Search

This paper proposes a posture recognition system that can be applied for medical surveillance. The proposed method estimates human posture using mobilenetV2 and long short-term memory (LSTM) to extract the important features of an image. The output of the system was a fully estimated skeleton. We used seven human indoor postures, including lying, sitting, crouching, standing, walking, fighting, and falling, and classified them. The output results are the extraction of the human skeleton and the corresponding labels for the poses. We first experiment with classification using machine learning. The system only achieves approximately 88% accuracy because it is not able to classify similar postures, such as standing and walking. This difference can be caused by the extraction of features for static images, and the machine learning classification algorithm has not reached accuracy with training data. Therefore, we proposed the integration of the LSTM model into the proposed system. LSTM learns the features of the skeleton and provides classification results for postures. As a result, our system improved the accuracy by up to 99%. Similar postures, such as standing and walking, have improved accuracy by up to 7%. In addition, we performed the system on the Jetson Nano hardware. The results show that it can run on a low-profile (44% CPU and 2.1 frames per second) that is capable of applications for remote patient monitoring devices.

Adversarial machine learning defenses have primarily been focused on mitigating static, white-box attacks. However, it remains an open question whether such defenses are robust under an adaptive black-box adversary. In this paper, we specifically focus on the black-box threat model and make the following contributions: First we develop an enhanced adaptive black-box attack which is experimentally shown to be $\geq 30\%$ more effective than the original adaptive black-box attack proposed by Papernot et al. For our second contribution, we test 10 recent defenses using our new attack and propose our own black-box defense (barrier zones). We show that our defense based on barrier zones offers significant improvements in security over state-of-the-art defenses. This improvement includes greater than 85% robust accuracy against black-box boundary attacks, transfer attacks and our new adaptive black-box attack, for the datasets we study. For completeness, we verify our claims through extensive experimentation with 10 other defenses using three adversarial models (14 different black-box attacks) on two datasets (CIFAR-10 and Fashion-MNIST).

In this paper, the advantages of multiple reconfigurable intelligent surfaces (RISs) assisted wireless systems and direct link between two terminals are exploited by considering a cooperative communication system where reflecting links created by multiple RISs and the direct link are combined at the receiver. We mathematically analyze the performance of the cooperative multiple RISs - direct link (RIS-D) system over Nakagami- $m$ fading channels by obtaining the symbol error probability (SEP) expression. In addition, the SEP of the single-input single-output (SISO) system without RISs (i.e., there is only direct link between two terminals) is also obtained for convenience in comparison the performance of the cooperative RIS-D and SISO systems. We demonstrate the correctness of the derived expressions via Monte-Carlo simulations. Numerical results show that the SEP of the cooperative RIS-D system significantly lower than that of the SISO system for a certain transmission power of the transmitter. Also, for a certain SEP target, the cooperative RIS-D system can greatly save energy consumption in comparison with the SISO system. Specifically, increasing the number of RISs or the number of reflecting elements (REs) in the RISs can remarkably reduce the SEP of the cooperative RIS-D system. On the other hand, the impacts of the system parameters such as the number of RISs, the number of REs, and the Nakagami- $m$ fading channels are fully investigated to clearly indicate their influences on the SEP of the cooperative RIS-D system.

This paper proposes a new online auto-tuning method to improve the accuracy and reduce the tuning time of permanent magnet synchronous motor (PMSM) drives. Under varying loads, the ability to tune the controllers of PMSM drives using optimal tuning time is crucial. However, direct tuning of controller parameters using estimated parameters or conventional particle swarm optimization (PSO) methods do not satisfy the performance criteria. To solve this problem, the new method combining mechanical parameter estimation (MPE) and multi-layer particle swarm optimization (MLPSO) with K-means clustering (KMC) and an adaptive learning strategy (ALS) is proposed. First, the combination of an MPE method with a lookup table (LUT) for initial parameter selection is introduced to reduce the iteration time. Then, the MLPSO-KMCALS method is proposed as an improvement over the conventional PSO method by increasing the number of layers, grouping the swarm into several subswarms, and using the ALS for each particle to increase the population diversity and optimize the controller parameters within the shortest possible amount of time. Finally, a disturbance load torque observer is applied to compensate for the effect of external disturbances after tuning. The effectiveness of the proposed method is validated through experiments conducted under practical conditions.

This paper presents a dual-circularly polarized (dual-CP) antenna with simple configuration, low-profile, and wide bandwidth. Its primary radiating element is an aperture-coupled patch loaded with a stacked patch for a wide operational bandwidth. A single-section hybrid coupler is utilized as a feeding structure of the antenna to generate dual-CP radiation. Different from the priors normally requiring both wideband radiator and wideband feeding structure, this work demonstrates that the wideband dual-CP radiation can be obtained by combining a wideband radiator and a conventional hybrid coupler. A design operating at the center frequency of 5.5 GHz has been fabricated and tested. The measurements result in a −10-dB reflection coefficient bandwidth of 30.0% (4.66 – 6.31 GHz), a 10-dB isolation bandwidth of 30.9% (4.66 – 6.35 GHz), a 3-dB axial ratio (AR) bandwidth of 29.1% (4.70 – 6.30 GHz), and a realized broadside gain of 5.5 – 7.05 dBic. For a more robust dual-CP radiation, a low-sidelobe 4 $\times $ 4 element array of the proposed antenna has been designed, fabricated, and measured. Compared to the single element design, the array prototype yields similar impedance and 3-dB AR bandwidths, but a higher isolation. The measurements on the array prototype result in an isolation of $\geq 14$ dB, a peak gain of 16.3 dBic, a side-lobe level of $\leq -20$ dB, and a cross-polarization of $\leq -16$ dB across the frequency range of 4.7 – 6.3 GHz (29.1%).

The disturbance and uncertainty of the motor drive systems are very complicated terms. There is no exception for the slotless-self bearing motor (SSBM), where the perturbations of the bearing motor are mainly came from the outside as the wind affect, from inside as the thermal changing of the coils, and incorrect modeling of the winding processes. First, to delete these inversed terms, this paper proposes a new super-twisting disturbance observer (STDOB) to obtain the desired goal of the robust control design. The proposed disturbance observer was based on the information of measured and estimated states with the aim of softening the cost of the measurement. Second, to estimate the velocities and accelerations of the movements on $x-$ and $y-$ axes, the stability concept of homogeneous function-based was used to design the fixed-time state observers (FTSOBs) for these axes. The state of the rotational operation on $\omega -$ axis was estimated with a fixed-time state observer. Third, to control the positions and rotational speed, a variable boundary layer thickness (VBLT) fixed-time sliding mode control (FTSMC) was designed to force these positions and speed states converge to the desired goals. Finally, the stability of the proposed control algorithm was theoretically verified by using Lyapunov condition and simulation of MATLAB software. The obtained states were acceptably stable with small overshoots, small settling-times, and stable steady-states.

In recent decades, digital transformation has received growing attention worldwide, that has leveraged the explosion of digitized document data. In this paper, we address the problem of parsing publications, in particular, Vietnamese publications. The Vietnamese publications are well-known with high variant, diverse layouts, and some characters are equivocal in the visual form due to accent symbols and derivative characters that pose many challenges. To this end, we collect the UIT-DODV-Ext dataset: a challenging Vietnamese document image including scientific papers and textbooks with 5,000 fully annotated images. We introduce a general framework to parse Vietnamese publications containing two components: page object detection and caption recognition. We further conduct an extensive benchmark with various state-of-the-art object detection and text recognition methods. Finally, we present a hybrid parser which achieves the top place in the benchmark. Extensive experiments on the UIT-DODV-Ext dataset provide a comprehensive evaluation and insightful analysis.

The Doppler effect critically degrades the performance of orthogonal frequency division multiplexing (OFDM) systems in general. This problem is significantly worse for underwater acoustic (UWA) communication systems due to the distinct characteristics of the underwater channel, resulting in the loss of orthogonality among sub-carriers. In order to compensate Doppler shifts, including phase noise and multipath channels in realistic communication scenarios, the joint of channel estimation and ICI reduction is often performed. However, the accuracy depends on the channel estimation and the FFT size, while this leads to increased computational complexity at the receiver. To achieve this dual goal in the actual underwater communication environment, a novel pilot structure in the frequency domain has been applied to overcome the channel impulse response (CIR) variation in a block period. The coarse Doppler shift is firstly estimated by using the received pilot signal. Afterward, the study takes advantage of the flexibility provided by non-uniform fast Fourier transform (NFFT) in choosing the sampling points to construct a fast and stable Doppler frequency Compensation Matrix-based NFFT (DCMN) to fine compensate the Doppler phase shift. Finally, this study shows the improvement of the proposed method’s performance by actual experimental measurements and simulations.

Unmanned aerial vehicles, especially quadcopters, play key roles in many real-world applications and the related quadcopter autonomous control algorithms have attracted a great deal of attention. In this paper, we address the vision-based autonomous landing problem of a quadcopter on a ground moving target. Firstly, we propose a disturbance observer-based control algorithm, consisting of a nonlinear disturbance observer and robust altitude and attitude controllers. This algorithm is based on the quadcopter dynamics model, and its stability is strictly proved using Lyapunov’s theory. Secondly, we develop an autonomous landing planner which we test for various landing scenarios to deliver improved reliability and accuracy of the landing mission. These theoretical studies are complemented by a numerical feasibility study, before demonstrating the effectiveness of our approach under actual flight conditions with an experimental quadcopter platform.

Text-based Image Captioning has been a novel problem since 2020. This topic remains challenging because it requires the model to comprehend not only the visual context but also the scene texts that appear in an image. Therefore, the ways image and scene texts are embedded into the main model for training is crucial. Based on the M4C-Captioner model, this paper proposes the simple but effective EAES embedding module for effectively embedding images and scene texts into the multimodal Transformer layers. In detail, our EAES module contains two significant sub-modules: Objects-augmented and Grid features augmentation. With the Objects-augmented module, we provide the relative geometry feature, representing the relation between objects and between OCR tokens. Furthermore, we extract the grid features for an image with the Grid features augmentation module and combine it with visual objects, which help the model focus on both salient objects and the general context of an image, leading to better performance. We use the TextCaps dataset as the benchmark to prove the effectiveness of our approach on five standard metrics: BLEU4, METEOR, ROUGE-L, SPICE and CIDEr. Without bells and whistles, our method achieves 20.21% on the BLEU4 metric and 85.78% on the CIDEr metric, 1.31% and 4.78% higher, respectively, than the baseline M4C-Captioner method. Furthermore, the results are incredibly competitive with other methods on METEOR, ROUGE-L and SPICE metrics. Source code is available at https://github.com/UIT-Together/EAES_m4c.

Channel congestion has been an open challenge for vehicular networks due to the limited resource of communication channels. Explosion of channel access requests from a massive number of transmitter vehicles can exhaust bandwidth and then degrade transmission quality. The rapid drop of messages (because of the high bit error rate in the transmission congestion condition) can threaten the safety of connected vehicles. Maintaining congestion-free communications is then essential to improve the reliability for vehicular networks, including Cellular-V2X (C-V2X)-based cooperative intelligent transport systems and road-safety applications. In this work, we present a novel intelligent transmission control model, namely DEEPCUT, to automatically adjust the message broadcasting rate of a transmitter vehicle. DEEPCUT works based on a Double Deep Q-learning Networks with Prioritized Experience Relay framework. DEEPCUT encourages the transmitter vehicle to (1) reduce its broadcasting rate if the vehicle is maintaining a safe distance from its neighbors and (2) increase the rate if the vehicle is approaching the others at a high-risk distance, all done by using reward/punish strategies. The evaluation results show that DEEPCUT can cut up 16% redundant data while increasing 22% packet reception rate compared with baseline models, particularly in crowd vehicular communications. Our risk-based transmission control can be an excellent complement to address the congestion when the channel cannot satisfy every vehicle’s resource requests. At best, the risk assessment-based approach in our congestion control method can provide a novel material to enhance Decentralized Congestion Control (DCC) for 5G V2X sidelink in the coming specifications.

In developing countries as well as in Vietnam, infrastructure for two-wheeled electric vehicles is necessary but has not drawn much attention. Sustainable transportation in urban traffic requires sustainable solutions to first-mile/last-mile services which are essential for eco-friendly and efficient connection to the public transport system. Hence, electric two-wheeler charging stations which coordinate charging a large number of electric bicycles/electric motorcycles at public parking or transit hubs should be considered. In this study, the authors propose a novel approach to effectively allocate power to electric two-wheelers in a charging station which is interconnected with the distribution grid, a photovoltaic system and conventional loads of a building. With a three-stage of allocation, the proposed algorithm proves its effectiveness in reducing load variance as well as peak shaving and valley filling while relieving the complexity of solving multivariable optimal problems. The idea also contributes to increasing the penetration rate of both electric vehicles and distributed photovoltaic sources while alleviating adverse effects of these emergent factors.

The blossom of renewable energy worldwide and its uncertain nature have driven the need for a more intelligent power system with the deep integration of smart power electronics. The smart inverter is one of the most critical components for the optimal operation of Smart Grid. However, due to the deep information and communication technology (ICT) infrastructure implementation that most inverter-based smart power systems tend to have, they are vulnerable to severe external threats such as cyberattacks by hackers. This paper presents a comprehensive review of the system structure and vulnerabilities of typical inverter-based power system with distributed energy resources (DERs) integration, nature of several types of cyberattacks, state-of-the-art defense strategies including several detection and mitigation techniques, and an overview and comparison of testbed and simulation tools applicable for cyber-physical research. Finally, challenges, unsolved problems, and future direction of the field are discussed and concluded at the end of the journal. This paper provides an all-inclusive survey at the state of the art smart grid cybersecurity research and paves the path for potential research topics in the future.

The requirement of high quality of service (QoS) in multi-priority industrial and domestic sensor networks poses new challenges to the increasing adoption of Internet of Things (IoT). In Multi-event Wireless Sensor Networks (MWSNs), nodes generate different types of data packets i.e., urgent (high priority) or normal (low priority), with different traffic proportions. High priority packets require an assurance of faster transmission and higher reliability in the network. In the literature, the existing medium access control (MAC) protocols for MWSNs have limited consideration in supporting data priority with different traffic proportions. Therefore, this paper proposes an energy efficient MAC scheme that incorporates multi-priority of data packets with dynamic traffic proportion, called PriTraCon-MAC. PriTraCon-MAC supports multi-events by considering four different priority levels of data packets and uses a novel approach that adjusts the contention window adaptively. Due to that, Request-To-Send frame of higher priority data can be sent earlier in the contention window, resulting in the corresponding faster acceptance by the receiver. Furthermore, mathematical delay analysis with different priority traffic proportions has also been undertaken. In addition, PriTraCon-MAC has been implemented in OMNET++ Castalia and its performance has been evaluated in terms of packet delay, reliability, and energy consumption, and compared with the existing Timeout Multi-priority based-MAC (TMPQ-MAC) under various network conditions. The simulation results demonstrate that PriTraCon-MAC offers lower average delay and achieves significantly higher packet success rate, while reducing energy consumption by up to 80% when compared with TMPQ-MAC protocol.

Accurate alignment of multi-session medical imaging is essential to the analysis of disease progression. By comparing the magnetic resonance imaging (MRI) data captured before and after a course of neoadjuvant chemoradiation (nCRT) treatment, physicians are able to evaluate the tumor response for further treatment of the disease. However, rectal MRI data captured in multi-session are often misaligned and not guaranteed to have one-to-one correspondence, making it challenging for physicians to observe the treatment response of tumor. To address this issue, we propose an unsupervised learning based volume registration framework, which enables accurate alignment even under a high degree of deformation between multi-session rectal data. Moreover, it works without the assumption of one-to-one correspondence between multi-session data, and hence is a general solution to rectal MRI volume registration. The experimental results show that the proposed registration framework accurately aligns rectal cancer images and outperforms other state-of-the-art methods in medical image registration. By providing accurate registration, it can potentially increase the efficiency and reduce the workload for physicians to evaluate the rectal tumor response to nCRT.

A good suspension system is of paramount importance to the operating performance of a vehicle and, consequently, to the safety and driving comfort of the passengers. Nevertheless, suspension systems are commonly susceptible to nonlinearity, parameter uncertainty, and exogenous perturbation, which can easily impair their effectiveness. This study first employs a full state feedback super twisting control (FS-STC) to stabilize both vertical displacement and pitch angle of a half-car suspension system in the presence of disturbances. FS-STC inherits the robust property of sliding mode control (SMC) while effectively attenuating the chattering phenomenon as one of its attractive features. However, FS-STC strictly requires both direct displacement and velocity state feedback, which implies additional sensors have to be installed, thus increasing the complexity of the physical structure and being prone to measurement noises. Therefore, a higher order sliding mode observer (HOSMO) based STC (HOSMO-STC) and an unscented Kalman filter (UKF) based STC (UKF-STC) are subsequently proposed to tackle this state availability problem. HOSMO estimates velocity states, thus reducing the dependence on state feedback for STC design. Meanwhile, UKF implementation takes further actions by utilizing more common and easily accessible relative displacements such as suspension strokes to estimate all concerned system states. Comparative simulation results demonstrate that UKF-STC offers better performance in terms of both convergence accuracy and chattering alleviation compared to FS-STC and HOSMO-STC while requiring the least information of state feedback.

Polarization-control devices have attracted considerable interest, however, most of the polarization converters operating at lower frequencies have a heavy design and narrow bandwidth which limits their practical applications. Here we report a simple design of an ultra-wideband and lightweight polarization converter for applications in the S- and C-bands. The proposed converter is designed based on a metasurface structure with the dielectric layer modified to hollow structure to obtain a lightweight design even working at such low frequency. Theoretical analysis and simulation results indicate that the converter can convert the orthogonal polarization transformation of reflected wave. Furthermore, the measurement results show good agreement with the simulation results. The proposed polarization converter can achieve a polarization conversion ratio above 90% in an ultra-wide frequency range from 2 to 8.45 GHz due to multi-resonance modes. These performances are going beyond state of the art in terms of bandwidth and lightweight design, thus it can be applied in various applications in the operating bands.

Thanks to its ability to provide immersive experience to users, 360-degree video has become one of the key enablers of Virtual Reality. However, the huge data size of 360-degree video poses a challenging problem for live streaming of this special type of video over resource-constrained networks. In this paper, we propose a novel framework for live 360-degree video streaming to multiple users over mobile networks. Our proposed framework jointly utilizes Scalable Video Coding and multicast to deliver 360 video to users in a bandwidth-efficient manner. In particular, 360-degree video is split into small parts called tiles, each is encoded into multiple layers using Scalable Video Encoding. The proposed framework then selects suitable tile layers to maximize the overall Quality of Experience of all users. To support real-time adaptation, we design a Linear Regression-based algorithm to estimate the weights of tiles of individual users. In addition, an efficient algorithm for deciding the suitable tile layers and their transmission modes is proposed. Experimental results show that the proposed method can significantly improve the average viewport quality compared to state-of-the-art methods.

Domain adaptation methods in machine learning deal with the domain shift issue by aligning source and target data representation. This paper proposes a novel domain adaptation method for semantic segmentation that exploits the Fourier transform on chromatic space to improve the quality of style transfer, and generates pseudo-labels for self-training by combining the results from different teachers obtained at different rounds of self-training. Our method also applies class-level adversarial learning to achieve a more fine-grained alignment between the two domains, and a late fusion with a depth-estimation model to improve its segmentation outputs. Experiments show that our method yields superior performance in terms of accuracy compared to other existing state-of-the-art methods.

In this paper, a dual-polarized filtering Fabry-Perot antenna (FPA) with high-isolation is proposed. It consists of a feed element of differential-fed dual-polarized square patch and a partially reflecting surface (PRS). The patch is capacitively coupled with T-shaped resonators and shorting-vias to obtain broadband characteristic and radiation nulls. The PRS structure is composed of two complementary metasurface layers, which are selected for achieving a positive reflection phase gradient within the broad frequency range. More interestingly, thanks to the frequency selectivity feature, the PRS significantly enhances the filtering characteristic of the FPA. For verification, a prototype of the proposed antenna operating at the 5.5-GHz center frequency has been fabricated and measured. The prototype with an overall size of $\sim 2.0\lambda _{\text {min}} \times 2.0\lambda _{\text {min}} \times 0.49\lambda _{\text {min}}$ ( $\lambda _{\text {min}}$ is the free-space wavelength referring to the lowest operational frequency) result in an impedance bandwidth of 17.1% ( $5.02-5.96$ GHz) for 10-dB return loss and a high isolation of $\ge45$ dB. Moreover, the far-field measurements result in a good dual-polarized radiation with the peak gain of 13.0 dBi, cross-polarization level of $\le $ –25 dB within the passband, and out-of-band suppression level of $\ge20$ dB.

This paper proposes a new method to design a single-layer dual-band circularly polarized (CP) patch antenna with a small frequency ratio. The design consists of one or several pairs of rectangular patches proximity-fed by a 50- $\Omega $ microstrip line with an open-circuit termination. By exploiting both capacitive and inductive coupling mechanisms, and both orthogonal radiating modes of these patches, the antenna can be designed to operate at two close resonance frequencies. Due to its simple and single-layer structure, the antenna can be easily adapted with meshed configuration, which is suitable for transparent devices. For verification, a dual-band CP meshed patch antenna with a frequency ratio of 1.12 and two pairs of patch is designed, fabricated, and tested. The measurements show that the antenna prototype provides a $|S_{11}| < -10$ -dB bandwidth of 4.82–5.03 GHz (210 MHz) and 5.49–5.78 GHz (290 MHz), axial ratio < 3-dB bandwidth of 4.88–4.93 GHz (50 MHz) and 5.50–5.57 GHz (70 MHz), and broadside realized gains of 9.0 dBic and 8.6 dBic for the lower and upper bands, respectively.

This study investigates a combined system comprising non-orthogonal multiple access (NOMA) and beamforming in a downlink network. To fully exploit the advantages of NOMA, user (UE) pairing and beamforming design are jointly optimized via a generalized model for UE association, subject to energy efficiency maximization. Owing to the combination of binary variables and nonconvex constraints, the resulting optimization problem belongs to the class of mixed-integer nonconvex programming. An innovative algorithm, integrating the inner-approximation and Dinkelbach methods, is proposed herein to address a nonconvex fractional function. By introducing a pairing matrix and relaxing the binary variables into continuous ones, our approach is capable of reaching an optimal solution, where two arbitrary UEs are optimally paired regardless of geographical or spatial constraints. For practical scenarios, we further propose a robust design to manage the effect of channel estimation errors under settings involving channel uncertainty. Numerical results show that our proposed designs, even with the presence of the imperfect channel state information at the base station, significantly outperform the conventional beamforming and existing pairing schemes.

We are now witnessing the rapid growth of advanced technologies and their application, leading to Smart Manufacturing (SM). The Internet of Things (IoT) is one of the main technologies used to enable smart factories, which is connecting all industrial assets, including machines and control systems, with the information systems and the business processes. Industrial Control Systems of smart IoT-based factories are one of the top industries attacked by numerous threats, especially unknown and novel attacks. As a result, with the distributed structure of plenty of IoT front-end sensing devices in SM, an effectively distributed anomaly detection (AD) architecture for IoT-based ICSs should: achieve high detection performance, train and learn new data patterns in a fast time scale, and have lightweight to be deployed on resource-constrained edge devices. To date, most solutions for anomaly detection have not fulfilled all of these requirements. In addition, the interpretability of why an instance is predicted to be abnormal is hardly concerned. In this paper, we propose the so- called FedeX architecture to address those challenges. The experiments show that FedeX outperforms 14 other existing anomaly detection solutions on all detection metrics with the liquid storage data set. And with Recall of 1 and F1-score of 0.9857, it also outperforms those solutions on the SWAT data set. FedeX is also proven to be fast in terms of training time of about 7.5 minutes and lightweight in terms of hardware requirement with memory consumption of 14%, allowing us to deploy anomaly detection tasks on top of edge computing infrastructure and in real-time. Besides, FedeX is considered as one of the frameworks at the forefront of interpreting the predicted anomalies by using XAI, which enables experts to make quick decisions and trust the model more.

In this paper, a novel and effective method for designing the pulse shaping filter for multicarrier faster-than-Nyquist with offset quadrature amplitude modulation (MFTN-OQAM) systems is proposed. The connection between the signal-to-interference ratio (SIR) and the filter coefficients is first established. Then, for a desired overall compression level taking into account compressions in both frequency and time domains, a simple convergence search is suggested to jointly find the optimal values of time and frequency compression factors as well as the filter coefficients to maximize the SIR under a spectrum localization constraint. The obtained results show that higher SIRs, and consequently better bit error rates, can be achieved by the proposed filters over the Martin filter that is commonly used in the filter-bank multicarrier OQAM (FBMC-OQAM) systems (i.e., without time or frequency compression). Moreover, when applying our method to FBMC-OQAM systems, the obtained results show that the original Martin filter is suboptimal as shorter filters are found having the same SIR, which translates to lower implementation cost and latency.

This paper investigates performance of simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-enabled multiple two-way full-duplex device-to-device (D2D) communication systems over Rayleigh fading channels under optimal and uncertain phase shift alignments. We derive closed-form expressions for outage probability (OP), sum throughput, ergodic capacity (EC) and energy efficiency. To gain insights, we quantify and reveal some useful guidelines for the performance behavior of the OP and the EC, such as diversity order and ergodic slope from high transmit power configuration. In addition, some critical points also deduced for the sum throughput and the system energy efficiency. Moreover, the impacts of the transmit power configurations, RIS deployments, allocating target data rate transmission, and the number of user deployments on the system performance are examined. Finally, we present some extensive simulations using Monte-Carlo method to corroborate the accuracy of the theoretical analysis.

The volume and complexity of publicly available real estate data have been snowballing. As a result, information extraction and processing have become increasingly challenging and essential for many PropTech (Property Technology) companies worldwide. The challenges are even more pronounced with languages other than English, such as Vietnamese, where few studies in this field have taken place. This paper presents an end-to-end framework for automatically collecting real estate advertisement posts from different data sources, extracting useful information, and storing computed data into proper data warehouses and data marts for the Vietnamese advertisement posts in real estate. After that, one can serve aggregated data for other descriptive and predictive analytics. We combine two models for constructing the most appropriate extraction step: Noise Filtering and Named Entity Recognition (NER). These models can help process initial input data and extract all helpful information. The experiment results show that using $\text{PhoBERT}_{large}$ can achieve the best performance compared to other approaches. Furthermore, we can obtain the corresponding F1 scores of the Noise filtering module and the NER module as 0.8697 and 0.8996, respectively. Finally, we utilize Superset for implementing analytic dashboards to visualize the predicted results and serve for further analysis and management processes.

Fulcrum Random Linear Network Coding (RLNC) combines outer coding in a large Galois Field, e.g., $GF(2^{8})$ , with inner coding in $GF(2)$ to flexibly trade off the strong protection (low probability of linear dependent coding coefficients) of $GF(2^{8})$ with the low computational complexity of $GF(2)$ . However, the existing Fulcrum RLNC approaches are generation based, leading to large packet delays due to the joint processing of all packets in a generation in the encoder and decoder. In order to avoid these delays, we introduce Fulcrum Sliding Window (FSW) coding. We introduce two flavors of FSW: Fulcrum Non-systematic Sliding Window (FNSW), which divides a given generation into multiple partially overlapping blocks, and Fulcrum Systematic Sliding Window (FSSW), which intersperses coded packets among the uncoded (systematic) transmission of the source packets in a generation. Our extensive evaluations indicate that FSSW substantially reduces the in-order packet delay (for moderately large generation and window sizes down to less than one fourth) and more than doubles the encoding and decoding (computation) throughput compared to generation-based Fulcrum.

To respond to the ever-increasing requirements of automotive electronics, universities need to adapt and play an active role in creating a new research environment and provide proper education to the young generation, which will face challenging tasks in this rapidly developed field. Vehicular technology education is crucial for the next generation of automotive electronics developers and engineers. Several education models, such as problem- and project-based learning (PPBL), academia–academia collaboration, and academia–industry collaboration, are highlighted as successful examples. [ABSTRACT FROM AUTHOR]

Hybrid beamforming for multiple-input/multiple-output (MIMO) communications is an attractive technology for realizing extremely massive MIMO systems envisioned for future wireless communications in a scalable and power-efficient manner. However, the fact that hybrid MIMO systems implement part of their beamforming in analog and part in digital makes the optimization of their beampattern notably more challenging compared with conventional fully digital MIMO. Consequently, recent years have witnessed growing interest in using data-aided artificial intelligence (AI) tools for hybrid beamforming design. This article reviews candidate strategies to leverage data to improve real-time hybrid beamforming design. We discuss the architectural constraints and characterize the core challenges associated with hybrid beamforming optimization. We then present how these challenges are treated via conventional optimization, and identify different AI-aided design approaches. These can be roughly divided into purely data-driven deep learning models and different forms of deep unfolding techniques for combining AI with classical optimization. We provide a systematic comparative study between existing approaches, including both numerical evaluations and qualitative measures. We conclude by presenting future research opportunities associated with the incorporation of AI in hybrid MIMO systems. [ABSTRACT FROM AUTHOR]

Load forecasting is more important than ever to enable new monitor and control functionalities of distribution networks aiming to mitigate the impact of the energy transition. Load forecasting at medium voltage (MV) level is becoming more challenging, because these load profiles become more stochastic due to the increasing penetration of photovoltaic (PV) generation in distribution networks. This work combines medium to low voltage (MV/LV) transformer loadings measured with advanced metering infrastructure (AMI) and machine learning (ML) algorithms to propose a new clustering based day-ahead aggregated load forecasting approach. This four-step approach improves the day-ahead load forecast of a city. First, MV/LV transformer loadings are clustered based on the shape of their load pattern. Second, a gradient boosting algorithm is used to forecast the load of each cluster and calculate the related feature importance. Third, feature selection is applied to improve the forecast accuracy of each cluster. Finally, the day-ahead load forecast of all clusters are aggregated. The case study presented uses 519 measured MV/LV transformer loadings in a city to perform 30 day-ahead load forecasts. Compared against the day-ahead aggregated load forecast without clustering, the average normalized root mean squared error (NRMSE) reduced 12.7 %, the average mean absolute percentage error (MAPE) reduced 18.2 % and the average Pearson Correlation Coefficient (PCC) increased 0.37 %. The 95 % confidence interval of the difference between the average NRMSE, MAPE and PCC without clustering and with the proposed method indicates a statistically significant improvement in accuracy.

To meet the increasing demand for group activities, single-user recommender systems need to be scaled up to provide recommendations to groups of users. This issue is solved by aggregating item preference profiles of individual group members into a single item preference profile, thereby allowing recommendations to be created for this item preference profile. In this paper, we introduce the concept of deep profiles of users, and we propose group recommendation methods based on the aggregation of group members’ deep profiles, instead of item preference profiles as in previous studies. The term deep profile refers to the users’ profiles that lie deep within the recommendation algorithms. Experiments have shown that group recommendations based on deep profiles give higher efficiency in terms of F1-score and nDCG than those based on item preference profiles.

To develop emerging transmission techniques for potential applications of Internet of Things (IoT), the system performance analysis of a cognitive radio (CR)-enabled ambient backscatter (AmBC) system will be studied in this paper with functionality of non-orthogonal multiple access (NOMA). In the proposed scheme, a base station communicates with two destinations via a designated backscatter device. It is assumed that the relay node is fitted with two different interfaces and can simultaneously collect/decode and backscatter the received source signals. Such transmission mechanism benefits to design various applications in IoT as well as wireless systems with improved performance. To exhibit system performance metrics, the outage probability and the ergodic capacity of the recipient nodes are derived analytically. Furthermore, it is shown that employing AmBC NOMA together with CR for secondary communication can significantly improve overall network performance in terms of the achievable throughput in delay-limited and delay-tolerant modes and outage probability. Numerical results show that: 1) The proposed system can improve the spectrum efficiency by employing both CR and NOMA techniques; 2) Compared with the orthogonal multiple access (OMA)-aided AmBC systems, the considered CR AmBC system relying on NOMA can obtain better reliability in the whole range of SNR; 3) There are error floors for the outage probability in the high SNR regime due to required target rates; 4) There exists a trade-off between system performance of IoT devices and power allocation coefficients associated with NOMA; 5) We find energy efficiency factor as evident of further improvement in such system.

The objective of depth completion is to generate a dense depth map by upsampling a sparse one. However, irregular sparse patterns or the lack of groundtruth data caused by unstructured data make depth completion extremely challenging. Sensor fusion using both RGB and LIDAR sensors can help produce a more reliable context with higher accuracy. Compared with previous approaches, this method takes semantic segmentation images as additional input and develops an unsupervised loss function. Thus, when combined with supervised depth loss, the depth completion problem is considered as semi-supervised learning. We used an adapted Wasserstein Generative Adversarial Network architecture instead of applying the traditional autoencoder approach and post-processing process to preserve valid depth measurements received from the input and further enhance the depth value precision of the results. Our proposed method was evaluated on the KITTI depth completion benchmark, and its performance in depth completion was proven to be competitive.

This paper presents an online monitoring system to classify the leakage/discharge current of the insulator in coastal sites using Bidirectional Long short-term memory (Bi-LSTM) models on a web-based service. The remote monitoring methodology uses environmental parameters to classify the peak level of leakage/discharge current of the 15kV and 25kV distribution insulators. The sequential weather data, the humidity, temperature, rainfall, dew point, solar illumination, wind speed, air pressure, and wind direction are automatically collected hourly in real-time and transferred to data servers. The hyperparameter optimization for the structure of Bi-LSTM is utilized through the grid search capability in a deep learning machine. The optimized design of Bi-LSTMs improves the performance and accuracy in predicting the leakage/discharge current classification for HDPE and SR of 15kV and 25kV insulators. Compared with persistent methodologies such as recurrent neural networks (RNN), long short-term memory (LSTM), and the gated recurrent unit (GRU), the Bi-LSTM has better performance and higher accuracy in predicting the leakage/discharge levels, which are utilized to evaluate the surface pollution of insulators. The optimized structure of the proposed Bi-LSTM model could achieve a maximum improvement of 49.529% error, 12.761% accuracy, 72.736% error, and 36.641% accuracy for training and validating data compared with other models. Moreover, a web-based service is developed for maintenance staff to interact with all current and predicted status of insulators. This online leakage current classification has been installed in Mailiao Township in Taiwan and could establish a reasonable maintenance mechanism for 15kV and 25kV distribution insulators.

Dish understanding from digital media is an interesting problem, but it also contains a big challenge. The challenge comes from the complexity of ingredients in the dish. With the development of deep learning, several effective tools can solve the problem partially. In this work, the task of dish recognition is considered. A novel dish recognition method based on EfficientNet architecture and transfer learning is proposed. First, we modify the EfficientNet-B0 by adding several important layers. Second, we use transfer learning to utilize optimal parameters obtained from pretraining the model on ImageNet, and then retrain it on a new dataset of dish images, i.e., UEH-VDR dataset. The UEH-VDR dataset contains images about Vietnamese dishes collected from various sources. Experimental results show that the proposed method can achieve an accuracy of 92.33% for the task of recognizing a dish. It also works more effectively than other models based on popular convolutional neural networks such as VGG and ResNet. In addition, a mobile application is also developed based on the trained data to serve visitors who want to discover the Vietnamese culinary culture.

The evolution of automotive technology will eventually permit the automated driving system on the vehicle to handle all circumstances. Human occupants will be just passengers. This poses security issues that need to be addressed. This paper has two aims. The first one investigates strategies for robustifying scene analysis of adversarial road scenes. A taxonomy of the defense mechanisms for countering adversarial perturbations is initially presented, classifying those mechanisms in three major categories: those that modify the data, those that propose adding extra models, and those that focus on modifying the models deployed for scene analysis. Motivated by the limited number of surveys in the first category, we further analyze the approaches that utilize input transformation operations as countermeasures, further classifying them in supervised and unsupervised methods and highlighting both their strengths and weaknesses. The second aim of this paper is to publish CarlaScenes dataset produced using the CARLA simulator. An extensive evaluation study, on CarlaScenes, is performed testing the supervised deep learning approaches that have been either proposed for image restoration or adversarial noise removal. The study presents insights on the robustness of the aforementioned approaches in mitigating adversarial attacks in scene analysis operations.

This study proposes an optimized model of a micro-energy network (MEN) that includes electricity and natural gas with integrated solar, wind, and energy storage systems (ESSs). The proposed model is based on energy hubs (EHs) and it aims to minimize operation costs and greenhouse emissions. The research is motivated by the increasing use of renewable energies and ESSs for secure energy supply while reducing operation costs and environment effects. A general algebraic modeling system (GAMS) is used to solve the optimal operation problem in the MEN. The results demonstrate that an optimal MEN formed by multiple EHs can provide appropriate and flexible responses to fluctuations in electricity prices and adjustments between time periods and seasons. It also yields significant reductions in operation costs and emissions. The proposed model can contribute to future research by providing a more efficient network model (as compared with the traditional electricity supply system) to scale down the environmental and economic impacts of electricity storage and supply systems on MEN operation.

In recent years, many pulse width modulation techniques have been explored for three-level impedance-source inverters. Among them, a space vector modulation (SVM) technique using upper/lower shoot-through (UST/LST) insertion provides high voltage gain and satisfactory output voltage quality. This paper further introduces a new SVM control method to reduce the magnitude of common-mode voltage (CMV) without affecting the output voltage quality and voltage gain. With this approach, only small vectors with low magnitudes of CMV are adopted to synthesize an output voltage vector. The UST and LST states are also inserted to these small vectors to boost the DC-link voltage in high voltage gain and high modulation index. The comparison of CMVs between this strategy and other schemes is presented to demonstrate the effectiveness of the proposed method. The simulation and experiments are conducted to verify the accuracy of the theory.

Ultra-dense networks (UDNs) have been employed to solve the pressing problems in relation to the increasing demand for higher coverage and capacity of the fifth generation (5G) wireless networks. The deployment of UDNs in a very large scale has been envisioned to break the fundamental deadlocks of beyond 5G or the sixth generation (6G) networks and deliver many more orders of magnitude gains that today’s technologies achieve. However, the mathematical tool to optimize the system performance under the stringent radio resource constraints is widely recognized to be a formidable challenge. System-level performance optimization of current UDNs are usually conducted by relying on numerical simulations, which are often time-consuming and have become extremely difficult in the context of 6G with extremely high density. As such, there is an urgent need for developing a realistic mathematical model for optimizing the 6G UDNs. In this paper, we introduce challenges as well as issues that have to be thoroughly considered while deploying UDNs in realistic environment. We revisit efficient mathematical techniques including game theory and real-time optimization in the context of optimizing UDNs performance. In addition, emerging technologies which are suitable to apply in UDNs are also discussed. Some of them have already been used in UDNs with high efficiency while the others which are still under investigation are expected to boost the performance of UDNs to achieve the requirements of 6G. Importantly, for the first time, we introduce the joint optimal approach between realtime optimization and game theory (ROG) which is an effective tool to solve the optimization problems of large-scale UDNs with low complexity. Then, we describe two approaches for using ROG in UDNs. Finally, some case study of ROG are given to illustrate how to apply ROG for solving the problems of different applications in UDNs.

As research attention in deep learning has been focusing on pushing empirical results to a higher peak, remarkable progress has been made in the performance race of machine learning applications in the past years. Yet deep learning based on artificial neural networks still remains difficult to understand as it is considered as a black-box approach. A lack of understanding of deep learning networks from the theoretical perspective would not only hinder the employment of them in applications where high-stakes decisions need to be made, but also limit their future development where artificial intelligence is expected to be robust, predictable and trustable. This paper aims to provide a theoretical methodology to investigate and train deep convolutional neural networks so as to ensure convergence. A mathematical model based on matrix representations for convolutional neural networks is first formulated and an analytic layer-wise learning framework for convolutional neural networks is then proposed and tested on several common benchmarking image datasets. The case studies show a reasonable trade-off between accuracy and analytic learning, and also highlight the potential of employing the proposed layer-wise learning method in finding the appropriate number of layers in actual implementations.

As a result of the explosion of security attacks and the complexity of modern networks, machine learning (ML) has recently become the favored approach for intrusion detection systems (IDS). However, the ML approach usually faces three challenges: massive attack variants, imbalanced data issues, and appropriate data segmentation. Improper handling of the issues will significantly degrade ML performance, e.g., resulting in high false-negative and low recall rates. Despite many efforts have done in the literature, detecting security attacks in a complicated network environment with imperfect data collection is still an open issue. This work proposes a machine learning framework with a combination of a variational autoencoder and multilayer perceptron model to deal with imbalanced datasets and detect the explosion of attack variants on the Internet. The detection engine also includes an efficient range-based sequential search algorithm to address the segmentation challenge in data pre-processing from multiple sources (network packets, system/statistic logs) effectively. Our work is the first attempt to demonstrate the effect of using an appropriate combination of ML models for boosting IDS detection capability in a heterogeneous environment, where data collection imperfection is common. Experimental results on a public system log dataset (e.g., HDFS) show that our method gains approximately as much as 97% on F1 score and 98% on recall rate, a promising result compared to the same measurement of other solutions. Even better, we found that the proposed treatment of imbalanced datasets can improve up to 35% on the F1 score and 27% on recall rate. The testing results also indicate that our model can detect new attack variants.

Reconfigurable intelligent surfaces (RIS), a device made of low-cost meta-surfaces that can reflect or refract the signals in the desired manner, have the immense ability to enhance the data transmission from the sender to the receiver. The concept of RIS is inspired by a smart radio environment or programmable radio environment. The introduction of this device in wireless communications aids in reducing the hardware requirements, energy consumption, and signal processing complexity. The integration of this device with various emerging technologies such as multiple-input multiple-output (MIMO) systems, non-orthogonal multiple access (NOMA) technique, physical layer security, etc., has increased its potentiality in terms of performance enhancement. One such integration could be studied, i.e. RIS-assisted unmanned aerial vehicles (UAVs). The UAVs exhibit aiding capability in various services to our society such as real-time data collection, traffic monitoring, military operations & surveillance, medical assistance, and goods delivery. Despite the positive appeal, the UAV has its limitations such as fuel efficacy, environment disturbances, limited network capability, etc. Considering these scenarios, the RIS can provide assistance to UAVs to enhance their performance when integrated. There is a limited number of articles and researches that consider UAV-assisted RIS systems. This article provides a detailed survey on RIS-assisted UAV systems considering multiple contexts such as optimization, communication techniques, deep reinforcement learning, secrecy performance, efficiency enhancement, and the internet of things. Finally, we draw attention to the open challenges and possible future directions of UAV-assisted RIS systems in phase shifting, channel modeling, energy efficacy, and federated learning.

The Internet of Things (IoT) infrastructure requires billions of devices that must ideally be self-powered. Ambient RF and thermal energy have great potential since they are both available continuously throughout the day. An RF harvester is a rectenna that is a combination of a receiving antenna and a rectifier. Thermal energy harvesters (TEH) are typically static type, with a fixed hot source at one end and a cold source at the other. Here, we present a transient type TEH that generates energy from diurnal cycle temperature fluctuations. Smart integration is achieved by designing the antenna to also act as the heatsink for the TEH. The antenna must be optimized while considering the electromagnetic radiation as well as the heat transfer performances. Thus, two simulators, Ansys HFSS and Ansys Fluent, were employed. The antenna operates at GSM900, GSM1800, and 3G bands simultaneously, with measured gains of 3.8, 4, and 5.3 dB, respectively, which have increased by ~3–4 dB (radiation efficiency doubled from ~40% to ~80%) compared to the flat antenna (with no heatsink fins). The TEH is in the form of a square box where two identical rectennas cover the four sides. Through RF field testing, ~250 mV is consistently collected at any instance ( $6.25 {\mu }\text{W}$ for a 10 $\text{k}\Omega $ load). Without the heatsink antenna, the average power collected from the TEH is $13.6 {\mu }\text{W}$ , which increases by 2.3 times when the heatsink antenna is integrated, highlighting the utility of this co-design and monolithic integration, which enhances both RF and thermal harvested powers.

Aircraft trajectory prediction is a challenging problem in air traffic control, especially for conflict detection. Traditional trajectory predictors require a variety of inputs such as flight-plans, aircraft performance models, meteorological forecasts, etc. Many of these data are subjected to environmental uncertainties. Further, limited information about such inputs, especially the lack of aircraft tactical intent, makes trajectory prediction a challenging task. In this work, we propose a deep learning model that performs trajectory prediction by modeling and incorporating aircraft tactical intent. The proposed model adopts the encoder-decoder architecture and makes use of the convolutional layer as well as Gated Recurrent Units (GRUs). The proposed model does not require explicit information about aircraft performance and wind data. Results demonstrate that the provision of enriched aircraft intent, together with appropriate model design, could improve the prediction error up to 30% at a prediction horizon of 10 minutes (from 4.9 nautical miles to 3.4 nautical miles). The model also guarantees the mean error growth rate with increasing look-ahead time to be lower than 0.2 nautical miles per minute. In addition, the model offers a very low variance in the prediction, which satisfies the variance-standard specified by EUROCONTROL (EU Organization for Safety and Navigation of Air Traffic) for trajectory predictors. The proposed model also outperforms the state-of-the-art trajectory prediction model, where the Root Mean Square Error (RMSE) is reduced from 0.0203 to 0.0018 for latitude prediction, and from 0.0482 to 0.0021 for longitude prediction in a single prediction step of 15 seconds look-ahead. We showed that the pre-trained model on ADS-B data maintains its high performance, in terms of cross-track and along-track errors, when being validated in the Bluesky Air Traffic Simulator. The proposed model would significantly improve the performance of conflict detection systems where such trajectory prediction models are needed.