Search

Due to the increasingly growing demand for user identification on cell phones, PCs, laptops, and so on, face anti-spoofing has risen to significance and is an active research area in academia and industry. The detection of the real face then recognize it present an important challenge regarding the techniques that can be used to spoof any recognition system like masks, printed photos. This paper we present an anti-spoofing face method to solve the real-world scenario that learns the target domain classifier based on samples used for training in a particular source domain. Specifically, with the conventional regression CNN, the Spatial/Channel-wise Attention Modules were introduced. Two modules, namely the Spatial-wise Attention Module and the Channel-wise Attention Module, were used at spatial and channel levels to improve local features and ignore the irrelevant features. Extensive experiments on current collections with benchmarks datasets verifies that the recommended solution will significantly benefit from the two modules and better generalization capability by providing significantly improved results in anti-spoofing.

This study shows the agile management of a project using the Scrum methodology. The main objective is to analyze the use of Scrum for life, specifically in scientific production when writing a Paper. The results obtained recognize that the common practices of the agile approach such as revision and retrospective allow that the writing of the article can be improved throughout its development and that the collaboration between the participants is efficient and effective. Likewise, a consistent and effective alternative is provided to the scientific community for the administration in the development of scientific products.

In this work, capacitive touchpads are presented based on fabrication processes that are inexpensive, lab and solvent-free. Pencil and paper have been used to fabricate the sensors, and these materials are economical, biodegradable, and readily available. A touchpad comprising a combination of five sensors has been developed, and analyzed its working by interface controlling for LED's toggle operation. LCD and android app interface have been used to display the result. The sensor's response has been sensed for a gloved finger touch for sensors. The sensor's performance remained consistent, and the functionality was similar to common buttons and sensors. These sensors are recyclable and can be fabricated with ease. This can contribute to potential developments in smart packaging, medical and healthcare applications, and thin human-machine interfaces.

Drowsiness can be dangerous when performing tasks that require constant attention, such as driving a vehicle. Sleepiness is correlated with a variety of physiological variables, such as eye closing, head movements, pulse rate, eye twitch rate, etc. Also, the yawn can be considered as an accurate indicator of drowsiness and fatigue. Yawning detection is very important for the safety purpose of drivers as it will let the driver know if he/she is getting drowsy. Driving at that moment may not be safe. Several automatic yawning detection techniques have been developed for driver's drowsiness monitoring system. Nevertheless, correctly detecting the yawning of the driver and predicting exhaustion in real-time situations is still a crucial challenge. In this paper, we will be reviewing various existing machine learning approaches for driver's yawning detection. In previous approaches, various classical machine learning algorithms such as viola-Jones, contour activation algorithm and SVM have been used for yawning detection, but these approaches failed to predict yawning in realtime situations. Using Deep learning techniques, we can make a real-time yawn detection system with high accuracy. We find that some precious Deep learning algorithms like CNN, RNN, LSTM, Bi-LSTM can detect the patterns with high accuracy. After the comparison of various algorithms and techniques, we find that with the help of Deep learning algorithms the yawning can be detected in real time with high accuracy.

Human recogntion technologies are gaining significant research attention, where the model can be trained to be more precise to recognize the poses performed by objects. Activity identification is a kind of problem, which needs more research consideration and improvement. There is also an increasing need to detect different poses of objects. To tackle the issue, different sensors like Gyroscope and Accelerometer are required to classify the data in the form of images using machine learning algorithms like SVM and CNN. These approaches help us in implementing real-time applicaions such as health monitoring, tracking sports activity and security. The paper also discusses about its benefits, limitations and prominent approach for human activity recognition.

The power density and total power from the package have been increased significantly. The rapid rate of power density increases from one generation to the next not only happens at the device level but also happens at the localized area on the devices. The objective of this study is to build more understanding of the localized high-power density in silicon blocks with respect to the impact in temperature gradient and reliability of the silicon. Conventionally, individual design block power usage numbers are often reported from circuit simulations without considering physical location of transistors and metal wires. However, in a silicon block design, certain transistors may consume significantly more power than others due to higher load capacitance, current draw, frequency, or toggle rate. Therefore, the power distribution is uneven within the design block even in the same use case. In order to gain a better understanding of power density peak spot inside of a single design block, in this study, the transistor activity is dynamically analyzed with circuit simulation vectors of an actual Versal product use case and the spatial power distribution of a 0.5mm x 0.5mm design block is reported in an granularity of 10um x 10um and turn x turn respectively. The analysis results show that the local heat flux can be several magnitudes higher when the design block is analyzed in more granularity. The local heat flux creates local hot spot that was overlooked. The max hotspot temperature would not be captured by the sensor because this specific hotspot sensor location would not be known at first hand due to uncertainty of the exact location created by the local hot spot. Understanding that there can be temperature sensors located some distance(700um) away from hotspot, reporting out the temperature, the device max temperature can be higher than the system monitor readout. Given this max temperature discrepancy, a correlation for max temperature based on heat flux density, source area, heat sink resistance, and senor location would be defined to estimate the max temperature. When the semiconductor chip has hot spots, it affects reliability of the chip. Most of reliability mechanisms result in shorter lifetime with higher temperature and temperature gradients. When a small portion of product has local temperature increase by whatever reason, long-term circuit performance should be altered due to enhanced or reduced aging effect. The device long term reliability is reassessed based on the new discovered localized hot spot.

Rate-Splitting Multiple Access (RSMA) is a flexible and robust multiple access scheme for downlink multi-antenna wireless networks. RSMA relies on Rate-Splitting (RS) at the transmitter and Successive Interference Cancellation (SIC) at the receivers. In this work, we study the performance of RSMA in the scenarios related with the important core services of New Radio (NR) and 6G, namely, enhanced Ultra-Reliable and Low-Latency (URLLC) and enhanced Mobile Broadband Communications (eMBB). We present the optimal system designs employing RSMA that target short-packet and low-latency communications as well as robust communications with high-throughput under the practical and important setup of imperfect Channel State Information at Transmitter (CSIT) originating from user mobility and latency/delay (between CSI acquisition and data transmission) in the network. We demonstrate via numerical results that RSMA achieves significantly higher performance than Space Division Multiple Access (SDMA) and Non-Orthogonal Multiple Access (NOMA), and is capable of addressing the requirements for enhanced URLLC and eMBB in 6G efficiently.

We consider an edge computing framework for online distributed learning in UAV swarm, which allows swarm to collect data and train machine learning (ML) models over the air. Usually, UAVs are used as sensors to collect and transmit data to ground for further processing. However, the overhead for raw data transmission are becoming unaffordable as the task gets complex and the timeliness gets increasingly rigorous. Therefore, We have proposed a new distributed online learning scheme to allow each UAV collect data locally as well as perform ML model training onboard. We have theoretically proved the convergence of the ML model to be trained, and the performance gap between central training method and proposed scheme is analyzed. Furthermore, we have bridged the sensing and transmission via the definition of importance of each UAV based on its contribution in training, where we extend the widely-used timeliness metric Age of Information (AoI) to Age of Updates (AoU), and thus the data freshness and data value in training process are combined together. The performance of the proposed distributed learning scheme is validated, and the difference between AoI and AoU are compared via simulations.

All generations of renewable energy systems stepped into a huge evolution around the world. Countries with big potential of gaz and petroleum energies are finally concerned by the potential of this sector and are currently developing it. Algeria has a huge potential in fossil fuels, however, the government aim to exploit the renewable energy potential of the country. Focus is now on the solar energy above other energies sources such as wind energy. This paper will explore the potential of wind energy in Algeria and go through the challenge facing it development. Some recommendations aiming to develop and implement the wind energy are provided.

A waste carbon copies paper (WCCP) can be applied counter electrode, it made from carbon. In this work, The WCCP film counter electrode is particularly focused under the condition rate at 1 g of WCCP. A WCCP film counter electrode will be used to coat on the fluoride-doped tin oxide (FTO) glass and the anode of DSC use TiO 2 film coated on FTO glass. The both electrodes were coated on FTO glass by size 2x3 cm2. The DSC with WCCP film counter electrode is investigated by solar simulator at 250 W/m2. The morphological properties of WCCP film counter electrode were confirmed by X-ray diffraction (XRD) and scanning electron microscope (SEM). There has been found that open circuit voltage (V OC ) was 0.568 V. The short circuit current (I SC ) was 145.23 μA. The use WCCP film counter electrode is sufficient to be built and tested cathode electrode for DSCs application.

This paper is review on image fusion for different technique i.e. Non-subsampled Contourlet Transform (NSCT) and Neural Network (NN). NSCT is the pyramid structure and gives the multiscale feature of the image. NSCT used to directional filtering structure and used to filter bank. Neural network is a progression of calculations that tries to perceive fundamental connections in a bunch of information through an interaction that impersonates the manner in which the human mind works. The implemented algorithm will applied to MRI image and with the help of NN to calculated accuracy. The previous algorithm was work of different technique i.e. principal component analysis, discrete wavelet transorm and curvelet transform. The implemented algorithm will implement MATLAB software and calculate mean square error and peak signal to noise ratio.

This paper proposes an intelligent cockpit based on artificial intelligence technologies such as voice recognition ASR, TTS, NLP, and multi-round dialogue. The cockpit monitors the progress of the distribution network virtual dispatcher and analyzes the decision engine, and replays the virtual dispatcher online. The implementation process and thinking progress of the virtual dispatcher are overall monitored, and the online monitoring and safety monitoring of the operating personnel are supported.

In modern, real-time heterogeneous systems, ensuring the predictability of interconnects is becoming increasingly important. Existing interconnects are mainly designed to achieve high throughput, with their micro-architectures usually based on FIFO queues. This FIFO-based design prevents prioritization of transactions based on their importance, leading to difficulties in ensuring transaction predictability, especially in a system with a large number of system components. In this paper, we introduce AXI-InterconnectRT, a real-time AXI interconnect for heterogeneous SoCs, which redefines the micro-architecture of interconnects by enabling random accesses of buffered transactions and organizing transactions using dedicated hardware units. With the new micro-architecture, AXI-InterconnectRT can manage transactions based on their importance, guaranteeing their predictability.

Time synchronization is a critical task in robotic computing such as autonomous driving. In the past few years, as we developed advanced robotic applications, our synchronization system has evolved as well. In this paper, we first introduce the time synchronization problem and explain the challenges of time synchronization, especially in robotic workloads. Summarizing these challenges, we then present a general hardware synchronization system for robotic computing, which delivers high synchronization accuracy while maintaining low energy and resource consumption. The proposed hardware synchronization system is a key building block in our future robotic products.

Real time operating system (RTOS) is an operating system which is designed to execute tasks in a guaranteed time. In this paper, we introduce SylixOS - a powerful and open source RTOS. As an industrial RTOS, SylixOS has been used in aerospace, military and many more embedded devices. We introduce the development path and some advanced features of SylixOS. In addition, SylixOS also supports the multiprocess. We discuss how SMP scheduling algorithms can be applied in SylixOS. Performance evaluations of SylixOS are presented to assess the real-time efficiency of SylixOS.

Improper Input Validation (IIV) is a software vulnerability that occurs when a system does not safely handle input data. Even though IIV is easy to detect and fix, it still commonly happens in practice. In this paper, we study to what extent developers can detect IIV and investigate underlying reasons. This knowledge is essential to better understand how to support developers in creating secure software systems. We conduct an online experiment with 146 participants, of which 105 report at least three years of professional software development experience. Our results show that the existence of a visible attack scenario facilitates the detection of IIV vulnerabilities and that a significant portion of developers who did not find the vulnerability initially could identify it when warned about its existence. Yet, a total of 60 participants could not detect the vulnerability even after the warning. Other factors, such as the frequency with which the participants perform code reviews, influence the detection of IIV. Preprint: https://arxiv.org/abs/2102.06251. Data and materials: https://doi.org/10.5281/zenodo.3996696.

New generation airborne embedded system has deployed Graphical Processing Units (GPUs) to raise processing capability to meet growing computational demands. Applications in the airborne embedded system have strict real-time constraints. Therefore, it is necessary to accurately predict timing behaviors of those applications. Many previous work propose GPU performance models to estimate the execution time of applications. However, most of those models do not consider the impact of co-execution on the GPU performance. In this paper, we propose a workload-aware GPU performance model to predict the execution time of applications executed concurrently on a single GPU. Experimental results illustrate that the proposed model can achieve a 5.1%-11.6% prediction error in a real airborne embedded hardware platform.

Wireless communication base stations have strong requirements in flexibility, performance and real-time responsiveness. Building an application-specific lightweight LibOS on a general OS is a good technique to make a balance among real-time responsiveness, performance and flexibility. Compared to traditional ext-kernel systems, the fast progress of virtualization technology brings a more efficient implementation of LibOS. In this paper, we propose a virtualization assisted real-time LibOS system named Tenma for wireless Baseband Unit (BBU). In Tenma, lightweight process virtualization, enhanced isolation between host OS and guest OS, interrupt direct-pass and application specific lightweight scheduling are used to guarantee the requirements of performance and real-time responsiveness while keeping system flexibility. The Tenma system is already in commercial deployment. Experiments in practice prove that it can meet the requirements of wireless base station applications.

Internet-of-Things (IoT) devices can only afford very small batteries due to the size, weight, power and cost constraints. On the other hand, long battery life is expected for such devices, either for improving user experience or due to charging limitations. Therefore, it is critical to reduce the energy consumption of IoT devices. A good opportunity is to save energy on the system level, by letting the system to sleep occasionally. In this paper, we introduce Image Partitioning and Incremental Loading (IPIL), an operating-system-level sleep mechanism used in Huawei LiteOS to reduce the energy consumption of an IoT device. IPIL powers off both the CPU and the main memory during sleep to save as much energy as possible. Thus, system data have to be reloaded after the system wakes up, which may take a long time and thus reduce real-time responsiveness. To solve this problem, IPIL partitions the system image into clusters and incrementally loads the clusters on-demand, reducing the amount of data loading and time overhead in the wake-up step. Experimental results show that IPIL outperforms two widely adopted sleep approaches in terms of energy consumption. At the same time, IPIL offers satisfactory responsiveness with very short wake-up delay. We also present a case study on smart watch to demonstrate how battery life, as a core value of wearable devices, can be improved with IPIL provided by LiteOS.

Correctly specifying system requirement verification strategies plays an important role in Real-time embedded and safety-critical systems development. Traditionally, industries are using document-based/textual artefacts to capture system verification strategies in a way that is highly reliant on efficient communications between system requirement engineers and system verification engineers during systems development iterations. This paper describes a Model-Based System Verification Strategy specification framework and Tools used in Rolls-Royce. The contribution of this framework is 1. improving information reusability; 2. reduced risk of system verification and requirement inconsistency and redundancy 3. improved correlation between system requirements and verification artefacts by sharing the same system models 4. automated support for transformation from document-based system verification to model-based verification.

Constantly increasing botnets powered by vulnerable IoT devices perform record-breaking DDoS attacks to critical infrastructures. Therefore, it is imperative to find vulnerabilities in IoT devices ahead of attackers. In this paper, we first present a systematic analysis on various kinds of IoT malware to further explore the challenges of IoT Honeypot design. We then propose HoneyIoT, a scalable IoT Honeypot framework, which aims at attracting IoT attacks and recording malicious behaviors with configurable vulnerabilities and firmware support. During a 7day real-world industrial experiment, HoneyIoT observed over 12,500 malicious connections and 3,423 distinct login attempts.

We proposed a deep reinforcement learning (DRL) based approach to reconfigure the multicast sessions in an elastic optical network (EON) adaptively. Simulation results demonstrate that our proposal maintains the optimality of light-trees with less reconfigurations and reduces blocking probability.

In order to automate the task of correcting handwritten subjective answers, we envision a system that reads such answers and grades them. In this project, we have created a mobile application to achieve this task. We have used the RAKE algorithm to extract key-phrases, which are then embedded using Word2vec model and compared using Word Mover's Distance. Our created model achieved almost human like evaluation accuracy. We have also provided accessibility features to the users with the aim of making this application a one-stop solution to automatic evaluation of handwritten subjective answers.

Wireless communications via reconfigurable intelligent surfaces (RIS) are envisioned to offer unprecedented spectral efficiency gains by smartly inducing phase shifts on the impinging electromagnetic waves at the RIS to reconfigure the underlying propagation environment. This work investigates a multi-user downlink communication system in which the access point, equipped with multiple antennas, serves its associated single-antenna users equipments (UE) with the help of RIS. The proposal is a leakage-based design in which the transmit precoding, the power allocation, and the phase shifts of the RIS, are designed to maximize the minimum signal-to-Ieakage-plus-noise power ratio (SLNR) of the UEs subject to the constraint on the power budget of the access point. The merits of the SLNR-based approach are highlighted, e.g., reduced complexity, and simulation results show that the RIS-aided design can achieve massive MIMO gains with much fewer number of active antennas in the access point as compared to conventional relaying.

Sentiment analysis and opinion mining have developed significantly throughout the last decades. This study aims to assess, amongst many other aspects, people's feelings, opinions, and emotions of someone or something. This review's principal objective is to automatically assess the direction of the review and contrast this with the evaluation made by the reviewers. This will allow scientists to categorize and evaluate reviews cross-sectionally and encourage a scientific article's overall evaluation more critically. An EM-VC ensemble approach has been proposed for analyzing the reviews. This approach uses POS to achieve a syntactic sentence structure. With dictionaries, this syntactic structure enables the semantic orientation of the analysis via the scoring algorithm to be determined. To test the proposed approach's efficiency and performance about the baseline method, some tests have been performed using typical performance metrics like precision, accuracy, F1-score, and recall. The findings reveal that improvements are achieved in binary classification, ternary classification, & five-point classification compared to the existing HS-SVM method. Still, it provides a slight improvement in accuracy in the context of multiclass classification compared to others.

With the proliferation of different forms of digital communication, the opportunities for spreading Fake News are increasing constantly. More often than not people inform themselves about news and current events in social networks or blogs. By doing so, they run the risk of falling for and also – often unconsciously – spreading false information from manipulative news sites. The distribution of fake news itself is not a new phenomenon, but thanks to the advances in digital communication their circulation accelerates, leading to major impacts for the whole society. This lecture will take a closer look at several aspects. (a) What exactly are fake news? Which legal and content-related criteria must be fulfilled in order to identify fake news as such? (b) Modus Operandi: What are the typical methods of committing a crime? Moreover, the typical motives for spreading fake news are going to be examined in more detail.

In my talk I will discuss the AI and brain-computer interface (BCI) technologies overview in the context of integration with robotic and biological systems starting from classical Rosenblatt perceptron till current breaking through technologies of BCI that could change the scientific outlook of the field. This is what we currently broadly call now AI and neurotechnology. Two bright examples are ResNet inspired by the projections of a mammalian cortical column and U-Net with close to ResNet ideas and topology. The other interesting cortical column inspired NN architecture is hierarchical temporal memory introduced by the Numenta company. These brain or neuro inspired approaches are widely used in machine learning, computer vision, natural language processing, path planning and broader in robotics.In the neurotechnology field I want to reference a lot of interesting and projects dedicated to BCI. Starting from works of Miguel Nicolelis where researchers connected the robotic hand via a computer system with a motor cortex of a monkey and later Lebedev group demonstrated the adaptation of a mammalian brain to create the representation of robotic limb extending biological limbs. I want to pay the special attention to works of Kevin Warwick that implemented invasive nervous system to nervous system interface and the robot managed by neurons of a rat brain developed and trained during the experiment.The commercial company Neuralink recently demonstrated the most advanced 1536-channel BCI exploiting wireless Blue-tooth interface to record and process a brain activity. Elon Musk in his Neuralink presentation last year demanded that one of the nearest goals of the company is spinal cord injury and the recovery of the lost motor control of limbs. Several projects in neurorehabilitation targeted to the motor control demonstrated recently important success using medical programmed neurostimulators.The perspective but currently in early state of the development is the direction of neurosimulation based models that could be used for the robotic system control (exoskeleton) or as neuroprosthesis as the part of closed loop system for example projects simulating the spinal cord, where I can see interesting future opportunities both in medicine and robotics.

The aim of this work is to classify the dementia stage using microstructural white matter (WM) diffusion kurtosis indices. We develop various class imbalance machine learning methods in classifying the cognitive normals (CN), Alzheimer's disease (AD) and mild cognitive impairment (MCI) individuals. Diffusion-weighted images of 155 subjects were collected with 44 CN, 88 MCI and 23 AD individuals aged between 60 to 96 years. We first perform skull striping using FSL tool along with head motion and artifact correction. We calculate the mean diffusivity, fractional anisotropy from various brain regions of WM indices from diffusion kurtosis imaging (DKI) using tract based spatial statistics. Then we estimate the kurtosis fractional anisotropy of various WM regions. After estimation of all the regional KFA indices, we trained these indices using various class imbalance machine learning classification algorithms, as the sample size is different in all the three groups. We found that balanced random forest classifier was the best classifier with an accuracy of 74% when compared to other methods for 5-fold cross-validation. We conclude that class imbalance machine learning methods are potential in classifying cognitive normals, AD and cognitive impairment individuals.

Agile methodology has been a trending topic over the past two decades. Organizations are establishing agile transformation programs to reap its benefits, and cope with the everchanging business and technology landscape. Yet, making sense of the manifold domains contributing to the success of an agile transformation, and orchestrating them with a great synergy is not a simple task. To navigate in this complexity and establish roadmaps, organizations primarily need to understand where they stand. There are a multitude of Agile Assessment Models (AAM) that are designed with the intent of identifying the state of an agile transformation in an organization. Designing an AAM is a valuable step towards this need, however, such models tend to lack vital aspects. Although “What” is needed seems straightforward, in our research, we are trying to focus on “Why” lacking aspects are vital, and “How” to enable effective and efficient value creation through Agility.

The QNX operating system has emerged as a promising candidate as a base operating system for upcoming domain or vehicle integration computers in centralized automotive E/E. In this work, we look deeper in the Adaptive Partitioning Scheduler offered by QNX with the aim of assessing its suitability in providing temporal isolation and guaranteed execution behavior to different applications. With APS, QNX has introduced budget-based scheduling into a mainstream commercial OS and hence deserves merit. However we also found certain drawbacks in the APS scheduler and in order to mitigate the problems caused by them, we propose some guidelines for system designers to configure their systems efficiently.

Improper Input Validation (IIV) is a dangerous software vulnerability that occurs when a system does not safely handle input data. Although IIV is easy to detect and fix, it still commonly happens in practice; so, why do developers not recognize IIV? Answering this question is key to understand how to support developers in creating secure software systems. In our work, we studied to what extent developers can detect IIV and investigate underlying reasons. To do so, we conducted an online experiment with 146 software developers. In this document, we explain how to obtain the artifact package of our study, the artifact material, and how to use the artifacts.

The CAN (Controller Area Network) signals transmitted in vehicles have great analytical value with the quick development of complex automotive software. The boom in big data creates an opportunity to transmit CAN signals from the in-vehicle network to the big data cloud platform, through which the signal analysis can be conducted. The signals are transmitted from the in-vehicle network to TelematicsBOX via CAN bus and then sent to the big data cloud platform. When using the CAN bus for signal transmission of the in-vehicle network, signals larger than 1 byte need to be unpacked into several 1-byte signals. The general solution of automotive software manufacturers usually uses the model-based development method to unpack the CAN signals, but such method is inefficient. To solve this problem, we develop a CAN signal unpacking tool called AutoToolCSU, which is based on a configured template through a GUI (Graphical User Interface). Compared to the model-based development method, AutoToolCSU not only greatly improves the development efficiency of CAN signal unpacking but also interfaces with the standard development processes of automotive software manufacturers.

Intermittent systems enable continuous and accumulative process execution under constraint or unstable power supply. To enable intermittent computing, process status and data are typically checkpointed from volatile memory (VM) to nonvolatile memory (NVM) before running out of power. After power resumes, these logged data can be loaded back from NVM to VM for continuous execution. Nevertheless, existing approaches rarely considered the energy consumed during moving data and may waste precious power resource over data movement, instead of computation. Such observation motivates us to propose an energy-reduction on-chip memory management (ERCM2) scheme to utilize the high cell density and non-volatility of SpinTransfer Torque RAM (STT-RAM) for enabling a hybrid on chip memory architecture. The experimental results show that the proposed scheme can achieve the access performance close to conventional SRAM-based on-chip memory architecture with lower energy consumption.

Smart contracts have been plagued by security incidents, which resulted in substantial financial losses. Given numerous research efforts in addressing the security issues of smart contracts, we wondered how software practitioners build security into smart contracts in practice. We performed a mixture of qualitative and quantitative studies with 13 interviewees and 156 survey respondents from 35 countries across six continents to understand practitioners' perceptions and practices on smart contract security. Our study uncovers practitioners' motivations and deterrents of smart contract security, as well as how security efforts and strategies fit into the development lifecycle. We also find that blockchain platforms have astatistically significant impact on practitioners' security perceptions and practices of smart contract development. Based on our findings, we highlight future research directions and provide recommendations for practitioners.

Facing the difficulty and inefficiency of creating and maintaining High-Definition (HD) maps in our commercial deployments, we have developed an edge-based crowdsourcing task distribution framework for HD Map in autonomous driving. Our key observation is that: HD map data crowdsourcing exhibits the diminishing marginal utility thus there exists an inflection point for maximum utility, meanwhile its premature convergence of utility will leave some map updates not notified in time. Based on this observation, we develop a periodic crowdsourcing task distribution framework. It discretizes the demands for collecting source data into different periods and uses an optimal stopping rule to terminate the data collection for the maximum crowdsourcing utility. The experimental results verify that our crowdsourcing framework can achieve high time coverage and high efficiency with lower cost.

In autonomous driving, 3D object detection is es-sential as it provides basic knowledge about the environment. However, as deep learning based 3D detection methods are usually computation intensive, it is challenging to support realtime 3D object detection on edge-computing devices in selfdriving cars with limited computation and memory resources. To facilitate this, we propose a compiler-aware pruning search framework, to achieve real-time inference of 3D object detection on the resource-limited mobile devices. Specifically, a generator is applied to sample better pruning proposals in the search space based on current proposals with their performance, and an evaluator is adopted to evaluate the sampled pruning proposal performance. To accelerate the search, the evaluator employs Bayesian optimization with an ensemble of neural predictors. We demonstrate in experiments that for the first time, the pruning search framework can achieve real-time 3D object detection on mobile (Samsung Galaxy S20 phone) with state-of-the-art detection performance.

Graph neural networks (GNN) have achieved state-of-the-art performance on various industrial tasks. However, the poor efficiency of GNN inference and frequent Out-of-Memory (OOM) problem limit the successful application of GNN on edge computing platforms. To tackle these problems, a feature decomposition approach is proposed for memory efficiency optimization of GNN inference. The proposed approach could achieve outstanding optimization on various GNN models, covering a wide range of datasets, which speeds up the inference by up to 3×. Furthermore, the proposed feature decomposition could significantly reduce the peak memory usage (up to 5× in memory efficiency improvement) and mitigate OOM problems during GNN inference.

This research aims to study the aging behavior of mineral oil, Gas-to-Liquid, and natural ester oil on oil paper insulation. Experiments were carried out on oil insulation, which contained paper insulation, namely kraft paper (TUK), kraft paper (KP), pressboard (PB). Then the oil-paper insulation was placed in a sealed glass bottle & accelerated thermal aging for 1344 hours at a temperature of 150 °C. Observation of oil behavior was carried out at 336 hours intervals during aging. The aging behavior of the oil was observed by testing the oil moisture content, total acid number, breakdown voltage, color scale, interfacial tension, flash point, and Fourier Transform Infrared (FTIR). It was found that natural ester oil had better electrical performance than mineral oil and GTL. Meanwhile, natural ester oils due to their polar nature have different trends in both moisture and acid content when compared to hydrocarbon-based oils. It can be observed there is a relationship between moisture content, acidity, and breakdown voltage in all oils. The color scale shows changes with aging of the three oils, as well as the interfacial tension. Meanwhile, flash point shows the superiority of natural ester oil and GTL has a greater value than mineral oil. FTIR shows the occurrence of transesterification in natural ester oil in the 1700-1800 cm-1 spectrum where in mineral oil and GTL this is not found.

As one of the main equipment in an electric power system, transformer’s failure is very important to be avoided to maintain the continuity of the distribution of electric power. One of the most crucial diagnosis method that has been practiced widely for transformer is Dissolve Gas Analysis (DGA). Many studies have been done to develop the DGA method to determine the type of faults that occur in the transformer by using insulating oil as a diagnostic medium. During its development, the DGA method has been evolving, among those DGA methods, there is one most recent method that geometrically incorporates five combustible gas that dissolved in oil in order to determine the type of faults. These gases represent the degradation of insulating oil, while the gas that represents degradation of insulating paper is not included in this method. This study offers a way to incorporate CO gas which is an indicator of the degradation of insulation paper into a geometric method that uses six gas combinations instead of five. Determination of the type of faults is done by mapping the identified faults from the IEC TC 10 database into hexagon geometry. It is expected that later the results of this study can be used to determine the type of faults in the transformer both due to degradation of insulating oil and paper.

This paper presents a novel film type of flexible loudspeaker by ultra-thin piezoelectric PZT/Si bare chip integrated on flexible film substrate. The device is designed to operate without a closed membrane offering significant improvements with respect to manufacturability. We have revealed that a fully audible sound pressure level of around 40 dB at 6500 Hz was generated for a film with an ultrathin PZT/Si chip 1 × 5 mm2 in size. This technology is expected to be applied as electronic parts for surround or flexible parametric speakers for electronic paper and smartphones.

This paper assesses the feasibility of a novel dynamic spectrum sharing approach for a cellular downlink based on cognitive overlay to allow non-orthogonal cellular transmissions from a primary and a secondary radio access technology concurrently on the same radio resources. The 2-user Gaussian cognitive interference channel is used to model a downlink scenario in which the primary and secondary base stations are co-located. A system architecture is defined that addresses practical challenges associated with cognitive overlay, in particular the non-causal knowledge of the primary user message at the cognitive transmitter. A cognitive overlay scheme is applied that combines superposition coding with dirty paper coding, and a primary user protection criterion is derived that is specific to a scenario in which the primary system is 4G while the secondary system is 5G. Simulation is used to evaluate the achievable signal-to-interference-plus-noise ratio (SINR) at the 4G and 5G receivers, as well as the cognitive power allocation parameter as a function of distance. Results suggest that the cognitive overlay scheme is feasible when the distance to the 5G receiver is relatively small, even when a large majority of the secondary user transmit power is allocated to protecting the primary user transmission. Achievable link distances for the 5G receiver are on the order of hundreds of meters for an urban macrocell or a few kilometers for a rural macrocell.

This position paper presents and proposes new requirements for Privacy and Data Protection. We first raise misalignments of current Privacy regulations and argue that current regulatory approaches do not benefit individuals as much as expected to the point that it primarily shields large organisations from ethical management of personal data. From this assessment, we propose the Start-from-the-End (SftE, pronounced “soft”) approach to online Privacy. It puts the focus on the later stages of the lifecycle of Personal Data (such as the Right to Erasure), while removing focus from the points of collection of personal data. The ultimate goal is to reclaim straightforward enforcement and re-empower individuals in a way that is meant to be feasible and practical.

One of the most challenging goals of SISSY (Self-improving System Integration) is to create methods for self-integrating systems to improve their integration. The purpose of this short position paper is to stimulate a community-wide discussion and eventual experimentation about what it means to measure the degree of integration between entities and then to improve or deepen integrations. To start this discussion, we first note the importance of having varying degrees and kinds of integration, and then define three ways in which an entity could improve integrations within itself and with other entities.

Applications are not what they used to be. Modern HPC applications are increasingly a mix of heterogeneous tasks and services – both internal and external. This imposes new constraints and requirements on application and performance monitoring, which is fundamentally different from single-task monitoring. Runtime decisions are needed, and critically rely on monitored information to determine what to do. Thus, online monitoring and analytics must be first-class requirements of all applications and the runtime environments that support their execution. This position paper explores the implications of novel application requirements on future monitoring and profiling subsystems, in the context of ensemble computing.

In this short paper, we outline the challenges of Big Data processing for vehicle data, considering different use cases as well as different locations. On an architectural level, we show the benefits and possible limitations of the different approaches for in-vehicle processing, edge computing as well as cloud processing. We also discuss different approaches for Big Data Processing, including batch processing as well as distributed stream processing.