1-score of 0.99 for all classes. The Decision Tree model also shows robust performance for detecting Fault and Injection intrusions (AUC = 0.98), with an overall F1-score of 0.94. However, the LDA and SVM models do not perform well in detecting Injection intrusions with overall F1-scores of 0.83 and 0.86. Advances in machine learning can be used to improve smart grid security, reliability, and efficiency, according to this study. These findings highlight the potential of advanced machine learning techniques to enhance smart grid reliability and efficiency. Doi: 10.28991/HIJ-2024-05-03-018 Full Text: PDF

Home Air Quality Monitoring System

Abstract: The quality of indoor air is a critical determinant of health and well-being, particularly Given the considerable amount of time individuals invest indoors. Recognizing the pivotal role of air quality, this paper introduces a novel Home Air Quality Monitoring System (HAQMS) designed to provide real-time, accurate assessments of air quality within residential environments. The HAQMS integrates advanced sensors and IoT (Internet of Things) technologies to detect and quantify a wide range of air pollutants, including particulate matter (PM2.5 and PM10), volatile organic compounds (VOCs), carbon dioxide (CO2), carbon monoxide (CO), and ozone (O3).The system architecture is delineated into three primary components: the sensor array for pollutant detection, a data processing unit employing advanced algorithms for real-time data analysis, and a user interface for displaying air quality metrics and providing health recommendations. Utilizing machine learning techniques, the system not only reports currentair quality but also predicts future air quality levels based on historical data and trend analysis. This predictive feature is pivotal for proactive measures in maintaining indoor air quality.

A knowledge graph algorithm enabled deep recommendation system.

Personalized learning resource recommendations may help resolve the difficulties of online education that include learning mazes and information overload. However, existing personalized learning resource recommendation algorithms have shortcomings such as low accuracy and low efficiency. This study proposes a deep recommendation system algorithm based on a knowledge graph (D-KGR) that includes four data processing units. These units are the recommendation unit (RS unit), the knowledge graph feature representation unit (KGE unit), the cross compression unit (CC unit), and the feature extraction unit (FE unit). This model integrates technologies including the knowledge graph, deep learning, neural network, and data mining. It introduces cross compression in the feature learning process of the knowledge graph and predicts user attributes. Multimodal technology is used to optimize the process of project attribute processing; text type attributes, multivalued type attributes, and other type attributes are processed separately to reconstruct the knowledge graph. A convolutional neural network algorithm is introduced in the reconstruction process to optimize the data feature qualities. Experimental analysis was conducted from two aspects of algorithm efficiency and accuracy, and the particle swarm optimization, neural network, and knowledge graph algorithms were compared. Several tests showed that the deep recommendation system algorithm had obvious advantages when the number of learning resources and users exceeded 1,000. It has the ability to integrate systems such as the particle swarm optimization iterative classification, neural network intelligent simulation, and low resource consumption. It can quickly process massive amounts of information data, reduce algorithm complexity and requires less time and had lower costs. Our algorithm also has better efficiency and accuracy.

IoT-Based Rainfall Monitoring System to Detect Rainy Flood for Coastal Areas of Bangladesh

Flood disasters in Bangladesh’s coastal regions have been a major source of worry due to their damaging effects on both the natural world and people. For this natural disaster to be tracked and efficiently controlled, rainfall measurement must be precise and trustworthy. The rain gauge, which has acquired global acceptability, is the most commonly used instrument for measuring rainfall. This study presents the design and experimental evaluation of a cloud-based sensor-based rain gauge for the real determination of total rainfall and rate. An ultrasonic sensor is connected to a round water container, a round rainwater acquisition, a Bluetooth module, and a data processing unit for data collecting and processing in the proposed recommendation sensor-based rain gauge. Improved rainfall evaluation and detection methods have been created using cutting-edge sensor data processing techniques. Throughout a local rainfall event, the performance of the electronic sensor-based rain gauge is assessed and contrasted with that of a hand-made rain gauge. The experimental results show that the suggested system offers an accurate reading of rainfall quantity and rate per unit of time and area, real-time rainfall monitoring, and reliable and consistent analysis. The created method is made to maintain rainy floods in Bangladesh’s coastal regions by storing rainfall data. The system is anticipated to help the region establish an efficient rainy flood monitoring system. Overall, the suggested digital sensor-based rain gauge offers a useful and trustworthy instrument for managing and monitoring rainfall in Bangladesh’s coastal regions in real-time. GUB JOURNAL OF SCIENCE AND ENGINEERING, Vol 9(1), 2022 P 52-57

Periodic Energy Optimization Using IOT and ML

The rapid expansion of Internet of Things (IoT) applications across various sectors generates an enormous volume of continuous time-series data. However, transmitting this massive amount of sensor data from energy constrained IoT nodes poses a significant challenge. The continuous transmission of such data consumes vast amounts of energy.In this work, we present a solution to this problem by predicting the periodic behavior of sensor data through a higher-level view of continuous transmission data from nodes in IoT at server side. Our system is composed of an IoT sensor network and a data processing unit. The local sensor network: temperature and humidity data is collected using 4 different nodes, as well, which afterward this info is transferred into a data processing unit built on the Raspberry Pi device. We use the machine learning model Autoregressive Integrated Moving Average (ARIMA) on the processing unit. This model is then applied individually to the data from each of the four nodes, predicting processed sensor values in the future accurately. In short, after getting highly accurate prediction, then we settle down proper energy saving pattern which reduces the data transmission requirements hence results in energy saving pattern.By utilizing the predictive capabilities of the ARIMA model, we minimize the need for constant transmission of raw sensor data. Instead, we transmit only essential updates or deviations from the predicted values. This approach substantially reduces energy consumption by eliminating the transmission of redundant information. In summary, our project aims to overcome the energy limitations of IoT sensor nodes by leveraging predictive modelling techniques, specifically the ARIMA model. By accurately predicting periodic patterns in sensor data, we can optimize energy usage by transmitting only the necessary information, while still ensuring effective monitoring of temperature and humidity in the IoT network.

Entertainment robot based on IoT and VR interaction for motion training posture monitoring

With the development of artificial intelligence technology, the role of social robots in sports activities has gradually become prominent. Robots can help with intelligent control of sports venues, assist in motion guidance, and engage in emotional communication with users, helping athletes adjust their emotions. This article studies the motion training posture monitoring of entertainment robots based on the interaction between the Internet of Things and VR. A new high-intensity motion attitude monitoring system based on quantum dot photodetectors has been developed to improve monitoring accuracy and stability. Entertainment robots can simulate and analyze motion training postures, and provide relevant technical guidance. The quantum dot photodetectors were integrated into the sensor components, and the collected data was transmitted to the data acquisition and processing unit through signal transmission and processing modules. The collected data was processed and analyzed to achieve real-time monitoring and tracking of the target object’s attitude. Multiple experiments were conducted on the entire system, simulating various high-intensity exercise environments, and evaluating the performance parameters of the system. Through continuous optimization and improvement, the stability and accuracy of the system were ultimately ensured.

Inspection of Semi-Elliptical Defects in a Steel Pipe Using the Metal Magnetic Memory Method

Ferromagnetic pipes are widely used for fluid transportation in various industries. The failure of these ferromagnetic pipes due to surface defects can generate industrial accidents, economic losses, and environmental pollution. Non-destructive testing techniques are required to detect these surface defects. An alternative is the metal magnetic memory (MMM) method, which can be employed to detect surface flaws in ferromagnetic structures. Based on this method, we present an analysis of experimental results of the magnetic field variations around five different surface semi-elliptical defects of an ASTM A36 steel pipe. A measurement system of MMM signals is implemented with a rotatory mechanism, a magnetoresistive sensor, a data processing unit, and a control digital unit. The MMM method does not require expensive equipment or special treatment of the ferromagnetic structures. In order to research a potential relationship between the defect sample size and the measured MMM signals, variable defect dimensions are experimentally considered. According to these results, the shape and magnitude of the normal and tangential MMM signals are altered by the superficial semi-elliptical defects. In particular, the maximum and mean tangential components and the maximum and minimum normal components are related to the defect dimensions. The proposed measurement system can be used to study the behavior of magnetic field variations around surface defects of ferromagnetic pipes. This system can be adapted to measure the position and damage level of small defects on the surface of ferromagnetic pipes.

Energy-Efficient Edge and Cloud Image Classification with Multi-Reservoir Echo State Network and Data Processing Units.

In an era dominated by Internet of Things (IoT) devices, software-as-a-service (SaaS) platforms, and rapid advances in cloud and edge computing, the demand for efficient and lightweight models suitable for resource-constrained devices such as data processing units (DPUs) has surged. Traditional deep learning models, such as convolutional neural networks (CNNs), pose significant computational and memory challenges, limiting their use in resource-constrained environments. Echo State Networks (ESNs), based on reservoir computing principles, offer a promising alternative with reduced computational complexity and shorter training times. This study explores the applicability of ESN-based architectures in image classification and weather forecasting tasks, using benchmarks such as the MNIST, FashionMnist, and CloudCast datasets. Through comprehensive evaluations, the Multi-Reservoir ESN (MRESN) architecture emerges as a standout performer, demonstrating its potential for deployment on DPUs or home stations. In exploiting the dynamic adaptability of MRESN to changing input signals, such as weather forecasts, continuous on-device training becomes feasible, eliminating the need for static pre-trained models. Our results highlight the importance of lightweight models such as MRESN in cloud and edge computing applications where efficiency and sustainability are paramount. This study contributes to the advancement of efficient computing practices by providing novel insights into the performance and versatility of MRESN architectures. By facilitating the adoption of lightweight models in resource-constrained environments, our research provides a viable alternative for improved efficiency and scalability in modern computing paradigms.

Optimal decision-making method for equipment maintenance to enhance the resilience of power digital twin system under extreme disaster

Digital twins and the physical assets of electric power systems face the potential risk of data loss and monitoring failures owing to catastrophic events, causing surveillance and energy loss. This study aims to refine maintenance strategies for the monitoring of an electric power digital twin system post disasters. Initially, the research delineates the physical electric power system along with its digital counterpart and post-disaster restoration processes. Subsequently, it delves into communication and data processing mechanisms, specifically focusing on central data processing (CDP), communication routers (CRs), and phasor measurement units (PMUs), to re-establish an equipment recovery model based on these data transmission methodologies. Furthermore, it introduces a mathematical optimization model designed to enhance the digital twin system’s post-disaster monitoring efficacy by employing the branch-and-bound method for its resolution. The efficacy of the proposed model was corroborated by analyzing an IEEE-14 system. The findings suggest that the proposed branch- and-bound algorithm significantly augments the observational capabilities of a power system with limited resources, thereby bolstering its stability and emergency response mechanisms.

Pollution Based Ventilation Control in Automobiles

Abstract: Growing worries about air pollution have driven new approaches to confined spaces, such autos. The proposed system includes air quality sensors, ventilation control mechanisms, and data processing units inside the car. It continuously monitors the amount of CO2 and other air pollutants. It emits warning sounds and turns on automated window ventilation when thresholds are passed. Furthermore, the system proposes closing windows in high-pollution areas and detects foul odors to put on air fresheners accordingly. Alerts are sent by push notifications, sirens that sound, or dashboard lights. This innovation combines sophisticated ventilation control, real-time monitoring, and alerts to make driving safer while prioritizing both environmental sustainability and passenger safety

VOICE CONTROL CAR USING MACHINE LEARNING

The "Voice-Controlled Car Using Machine Learning" project aims to create an innovative and user- friendly vehicle control system by integrating machine learning technology. This system provides a hands-free driving experience, making it safer and more convenient for drivers. The project involves developing a smart car prototype equipped with various sensors and microcontrollers. Machine learning algorithms are implemented to recognize voice commands and translate them into specific actions for the vehicle, such as accelerating, braking, turning, and even parking. The core components of the system include a speech recognition module, a control interface, and a real-time data processing unit. The project not only demonstrates the potential of machine learning in the automotive industry but also promotes a more accessible and user-centric approach to vehicular control. Keywords: “Speech Recognition”, command interpretation”, audio processing”, Wake word Detection”.

System Design and Implementation for Machine Learning and Internet of Things Based Anomaly Detection in Patient Movement

Emerging trends in healthcare have seen an increased need for improved monitoring, and as a result, machine learning (ML) and the Internet of Things have been integrated to develop smart systems that can provide rapid and accurate health assessments. With an inclusive view on mobility anomaly detection in patients, this paper aims to improve hospital patient safety. Patients wear or place these things, and the IoT sensor network then gathers real-time motion data. The information is then moved to the central server. The system was taught the machine learning part using a large number of normal and abnormal patient movement data. It also shows a way to differentiate between common activities and those that might indicate ill health. It comprises an IoT sensor network, a data processing unit, and an ML anomaly discovery module. There are sensors on the Internet of Things that have a wide range of movement data. Then, these are run through an ML model that captures the spatial-temporal dependencies between different sets based upon more sophisticated techniques like the Naive Bayes Classifier. Where there are abnormalities discovered after analyzing them, an alert is sent out; this will help in quick emergency response by saving lives even before diseases become last-stage acute conditions

Novel Approach for Network Data Exchange Among Safety-Related Control Systems of Nuclear Reactors

Industrially mature digital networking technologies form the backbone of the distributed control architectures of nuclear reactors. Typically, hundreds of control systems responsible for assuring safer and secure plant operations are located in the field and interconnected via wired isolated local area networks, which are also connected to centralized servers and operating consoles located in the main control room. Many of such control systems require data processed by other systems situated far away in the plant, which mandates multiple concurrent reliable TCP sessions to be active or live for such field-level control systems. Multi-socket communication beyond 10 is practically difficult for embedded systems with limited memory and processing capabilities; hence, data accumulation, extraction, and routing are typically done with the help of commercially available high-performance servers and switches. The servers and operating consoles of a plant are generally Linux-based commercial-off-the-shelf (COTS) industrial computers, whose development lifecycle is not accessible for review and in accordance with safety guides; hence, qualifying such components for safety operations is extremely difficult. This paper presents a novel way of independently offloading safety-related data transactions from such COTS components (reducing dependency) with the help of very simple, reliable, and secure hardware, which is referred to as a data exchange and processing unit (DEPU). An approach is proposed to realize distributed control architecture with this proposed hardware (to simplify qualification tasks) along with existing servers and operating consoles without compromising powerful and convenient human-machine interface as well as the data logging capabilities of such components. Hardware and software design perspectives, performance characterization, and qualification testing of the DEPU in accordance with nuclear power plant instrumentation and control system design guidelines are also briefed in this paper.

D 2 Comp: Efficient Offload of LSM-tree Compaction with Data Processing Units on Disaggregated Storage

LSM-based key-value stores suffer from sub-optimal performance due to their slow and heavy background compactions. The compaction brings severe CPU and network overhead on high-speed disaggregated storage. This paper further reveals that data-intensive compression in compaction consumes a significant portion of CPU power. Moreover, the multi-threaded compactions cause substantial CPU contention and network traffic during high-load periods. Based on the above observations, we propose fine-grained dynamical compaction offloading by leveraging the modern Data Processing Unit (DPU) to alleviate the CPU and network overhead. To achieve this, we first customized a file system to enable efficient data access for DPU. We then leverage the Arm cores on the DPU to meet the burst CPU and network requirements to reduce resource contention and data movement. We further employ dedicated hardware-based accelerators on the DPU to speed up the compression in compactions. We integrate our DPU-offloaded compaction with RocksDB and evaluate it with NVIDIA’s latest Bluefield-2 DPU on a real system. The evaluation shows that the DPU is an effective solution to solve the CPU bottleneck and reduce data traffic of compaction. The results show that compaction performance is accelerated by 2.86 to 4.03 times, system write and read throughput is improved by up to 3.2 times and 1.4 times respectively, and host CPU contention and network traffic are effectively reduced compared to the fine-tuned CPU-only baseline.

Design of Reconfigurable Logic Block Based Sequential Circuits Using Look Up Table Logics

Reconfigurable sequential circuits find applications in various digital systems, including communication networks, data processing units, embedded systems, and FPGA-based designs. Their ability to adapt and reconfigure their functionality onthe-fly allows them to accommodate dynamic requirements and optimize the use of hardware resources. Traditional implementations of sequential circuits involve static configurations, where the logic and functionality are fixed during synthesis. While these methods are straightforward to design and implement, they lack adaptability and cannot be modified without redesigning the entire circuit. The proposed method involves the utilization of a dedicated Reconfigurable Logic Block (RLB) within the sequential circuits, allowing for dynamic configuration changes without altering the overall circuit structure. The RLB can be programmed to provide different logic functions using look up tables, multiplexers, enabling the sequential circuit such as counters and shift registers to change its behaviour.

Test and performance of the CMS ECAL barrel data conversion and digital processing ASIC for HL-LHC

In preparation for the high luminosity upgrade of the Large Hadron Collider, a new data conversion and digital processing ASIC, called LiTE-DTU, was designed. Implemented in a 65 nm CMOS technology, the LiTE-DTU features two 12-bit 160 MS s-1 successive approximation register time-interleaved ADCs, a data processing unit and enables efficient data transmission at 1.28 Gbps.The LiTE-DTU has undergone extensive testing in both laboratory and beam tests, demonstrating excellent performance in the pre-production version, showing the production readiness of the design. The results of this comprehensive series of tests will be presented.

Recent advances in smart wearable sensors for continuous human health monitoring

In recent years, the biochemical and biological research areas have shown great interest in a smart wearable sensor because of its increasing prevalence and high potential to monitor human health in a non-invasive manner by continuous screening of biomarkers dispersed throughout the biological analytes, as well as real-time diagnostic tools and time-sensitive information compared to conventional hospital-centered system. These smart wearable sensors offer an innovative option for evaluating and investigating human health by incorporating a portion of recent advances in technology and engineering that can enhance real-time point-of-care-testing capabilities. Smart wearable sensors have emerged progressively with a mixture of multiplexed biosensing, microfluidic sampling, and data acquisition systems incorporated with flexible substrate and bodily attachments for enhanced wearability, portability, and reliability. There is a good chance that smart wearable sensors will be relevant to the early detection and diagnosis of disease management and control. Therefore, pioneering smart wearable sensors into reality seems extremely promising despite possible challenges in this cutting-edge technology for a better future in the healthcare domain. This review presents critical viewpoints on recent developments in wearable sensors in the upcoming smart digital health monitoring in real-time scenarios. In addition, there have been proactive discussions in recent years on materials selection, design optimization, efficient fabrication tools, and data processing units, as well as their continuous monitoring and tracking strategy with system-level integration such as internet-of-things, cyber-physical systems, and machine learning algorithms.