Search

Screen summarization aims to generate concise textual descriptions that communicate the crucial contents and functionalities of a mobile user interface (UI) screen. A UI screen consists of objects with a hierarchical structure that are tightly interconnected, and each object contains multimodal data such as images, texts, and bounding boxes. Considering these characteristics, previous works encoded the absolute position of objects at the view hierarchy to extract the semantic representation of the UI screen. However, the importance of the hierarchical dependency between objects in the UI structure was overlooked. In this study, we propose an object-wise hierarchical dependency-based Transformer named OHiFormer. OHiFormer considers the objects on the UI screen as tokens in natural language processing and leverages the Transformer to capture the mutual relationships between objects. Moreover, OHiFormer includes a modified self-attention mechanism using structural relative position encoding to represent the hierarchically connected UI. Experimental results demonstrate that OHiFormer outperforms benchmark models in the BLEU 1, BLEU 2, BLEU 3, BLEU 4, ROUGE-L, and CIDEr metrics by 3.63%, 2.1%, 0.12%, 1.8%, 2.38%, and 17.58%, respectively, on the Screen Summarization dataset. Furthermore, our proposed UI structural representation method achieves remarkable performance on complex UIs with numerous objects compared to other structural position encoding methods. Finally, a visualization of the self-attention heatmaps demonstrates how OHiFormer reflects the hierarchical dependencies between objects. By reflecting hierarchical dependencies hidden in the visual layout of the UI, OHiFormer not only improves the quality of summaries but also offers the potential for applications in mobile apps and systems containing numerous interactive objects.

Recently, intelligent reflective surface (IRS)-aided systems are becoming a prospective technology in realizing for sixth generation (6G) wireless communication era because of extremely low power transmission, seamless coverage and their superiority. These network systems can allow many users and devices to connect to each other, extending the coverage. To empower IRS-aided systems, non-orthogonal multiple access (NOMA) can be leveraged to work with IRS technique enabling further benefits such as mass connectivity, flexible resource allocation and improved performance. Increasing connected devices and expanding coverage means devices have the potential to interfere with each other. Recent studies focusing on researching and analyzing the performance of the IRS-supported NOMA network have not taken into account or not fully calculated the impact of interference on system performance. In this study, we first analyze the effect of co-channel interference (CCI) at users in downlink IRS-NOMA systems. In particular, the CCIs generated by the terminals deployed randomly in the coverage area affect the signal reception at the user in the downlink. In this network model, the channel conditions that follow the Rayleigh distribution and the CCI statistical model are independent and identically distributed. We analyze and evaluate network performance by extracting closed-form expressions of outage probability, ergodic capacity, total achievable rate then highlighting the adverse effects of CCI on IRS-NOMA. In addition, to improve the performance of the IRS-NOMA downlink, we present a framework of theorical analysis to look more insights of users’ performance, i.e. diversity order. Our analytical derivatives are verified through computer simulations based on Monte-Carlo and intuitive comparisons with the benchmarks.

The rapid evolution of wireless communication has affected unmanned aerial vehicles (UAV), which are expected to be used in diverse applications in smart cities, military operations, and cellular networks. To address the significant impacts of rapid wireless communication advancements, along with the escalating demand for user equipment (UE), multiple access technique approaches, such as non-orthogonal multiple access (NOMA), have been proposed. NOMA has the key distinguishing feature of supporting more UE, particularly UAV-enabled communication networks. Moreover, the successful implementation of such enhancements relies on the acquisition of high-quality predictions. These predictions, driven by in-depth insights derived from data, are facilitated by machine learning (ML). The integration of ML further enhances UAV capabilities to pave the way for optimized wireless communication. In this paper, we present a survey on the potential of NOMA techniques applied to UAVs using ML methods to enhance UAVs in wireless communication networks. Specifically, a basic overview of UAV and NOMA will first introduced. The role of NOMA in UAV networks is then divided into two categories: the principles and application of NOMA in UAV networks. Finally, implement ML on NOMA for UAVs by representing the diverse applications of ML systems. In addition, we highlight several open research problems as possible directions for future research.

This research extensively explores contemporary techniques in Underwater Wireless Power and Data Transfer (WPDT), as found in recent literature over the past 5 years. These techniques involve Magnetic Induction (MI), acoustic communication, and optical communication. The study aims to evaluate transmission efficiency, propose solutions to overcome range limitations, facilitate integration, and precisely map multilayer networks. The research methodology includes proposed solutions to address the limitations of each transmission technique, multilayer network mapping, and model validation. Evaluation is conducted on the relative efficiency of each method, including power loss at each layer, the connection matrix between layers, and transmission speed. Simulations reveal power loss at nodes in each layer and the connection matrix between layers. The power loss at each node shows random values, providing a realistic aspect close to real-world scenarios, offering in-depth insights into the characteristics and performance of multilayer networks in an underwater context. The research series, involving literature elaboration, technological approaches, illustrations, and simulations, demonstrates the effectiveness of the proposed model. Simulation results indicate the potential of this model in real-world scenarios, with tolerable power loss values, suggesting that multilayer networks could be a solution to classic challenges in underwater power and data transmission. The hope of this preliminary study is to provide new insights and make a significant contribution to understanding and designing reliable underwater power and data transfer systems in the future.

This article introduces an innovative approach to the navigation of autonomous underwater vehicles (AUVs) in inspection and data acquisition missions within underwater wireless sensor networks (UWSNs). We combine at least five underwater navigation techniques to accomplish this mission, adapted from related works. These five techniques are employed to address at least four main challenges identified in inspection and data acquisition missions in UWSNs involving the use of AUVs, namely: communication constraints, energy usage optimization, precise navigation, and effective data acquisition. Limited simulations conducted demonstrate the reliability of the proposed model. The model successfully navigates to the target point in 3D coordinates X Y Z, assuming the launch point as d0 (10 40 40), and reaches the q-goal target point (45 45 0) within 21 seconds, with the addition of uattractive and urepulsive (magnetic beacon attraction force and repulsion force as simulation of underwater current disturbance factor). Furthermore, in the inspection and data acquisition mission in UWSN simulated as node points (o) in pink, AUV (*) in blue effectively follows the predetermined points while acquiring data, as indicated by green lines (-) within just 5 seconds, achievable by increasing the value of $\alpha $ (angle of attack) of the target node to reduce delay time. The evaluation of the experimental simulations has raised issues and future research challenges, including the development of environmental simulation challenges that can closely resemble real conditions, the measurement of energy usage effectiveness to reach each target point, and the potential development of underwater recharging techniques. Furthermore, there is a need for advanced precise navigation and the advancement of effective data acquisition techniques.

In this paper, a multi-function embedded eye tracker system (ETS) has been presented and FPGA implementation of that has been described. The image processing algorithm is based on morphological operations to detect centers of pupil (PC) and corneal glint (GC) so that gaze of the user can be estimated by establishing a mapping between obtained PC-GC vector and screen. Moreover, the proposed ETS is capable to detect pupil size which is vitally important in understanding consciousness of the user since it varies in response to light condition. Taking advantage of optimum system design and parallel signal processing on high-speed digital units of the FPGA, the processing time of 1.8 ms was achieved that allows operation of the system up to 500 fps. The results of test with $176\times120$ images demonstrate maximum vertical or horizontal error of ±1 pixel. Experimental results imply versatility of the proposed low-power and high-accuracy ETS for real-time and accurate applications such as medical diagnosis and human-computer interface.

In today’s world, the importance of the Green Internet of Things (GIoT) in the transformed sustainable smart cities cannot be overstated. For a variety of applications, the GIoT may make use of advanced machine learning (ML) methodologies. However, owing to high processing costs and privacy issues, centralized ML-based models are not a feasible option for the large data kept at a single cloud server and created by multiple devices. In such circumstances, edge-based computing may be used to increase the privacy of GIoT networks by bringing them closer to users and decentralizing them without requiring a central authority. Nonetheless, enormous amounts of data are stored in a distribution mechanism, and managing them for application purposes remains a difficulty. Hence, federated learning (FL) is one of the most promising solutions for bringing learning to end devices through edge computing without sharing private data with a central server. Therefore, the paper proposes a federated learning-enabled edge-based GIoT system, which seeks to improve the communication strategy while lowering liability in terms of energy management and data security for data transmission. The proposed model uses FL to produce feature values for data routing, which could aid in sensor training for identifying the best routes to edge servers. Furthermore, combining FL-enabled edge-based techniques simplifies security solutions while also allowing for a more efficient computing system. The experimental results show an improved performance against existing models in terms of network overhead, route interruption, energy consumption, and end-to-end delay, route interruption.

Memristive crossbar arrays have gained considerable attention from researchers to perform analog in-memory vector-matrix multiplications in machine learning accelerators with low power and constant computational time. This work introduces a comprehensive framework for co-designing the software and hardware for deep neural networks (DNN) based on memristive and memcapacitive crossbars while considering various non-idealities. The model takes into account device-level factors, including conductance variation, cycle-to-cycle variation, device-to-device variation, peripheral circuits for error/weight gradient computation, and high tolerance. The overall neural network performance is thoroughly assessed by integrating these elements into a unified DNN training process. The proposed framework is implemented using a hybrid approach with Python and PyTorch. Performance evaluation was conducted using a simplified 8-layer VGG network on a measured $128\times 128$ array with weight resolution. Remarkably, the memristive and memcapacitive crossbar arrays achieved outstanding training accuracies of 90.02% and 91.03%, respectively, for the CIFAR-10 dataset. Additionally, detailed hardware estimation for both mem-elements devices is provided, enabling meaningful comparisons with prior works.

The underground utility tunnel is a facility for installing and managing public infrastructure such as electricity, water supply, and telecommunication, which are required for managing urban life. Due to the underground structure, airflow is different compared to general buildings, so heat and smoke are not smoothly removed in the event of a fire, making it difficult to evacuate safely of occupants and enter firefighters. Therefore, it is essential to accurately detect the fire location based on the current state of fire occurrence and to remove smoke efficiently. This study derived the optimum sensor location for detecting fire and exhaust fan operation for smoke removal by considering the ventilation system in the target underground utility tunnel. In the future, these results can be used as a response manual for underground facilities in case of fire. Analysis and understanding of how fire propagates inside critical infrastructures can prevent future accidents.

Retinal implants have been developed and implanted to restore vision from outer retinal degeneration, but their performance is still limited due to the poor spatial resolution. To improve the localization of stimulation, microelectrodes in various three-dimensional (3D) shapes have been investigated. In particular, computational simulation is crucial for optimizing the performance of a novel microelectrode design before actual fabrication. However, most previous studies have assumed a uniform conductivity for the entire retina without testing the effect of electrodes placement in different layers. In this study, we used the finite element method to simulate electric fields created by 3D microelectrodes of three different designs in a retina model with a stratified conductivity profile. The three electrode designs included two conventional shapes - a conical electrode (CE) and a pillar electrode (PE); we also proposed a novel structure of pillar electrode with an insulating wall (PEIW). A quantitative comparison of these designs shows the PEIW generates a stronger and more confined electric field with the same current injection, which is preferred for high-resolution retinal prostheses. Moreover, our results demonstrate both the magnitude and the shape of potential distribution generated by a penetrating electrode depend not only on the geometry, but also substantially on the insertion depth of the electrode. Although epiretinal insertions are mainly discussed, we also compared results for subretinal insertions. The results provide valuable insights for improving the spatial resolution of retinal implants using 3D penetrating microelectrodes and highlight the importance of considering the heterogeneity of conductivities in the retina.

In a non-cooperative communication environment, automatic modulation classification (AMC) is an essential technology for analyzing signals and classifying different kinds of signal modulation before they are demodulated. Deep learning (DL)-based AMC has been proposed as an efficient method of achieving high classification performance. However, most current DL-AMC methods have limited generalization capabilities under varying noise conditions, especially at low signal-to-noise ratios (SNRs). Therefore, these methods can not be directly applied to practical systems. In this paper, we propose a threshold autoencoder denoiser convolutional neural network (TADCNN), which consists of a threshold autoencoder denoiser (TAD) and a convolutional neural network (CNN). TADs reduce noise power and clean input signals, which are then passed on to CNN for classification. The TAD network generally consists of three components: the batch normalization layer, the autoencoder, and the threshold denoise. The threshold denoise component uses an auto-learning threshold sub-network to compute thresholds automatically. According to experiments, AMC with TAD improved classification accuracy by 70% at low SNR compared with a model without a denoiser. Additionally, our model achieves an average accuracy of 66.64% on the RML2016.10A dataset, which is 6% to 18% higher than the current AMC model.

We present monolithically integrated waveguide with thin isolated layers of dielectric material, which ensure efficient emission of guided light in multiple optical windows. The proposed waveguide is fabricated by using silicon oxynitride (SiON) which functions as the core guiding layer, and silicon nitride (SiN) which facilitates effective vertical emission of guided light via optical windows. The theoretical operating principle is analyzed by Finite-Difference-Time-Domain (FDTD) simulation, and the experimentally fabricated device is demonstrated to find the expected operational trend by measuring vertical emission of light through the optical windows. We further find that the physical origin of such efficient emission stems from the Fabry-Perot resonance induced laterally at the interface between the optical windows and cladding layers, which leads to the strong emission of light at the edge of optical windows. The output efficiencies of light emission from SiN optical windows, which are fabricated in cascaded fashion on top of the SiON waveguide, are 30, 45 and 55 percent at the wavelength of 457, 532, and 637 nm, respectively. The proposed design concept can be applied to various potential applications including optical integrated circuits with vertical interconnection, optical emission in wearable devices for virtual and augmented reality, and other emerging optical sensors.

A hardware-efficient implementation of a spiking cooperative network (SCN) for a real-time event-based stereo correspondence system is presented. While fully utilizing the advantage of event data, the proposed SCN design significantly reduces the amount of hardware resources by utilizing distinct properties of the SCN, such as the repeatability of synaptic connections and operations, through physical constraints. A stereo system consisting of a field-programmable gate array (FPGA) and a pair of dynamic vision sensors (DVSs) is implemented to demonstrate the SCN design. Stereo livestreamed event data are generated from the DVSs, and the SCN is implemented on an FPGA chip to process the event data. The SCN system has four cores, each comprising an array of 32 Coincidence-Disparity units that calculate the 32-level disparity in a semi-parallel manner. The system performance was evaluated experimentally to estimate the depth of objects moving at different speeds. A rotating drum with a diameter of 8 cm was used in the test. The median relative error of the estimated depth at a rotation speed of 16.7 Hz ranged from 7.3% to 10.6%.

This paper presents an online citizen science game called “Game and Repository for Aperture Solutions and Patterns (GRASP).” GRASP is uniquely tailored to an important task of designing a small-aperture antenna array by leveraging a mathematical structure called the Golomb pattern. In this application of GRASP, an optimal aperture pattern leads to space images of desired spectral resolution observed with the least-cost antenna aperture array via interferometry imaging techniques. Finding near optimal Golomb patterns is computationally very hard. GRASP citizen scientists are tasked with finding (near-)optimal Golomb patterns by selecting cells on a grid that obeys the rules of the game, ensuring Golomb properties to be satisfied. They can also submit known Golomb solutions with their citations, display existing solutions in the GRASP repository, and employ GRASP solutions for non-redundant aperture design of the interferometry and radar applications. In this paper, we introduce the concept of Golomb patterns, and present current design and framework of our citizen science platform GRASP as a means to solve, store, and display such patterns, as well as the motivation behind our work. In particular, we discuss the application of Golomb patterns in astrophysics’ inteferometry missions for small-aperture antenna array design.

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.

Graphs are used widely to model complex systems, and detecting anomalies in a graph is an important task in the analysis of complex systems. Graph anomalies are patterns in a graph that do not conform to normal patterns expected of the attributes and/or structures of the graph. In recent years, graph neural networks (GNNs) have been studied extensively and have successfully performed difficult machine learning tasks in node classification, link prediction, and graph classification thanks to the highly expressive capability via message passing in effectively learning graph representations. To solve the graph anomaly detection problem, GNN-based methods leverage information about the graph attributes (or features) and/or structures to learn to score anomalies appropriately. In this survey, we review the recent advances made in detecting graph anomalies using GNN models. Specifically, we summarize GNN-based methods according to the graph type (i.e., static and dynamic), the anomaly type (i.e., node, edge, subgraph, and whole graph), and the network architecture (e.g., graph autoencoder, graph convolutional network). To the best of our knowledge, this survey is the first comprehensive review of graph anomaly detection methods based on GNNs.

Femoral shaft fractures are correlated with frequent morbidity and mortality. It is a major musculoskeletal disorder caused by tremendous force applied to the femur. One of the most common surgical treatments for fixation is intramedullary nailing (IM), which utilises a specially designed metal rod and screws to be implanted into the medullary canal. However, severe bowing of the femur can result in a mismatch between the IM nail and the alignment of the femur. Such mismatch is a risk factor for anterior cortical perforation off the distal femur with subtrochanteric fractures and leg length discrepancy with fractures of the femoral shaft. Therefore, accurate three-dimensional (3D) preoperative planning is mandatory to facilitate the implant’s design based on the obtained geometric data, especially for the fractured bowed femur. This paper presents an automatic 3D femur shape analysis system (3D-FSA) based on the extracted skeleton of each individual patient to provide an accurate 3D preoperative simulation. The structure of the 3D femur was generated using a set of computed tomography (CT) images. By using the maximum-minimum centre approach for skeletonization, significant geometric and topological information was captured. The proposed approach can potentially assist in implant measurements.

Integration of Device-to-Device (D2D) communication into the cellular network can greatly enhance the spectrum utilization as well as creating additional communication opportunities. D2D can be combined with cognitive radio to further enhance spectrum utilization and cellular network performance. Unlike the traditional D2D communication approach which employs only cellular spectrum, cognitive D2D can enable sensing and utilizing non-cellular spectrum opportunistically as well, thus allowing to offload cellular base station traffic to non-cellular spectrum such as WiFi, Bluetooth, or TV white-spaces. Cognitive D2D users (cDUs) must vacate channel for primary users and hand-off to another secondary channel which renders it quite challenging to meet Quality of Service (QoS) requirements of multi-class cDUs in the presence of higher primary network load. In this work, an enhanced hybrid spectrum access scheme has been developed based on non-switching spectrum hand-off for multi-class DUs, utilizing both interweave and hybrid interweave underlay spectrum access strategies. Further, lower priority cDUs with non-real-time traffic remains in the system and wait for channels to become available, rather than being dropped from the system due to lack of secondary channels. A Continuous-Time Markov Chain (CTMC) has been developed to analyze the performance of the proposed scheme. For comparison, several cases, ranging from the simple cellular network to complex cellular-cognitive-D2D with hybrid-spectrum-access, have been analyzed. The main focus of the analysis is to compare the efficacy of enhanced hybrid spectrum access scheme with individual interweave and hybrid interweave underlay spectrum access strategies in terms of QoS provisioning for multi-class cDUs. The results depict improvement in throughput, spectrum utilization, and extended data delivery time for the proposed scheme and validate the suitability of the proposed scheme to meet QoS requirements for both delay-sensitive and delay-tolerant users of the multi-class cognitive D2D communication system.

In this paper, a hybrid satellite-terrestrial spectrum sharing system allows terrestrial secondary network to cooperate with a primary satellite network and to further provide higher spectrum efficiency. For massive connections design, we implement non-orthogonal multiple access (NOMA) technique to form cognitive radio based satellite-terrestrial (CR-NSHT) system relying on NOMA and further achieve more benefits compared with traditional schemes. The secondary network only remains its stable operation when the outage probability of such system is guaranteed, and thereby, to explore advantages of spectrum sharing opportunities. Considering Shadowed-Rician fading for satellite links, and Nakagami- $m$ as well as Rician fading for terrestrial links, we derive the closed-form expressions of the outage probability to evaluate the performance of secondary network. We consider further system performance metrics including ergodic capacity, energy efficiency and multi-user scenarios. We also demonstrate the impacts of power allocation factors, transmit signal-to-noise ratio (SNR) at the source, parameters of satellite links, target rates, and fading severity parameters on the system performance. Numerical and simulation results validate our analysis and highlight the performance gains of the proposed schemes for CR-NSHT with relay link serving secondary network and direct link serving the primary network.

4H-SiC is a wide-bandgap material that exhibits excellent high-temperature conductivity and high operating voltage. These characteristics can provide high electrostatic discharge (ESD) robustness in high voltage applications. However, a considerably wide range of snapback phenomena is triggered for 4H-SiC-based ESD protection devices owing to a high critical electric field. In this study, an ESD protection device based on a lateral insulated-gate bipolar transistor (LIGBT) with a new structure that creates an internal silicon-controlled rectifier (SCR) path is proposed. The proposed ESD protection device minimizes the effective base region of the NPN parasitic bipolar transistor to the gate length based on the internal SCR operation of the LIGBT. It also adjusts the emitter injection efficiency of the PNP parasitic bipolar transistor by introducing a segment topology and inserting an additional implant area. Consequently, the proposed ESD protection device significantly improves the wide range of snapback phenomena occurring in the 4H-SiC materials. A conventional SCR, the LIGBT, and the proposed protection device were fabricated using the 4H-SiC process under the same condition, and their electrical characteristics were comparatively analyzed using a transmission-line pulse system. Moreover, its high-temperature reliability was evaluated at 300–500 K to examine the compatibility with 4H-SiC devices and circuits that require a relatively high-temperature operation.

High-resolution (HR) 360° images offer great advantages wherever an omnidirectional view is necessary such as in autonomous robot systems and virtual reality (VR) applications. One or more 360° images in adjacent views can be utilized to significantly improve the resolution of a target 360° image. In this paper, we propose an efficient reference-based 360° image super-resolution (RefSR) technique to exploit a wide field of view (FoV) among adjacent 360° cameras. Effective exploitation of spatial correlation is critical to achieving high quality even though the distortion inherent in the equi-rectangular projection (ERP) is a nontrivial problem. Accordingly, we develop a long-range 360 disparity estimator (DE360) to overcome a large and distorted disparity, particularly near the poles. Latitude-aware convolution (LatConv) is designed to generate more robust features to circumvent the distortion and keep the image quality. We also develop synthetic 360° image datasets and introduce a synthetic-to-real learning scheme that transfers knowledge learned from synthetic 360° images to a deep neural network conducting super-resolution (SR) of camera-captured images. The proposed network can learn useful features in the ERP-domain using a sufficient number of synthetic samples. The network is then adapted to camera-captured images through the transfer layer with a limited number of real-world datasets.

Demand response (DR) is a voluntary program that encourages related stakeholders, in this case electricity consumers, to cut down on usage during periods of high electricity load. One key to fully exploiting DR is to encourage residential customers to join the DR program. Unlike in the DR programs for commercial and industrial customers, for the successful operation of the residential DR program, several issues have to be addressed, one of which is to provide a group-level incentive to participating customers. In particular, the issue comes up when the incentive calculated for a group is not equal to the aggregated incentives for each customer (i.e., non-equal incentive problem). The non-equal incentive problem deteriorates the successful operation of residential DR by decreasing the motivation of DR operators and customers. We first prove the non-equal incentive problem through mathematical and experimental methods. We then propose the novel single group-based indirect incentive calculation method. The basic idea of our approach is to indirectly calculate the incentive for each customer not using the customer’s data but using the data of other customers of the same DR group. Through experiments involving the electricity usage data of 42,193 households and the real DR events in Korea, we show that our method solves the non-equal incentive problem in most cases. Furthermore, our method improves the accuracy of the baseline estimation (used for calculating the contribution).

In realizing unsupervised pixel-precise anomaly localization by utilizing a generative model, a reference image must be generated (for comparison with an input image) by transforming abnormal patterns of an input image, if any, into normal patterns. In this study, a patch-level operation with adaptive patch control is proposed to improve anomaly localization by generating a better reference image. As a way to exploit a generative model, we divide an image into non-overlapped patches of the same size, generate patch-level reference images, and stitch the patch-level reference images into a single reference image. We then conduct anomaly localization by comparing an input image with the stitched, reconstructed image. To effectively apply the patch-level operation, we propose adaptive patch control to determine the number of non-overlapped patches to be applied. For this, we synthesize defective images using normal images and examine how well the candidate methods with different numbers of patches remove the synthesized defects. In the same way, we utilize adaptive patch control to select a promising model among the candidate generative models. Based on experiments conducted using the MVTec Anomaly Detection dataset, we demonstrate that our method outperforms previous existing methods. Under a real-world scenario, our method shows ROC AUC of 0.926, in contrast to the best value of 0.893 from existing studies. Furthermore, we prove the feasibility of the adaptive patch control by showing that the removal of the synthesized defects and the anomaly localization for real defective images are highly correlated.

Intelligent Transport Systems (ITS) require accurate information to be shared among vehicles and infrastructure nodes for applications including accident information or pre-crash warnings, to name a few. Due to its sensitive nature, ITS applications are vulnerable against data integrity attacks where nodes transmit false information that results in wrong decision making by the applications. A characteristic of such attacks is that the false transmitted information is significantly different than the actual information. In this paper, we propose an Outlier Detection, Prioritization and Verification (ODPV) protocol that efficiently isolates false data and improves traffic management decisions. ODPV uses the isolation forest algorithm to detect outliers, fuzzy logic to prioritize outliers and C-V2X communications to verify the outliers. Extensive simulation results verify the effectiveness of the proposed protocol to isolate the outliers.

In conventional cognitive radio (CR), the users in secondary network (SN) can only access the idle spectrum when users in primary network (PN) are absent. This novel strategy provides higher spectrum efficiency when detecting the presence of the PN. Hence, spectrum utilization of the traditional scheme can be further improved as exploiting application of non-orthogonal multiple access (NOMA). As combination of CR and NOMA, such CR-NOMA has been proposed to improve spectrum efficiency to adapt to requirements in 5G communications. In this study, the relaying scheme is employed in the SN of the proposed CR-NOMA and the relay is allowed to energy harvesting (EH) from the secondary transmitter to serve signal forwarding to distant secondary users. With this regard, the complex model of EH-assisted CR-NOMA is explored in outage behavior and throughput performance as awareness on imperfect successive interference cancellation (SIC) at the receiver. As most important results, the exact closed-form of the exact outage probability is derived for each NOMA destination by assuming that the channel coefficients among considered links follow Rayleigh distribution. Furthermore, performance gap between two NOMA users can be controlled by various parameters such as transmit power, energy harvesting coefficients and levels of imperfect SIC. Simulation results verify our analytical results.

In this article, a conceptual approach is presented for pre-fault detection regarding the waveform analytic aspects of distribution monitoring and measuring devices. Also included are event patterns specifically arranged by feature and classification methods. A waveform class patterning algorithm on time-series is applied experimentally to field waveforms that were obtained for several years. The waveforms are processed with consideration of waveform classification and event sequence processing because these detect fault-related phenomena. This approach demonstrates that conspicuous patterns in fault-related sequences can be discovered by the data-driven structure described in the paper by designing the event structure depending on the network configuration. The application is applied in a pattern-learning and pattern-detection process so that integrating both approaches provides a meaningful consolidation for detecting abnormal conditions on distribution lines. This event-based fault prevention is employed using actual acquisition data from a domestic-scale distribution system and a unique sequence model is constructed to determine normal and abnormal conditions. Event index manipulation analysis on different risk levels defines the pattern and its impact on monitoring results. The proposed model guides recognition of event patterns and waveforms that can be pre-emptively detected in advance of distribution line failures.

The future of wireless communications looks exciting with the potential new use cases and challenging requirements of future 6th generation (6G) wireless networks. Since the traditional wireless communications, the propagation medium has been perceived as a randomly behaving entity between the transmitter and the receiver, which degrades the quality of the received signal due to the uncontrollable interactions of the transmitted radio waves with the surrounding objects. The recent advent of reconfigurable intelligent surfaces (RIS) in wireless communications enables, on the other hand, network operators to control the radio waves (the scattering, reflection, and refraction characteristics) to eliminate the negative effects of natural wireless propagation. Recent results have revealed that non-orthogonal multiple access (NOMA) benefits from RIS mechanism which can effectively provide effective transmissions. Motivated by the potential of these emerging technologies, we study the impact of hardware impairment in RIS-aided NOMA system in term of performance metrics. We then derive analytical expressions of outage probability and throughput as main performance metrics. Simulations are conducted to validate the analytical expressions. We find that the number of meta-surfaces in RIS, transmit power at the base station, power allocation factors play important role to demonstrate improvement in system performance of RIS relying on NOMA compared with orthogonal multiple access (OMA). Numerical results are presented to validate the effectiveness of the proposed RIS-aided NOMA transmission strategies.

To improve the utilization of spectrum and system capacity, non-orthogonal multiple access (NOMA) is considered as a promising multiple access method for emerging communication technologies. Its advantageous benefits are implemented in two-way wireless networks. Power allocation factors assigned to multiple users play a key role in the successful deployment of fairness balance in NOMA. This paper considers fixed power allocation, and performance degradation is predicted in worse situation in which non-optimal power scheme and the imperfect channel state information (CSI) happen. More challenging issue happens in this system, and this paper addresses the situation of NOMA supporting multi-pair of users, in which group of users can communicate to other users to achieve acceptable throughout in two kinds of mode, i.e. delay-limited mode and delay-tolerant mode. Two performance metrics have been considered, namely, ergodic rate and outage probability. For both metrics, we derived closed-form analytical expressions as well as asymptotic performance. Numerical results show that the proposed system presents benefits to cellular networks by extending ability to serve more users who have different demands on data communication. Monte Carlo simulations are provided throughout to validate the accuracy of the derived analytical expressions.

Dual-directional silicon-controlled rectifiers (DDSCRs), which provide both positive and negative electrostatic discharge (ESD) surge paths, are ESD protection devices with an excellent area efficiency. However, DDSCRs have a low holding voltage for use in 5 V-class applications, with a relatively high on-state resistance because of the elongated ESD surge path compared to unidirectional SCRs. In this paper, we propose a novel DDSCR with a higher holding voltage and a better ESD tolerance than conventional low-triggering DDSCRs (LTDDSCRs), realized by operating two additional parasitic bipolar transistors. The proposed ESD protection device was developed through a 0.18-μm CMOS process, and a timeline pulse system was used to verify its properties. The measurement results show that the proposed ESD protection device exhibits an improved tolerance and a high holding voltage and is expected to be reliable in 5 V-class applications.

Mobile low-duty-cycle wireless sensor network (MLDC-WSN) is a new type of WSN that has emerged in recent years. It can deploy nodes on moving objects and use the object mobility to capture a wide range of dynamic environmental information. It has diverse applications and has the ability of mobility and gets into sleep for a long time. However, the mobility and sleeping features of MLDC-WSN nodes result in dynamic network topology changes, which restricts discovering their neighbor nodes quickly. Thus, the optimal distribution decision is not established by the nodes. To solve this problem, in this paper, a novel selectively proactive neighbor discovery algorithm is proposed. The proposed algorithm enables the nodes when they wake up to search their neighbor nodes, which prevent the long-time waiting delay in the traditional passive neighbor discovery mechanism. Moreover, the proposed algorithm can further reduce the delay and acquire the accurate neighbor discovery results by quickly determining the next neighbor set at the next moment due to the prediction of the movement speed and distance of neighbors. The analytical and simulation results show that the proposed algorithm can effectively reduce the network energy consumption and delay and can determine all the neighbor nodes quickly and accurately as compared with the existing algorithms.

The network onboarding of the Internet of Things (IoT) devices is a practically important issue to realize the IoT networks. The lack of appropriate and easy network onboarding schemes hinders the spread of the IoT networks and their related cloud services. To try and address this problem, various IoT network onboarding schemes have been developed, such as near-field communication, quick response code, ultrasound, and soft access point. In addition to these schemes, a network onboarding scheme based on Wi-Fi neighbor awareness networking (NAN), a recently developed low-power Wi-Fi technology, has received a lot of attention due to its low cost and ease-of-use characteristics. However, the radio frequency characteristics of the NAN-based scheme can cause various problems, including the unintended connection from neighboring devices. The development of the supporting technologies for the NAN-based network onboarding scheme is critically important. In this paper, we propose a convergence configurator detection scheme based on the time difference of arrival, received signal strength indicator, and data rate to reduce the neighboring issue of the NAN-based setup. Rather than using one parameter criterion, we use a weighted sum of various parameters to increase the reliability of the NAN-based onboarding. The numerical results validate the effectiveness of the proposed technique.

This study presents an automatic sleep-stage classification system based on utilizing compressive sensing (CS) for data reduction. The amount of electroencephalogram (EEG) signal data required for sleep-stage classification can be significantly reduced by applying CS at the cost of distortion in the reconstructed EEG signal. A neural network trained on features extracted from the reconstructed EEG signal was implemented as a classifier to avoid compromising the accuracy of classification in the presence of distortion in the reconstructed EEG signal. A radial basis function (RBF) neural network that uses a simple Manhattan distance function instead of complicated computation, such as vector multiplication matrix (VMM), was selected to reduce the hardware complexity. A classification method that utilizes information from the previous sleep stage based on the unique nature of human sleep was also presented and implemented on the proposed RBF classifier for further improvement of the classification accuracy. The classification system has been evaluated using EEG data from eight subjects in the Cyclic Alternating Pattern (CAP) sleep dataset. The measurement results showed that the classification system achieved high classification accuracy comparable to the previously reported sleep-stage classifiers that do not utilize signal compression.

We investigate the feasibility of improving the energy efficiency (EE) of massive multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems applied to a battery-limited Internet of Things (IoT) networks. Improving EE is especially important for battery limited IoT devices. We observe the uplink and downlink aspects of massive MIMO-OFDM-based IoT networks and categorize some of the effective methods to consider. As uplink aspect, we consider the uplink reference signal (RS) power control. Reducing uplink RS power could induce the battery saving of IoT devices but could cause an increase in channel estimation error. As downlink aspect, we consider the peak-to-average power ratio reduction of the OFDM signal and downlink transmitter power control. These techniques are well-known as effective EE improvement methods, but there is little work showing the actual EE gain in system perspective. In addition, we also consider the utilization of radio frequency energy transfer using unmanned aerial vehicles to extend the operating time of battery-limited IoT devices. We derive the theoretical closed-form approximations of spectral efficiency and EE when applying these methods and provide EE gains for various scenarios. Numerical results show that the theoretical analysis is in good agreement with the simulation results and thus can be used as useful tools to improve EE.

Media content sharing scheme among multiple devices is one of the most important service technologies in the Internet of Things (IoT) era. Typically, there are three kinds of schemes adopted for the content sharing among devices: screen mirroring, file casting, and content redirection. Among them, the content redirection in which a sink pulls the content from a cloud server based on universal resource locator information received from a source is a very promising technology in terms of resource usage. However, for the content redirection scheme, the link quality between the access point (AP) and sink must be reliable enough to share the high-quality multimedia services. The link quality problem usually occurs because AP and sink are installed without considering the link quality between the two devices. To solve this problem, in this paper, we introduce a multi-point content delivery scheme, which transfers the media content from both source and AP to sink simultaneously. The proposed scheme can operate in two modes, namely power combining mode that increases the received signal power by combining the transmitted power of both devices and bandwidth aggregation mode that increases the bandwidth of the received signal by aggregating the bandwidth of the transmitted signal of both devices. Numerical results show that using the proposed scheme, with 10 m of AP and sink separation distance, we can get 300% ~ 400% performance improvement; therefore, the proposed scheme can be a core technology for media content sharing in the IoT era.

Active matrix organic light emitting diode (AMOLED) displays are widely used in the displays of mobile devices. Whereas the display power accounts for a large portion of the total power of a mobile device, the computing power and battery size are limited. Thus, technologies for minimizing the computing overheads while reducing the power during video processing are important in mobile devices. In this paper, we propose a novel dynamic voltage scaling (DVS) method called ESC-DVS by designing low-overhead entropy-based scene change detection for video playback on mobile AMOLED displays. Its implementation using an MPEG4/H.264 player on an embedded board with a real mobile AMOLED display shows a power saving of 21.31% on average and higher accuracy compared to an existing power-saving scene change detection technique. In addition, the computing overhead decreases to 1/84 compared to an existing DVS technique while good human visual satisfaction is achieved.