Agriculture is considered one of the most critical domains for human beings. Currently, agriculture and supply chain businesses are becoming increasingly reliant on technology and interconnectivity. However, this inevitable necessity also brings about potential risks. Due to the widespread use of new technologies to improve productivity in agriculture, cybersecurity risks have inevitably emerged. To proactively mitigate the risks related to cybersecurity, it is important to understand the attack vectors, such as malware. Thus, malware analysis is indispensable for understanding the characteristics and behaviors of malware and developing defense strategies against malicious attacks based on data collected from the results.In the literature, two different techniques have been widely used to analyze malware: Static Analysis and Dynamic Analysis. While Static Analysis aims to investigate the structure of the malicious software at a low level, Dynamic Analysis focuses on the behavior of the malware when it is being executed. As executing malware may harm the host system, an isolated environment should be established for Dynamic Analysis.This study focuses on Malware Analysis in terms of the dynamic analysis approach. Additionally, an isolated, automated dynamic analysis platform was developed, and analysis was implemented in a virtual environment using Cuckoo and VirtualBox to collect data from analyzing various malware behaviors. This data can be used to create a partial test dataset that may contribute to developing defense strategies against malware.

An Improved Energy Detector (IED) based on Nakagami-m fading channel and Maximal Ratio Combining (MRC) is presented in this paper. With the MRC, the IED is used uniformly, exponentially, and Multiple Input Multiple Output (MIMO). An optimization problem involving 𝑁 secondary users (SU) and one primary user (PU) is the subject of our research. Using Kummer’s solution for Cognitive Radio (CR), we can derive the expression for the likelihood of miss detection using p-based Energy Detection (ED). In order to enhance the detection performance, we perform threshold operations such as addition, subtraction, mean, and median. The error rate is also evaluated in terms of how well it performs. The generated expressions are tested for validity and accuracy using simulations. Combined correlated fading and static distance have a significant effect on detection performance over faulty CSI is also demonstrated in this study. Further, the impact of correlation coefficient is also examine through the simulation. We see that the antennas connected in an exponential manner perform much better than antennas connected in a uniform manner.

Ultrasonic measurements at high magnetic field and low temperature necessitate ultra-fast excitations, ultra-high dynamic range of duty-cycle adjustments, as well as real-time, precise and automatic analysis. This paper proposes a dual-mode ultrasonic measurement system, which means the capability of carrying out acoustic measurements in both pulsed and static high magnetic field with low temperature. Based on the “counter-delay-chain” series structure and adaptive jitter compensation in the field-programmable-gate-array(FPGA), the timing control accuracy of 5 nanoseconds and the timing jitter below 160 picoseconds are realized for high-resolution excitation generation and precise sampling window adjustment. The duty cycle adjustment achieves a high dynamic range from 0.23 ppb to 10 % to realize both pulsed field mode and static field mode (denoted as PF-mode and SF-mode) ultrasonic measurements. The independent sampling window adjustment reduces the data size by more than 80%, extends the maximum testing duration and reduces the analysis consuming. Combined with the automatic cross-correlation analysis method, this system can automatically carry out PF-mode tests with a repetition rate above 100 kHz and SF-mode tests for hours with sound velocity accuracy of 2%. Experiments have proved the reliability and feasibility of the system and revealed its application prospect in condensed matter physics experiments.

In recent years, the analysis of movement patterns has increasingly focused on the individuality of movements. After long speculations about weak individuality, strong individuality is now accepted, and the first situation–dependent fine structures within it are already identified. Methodologically, however, only signals of the same movements have been compared so far. The goal of this work is to detect cross-movement commonalities of individual walking, running, and handwriting patterns using data augmentation. A total of 17 healthy adults (35.8 ± 11.1 years, eight women and nine men) each performed 627.9 ± 129.0 walking strides, 962.9 ± 182.0 running strides, and 59.25 ± 1.8 handwritings. Using the conditional cycle-consistent generative adversarial network (CycleGAN), conditioned on the participant’s class, a pairwise transformation between the vertical ground reaction force during walking and running and the vertical pen pressure during handwriting was learned in the first step. In the second step, the original data of the respective movements were used to artificially generate the other movement data. In the third step, whether the artificially generated data could be correctly assigned to a person via classification using a support vector machine trained with original data of the movement was tested. The classification F1–score ranged from 46.8% for handwriting data generated from walking data to 98.9% for walking data generated from running data. Thus, cross–movement individual patterns could be identified. Therefore, the methodology presented in this study may help to enable cross–movement analysis and the artificial generation of larger amounts of data.

In the past few years, accurate wind speed forecasting has become progressively important due to the growing demand for renewable energy sources. In this study, we propose a novel artificial model, Grey Wolf Optimization (GWO)-nested Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN)-Convolutional Neural Network (CNN)-Bidirectional Long Short-Term Memory (BiLSTM), for wind speed forecasting. The proposed model integrates three powerful techniques: CEEMDAN, CNN, and BiLSTM. Firstly, CEEMDAN with two nested layers is used to decompose the wind speed time series into a set of intrinsic mode functions (IMFs) to enhance the accuracy of the model. CNN is then applied to extract features from the IMFs, and BiLSTM is used to capture the temporal dependencies and make accurate predictions. To improve the performance of the model further, we also introduce GWO and apply it to select the best hyperparameters for the forecasting models based on decomposition results. The model is tested using wind speed data collected in Basel in 2022. The final combination result demonstrates that the proposed model achieve a better performance compared to several modern AI-base models in the wind speed forecasting field, with high accuracy reflected by a mean absolute percentage error (MAPE) of nearly 3%.

Light-fidelity (Li-Fi) is a promising solution to provide high-rate, secure, and green communications to be used in the next generation of wireless networks. Since visible or infrared (IR) light-emitting diodes (LEDs) are used as the optical source and have a nonlinear transfer function, the transmitted modulated signal can be distorted if the signal has a high peak-to-average power ratio (PAPR). Recently, a new modulation scheme called Hadamard-coded modulation (HCM) is proposed which has a low PAPR and produces symbols with discrete levels. In the HCM technique, the symbol levels can be created using multiple LEDs. Therefore, each LED operates in its linear region and just switches on or off. In this paper, a low-complexity transceiver architecture for the HCM-based communication links is proposed and a complete synchronization procedure based on the spread spectrum techniques is presented. Finally, the bit error rate of the system is evaluated by Monte Carlo simulations, and effects of system parameters such as preamble length, fall and rise time of optical devices, and timing jitter on the bit error rate of the link are discussed.

The development of the networked battlespace as a concept, and the wealth of wired and wireless communications to support it have opened the door to truly multidimensional radar systems. Whether as a static network of radars coordinating to provide national coverage, or a local network for UAV's providing a swarm aperture for rapidly reconfigurable airborne radar, there is an urgent need for a flexible radar channel model to empower the engineer and scientist to develop new systems, sensing algorithms and concepts of operation. LyceanEM, an open source electromagnetics model is presented as a solution to this issue, allowing for multi source, heterogeneous antenna arrays to be simulated together with a radar target.

Research in the field of automatic human emotion recognition has shown that changes in emotions can lead to physiological responses in humans. Typically, sensors placed on the body are used to collect physiological data for automatic emotion recognition. However, recent studies have demonstrated that heart rate can be estimated from human face videos by extracting photoplethysmographic (PPG) signals based on the RGB color changes in the face during the cardiac cycle. This study proposes a novel framework for classifying human emotions using contactless PPG signals. The study evaluated PPG signal extraction methods using a popular emotional database, and for classification, a deep learning architecture combining a 1D convolutional neural network (1DCNN) and a long short-term memory (LSTM) was adopted after normalization and signal segmentation. The proposed method achieved a recognition rate of 68.9\% and 54.2\% for binary classification of valence and arousal, respectively, with a signal segmentation of 5 seconds. It is important to note that the main aim of this study is to introduce a new approach to automatic emotion recognition using deep learning techniques based on contactless sensing of physiological signals.

We derive metrics for adaptation of antenna-element spacing in nonstationary pure line-of-sight MIMO (LOS-MIMO) links. An example of such a deployment is a high-frequency and very high-speed wireless link between an indoor base-station and an untethered mixed-reality headset worn by a user whose distance from the base-station is time-variant. Due to the narrow antenna beams required to close the link budget and highly directional propagation at high frequencies, achieving high transmission spectral efficiency must rely on LOS-MIMO transmission—realization of which depends critically on the distance between transmitter and receiver, as well as on the antenna element spacing at the transmitter and receiver. Therefore, metrics that facilitate quick adaptation of the antenna element spacings in response to changes in the LOS-MIMO link length are required. We derive these metrics for MIMO orders two, three and four (i.e., number of transceiver antennas) and show how the transmission rate of the LOS-MIMO link depends on them. We also show how they can be quickly computed from simple measurements of path phase.

In Cognitive Radio (CR), accurately detecting Primary User (PU) signal through spectrum sensing techniques is the foremost requirement. For Energy Detection method, selection of appropriate threshold is expected to improve efficiency of CR. This paper presents modified threshold approach that estimates the threshold parameters by incorporating the effect of Noise Uncertainty (NU). Authors have considered L correlated branches of multiple antenna receiver under the effect of mobility for spectrum sensing through Energy Detector (ED). The performance analysis is carried over Nakagami-m fading channel with constant, exponential and arbitrary correlation between the elements of antenna. The approach of Moment Generating Function (MGF) is considered to conveniently obtain statistical values of Signal to Noise Ratio (SNR). Further, lauricella and confluent hypergeometric functions are used for obtaining closed form of expression of Probability of Detection for Maximal Ratio Combining (MRC) diversity technique. It is observed that the performance of source algorithm is significantly improved using the modified threshold parameter over entire range of SNR. The further improvement for detection performance is observed with increase in number of branches in antenna. Moreover, at high value of fading parameter (m=4) for MRC diversity, the detection probability with exponential correlation between L elements is found superior compared to constant and arbitrarily correlated branches.

Driven by the Deep Learning (DL) revolution, Artificial   intelligence (AI) has become a fundamental tool for many Bio-Medical tasks, including AI-assisted diagnosis. These include analysing and classifying images (2D and 3D), where, for some tasks, DL exhibits superhuman performance. Diagnostic imaging, however, is not the only diagnostic tool. Tabular data, such as personal data, vital signs, and genomic/blood tests, are commonly collected for every patient entering a clinical institution. However, it is rarely considered in DL pipelines, although it carries diagnostic information. The training of DL models requires large datasets, so large that every institution might need more data that should be pooled from different sites. Data pooling generates newfound concerns about data access and movement across other institutions spawning multiple dimensions, such as performance, energy efficiency, privacy, criticality, and security. Federated Learning (FL) is a cooperative learning paradigm aiming at addressing these concerns by moving models instead of data across different institutions. This paper proposes a Federated multi-input model that leverages images and tabular data, providing a proof of concept of the feasibility of multi-input FL architectures. The proposed model was evaluated on two showcases: the prognosis of CoViD-19 disease and the patients’ stratification for Alzheimer’s disease. Results show that enabling multi-input architectures in the FL framework allows for improving the performance regarding both accuracy and generalizability with respect to non-federated models while ensuring security and data protection peculiar to FL.

Recent advances in telecommunication and machine learning (ML) have allowed for new smart and autonomous vehicle applications to improve road safety, environmental conditions, and traffic management through Vehicle to Vehicle (V2V) or Vehicle to Infrastructure (V2I) communication. However, with the rise of advanced cyber-attacks, the authenticity of a message guarantees its source but not its correctness. To mitigate the new sophisticated attacks, new Misbehavior Detection Systems (MDS), that use machine learning algorithms to detect misbehaving vehicles, have been proposed. This work provides first a comprehensive review of recent developments in ML-based MDS technology within a Vehicular Ad-Hoc Network (VANET) context, covering data collection, feature selection, model training, model evaluation and deployment. We survey useful public datasets and summarize recent studies. We report useful pieces of information for every work. In particular, we highlight the considered dataset for ML training, list the selected ML models, indicate the feature selection and dimensionality reduction techniques, recapitulate the main results, report the performance metrics and mention the deployment guidelines when applicable. Then, we compare the surveyed studies discussing not only the strength points but also their limitations. One of the key observations from the surveyed works is the absence of a quantitative analysis of the proposed models’ execution time, which is a crucial performance metric considering the limited on-board and edge computing resources. To develop a feasible ML-based MDS for V2X communication, it is essential to address this issue and propose a deployment strategy that optimizes the allocated resources for this technology. However, achieving this remains a challenge. Moreover, in view of the fact that data generation and analysis are critical phases in this technology. Also, using simulation has many advantages over the real data collection, we provide a tutorial on how realizing a useful dataset collection with popular open-source tools while considering exemplar types of attacks. Last, we demonstrate through the tutorial the use of the collected dataset for ML-based MDS model selection and training. An open source github repository is provided for regenerating the whole explained scenarios and modify according to the given research issue.

We used publicly available activity recognition datasets - UCF101 and HMDB51. UCF101 - K. Soomro, A. R. Zamir, and M. Shah, “Ucf101: A dataset of 101 human actions classes from videos in the wild,” arXiv preprint arXiv:1212.0402, 2012. HMDB51 - H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre, “Hmdb: a large video database for human motion recognition,” in Proceedings of the International Conference on Computer Vision (ICCV). IEEE, 2011, pp. 2556–2563.

With the gradual exit from fossil and nuclear based power generation, a replacement for the reducing grid inertia is urgently needed. Due to this reason, grid-forming inverter technology is discussed as possible future replacement but will need to be implemented in massive scales to be an adequate source of transient grid stability. Therefore, different projects, such as ‘Grid Control 2.0’ or ‘VerbundnetzStabil’ have looked into the potential for grid integration of grid-forming inverters down to the low-voltage level to utilize as much future renewable grid integration for inertia provision as possible. However, the incorporation of grid-forming converters that lack dependable islanding detection mechanisms may significantly diminish the efficacy of current procedures employed by existing grid-following converters. Consequently, islanded grids that emerge due to faults or planned maintenance may not be reliably deactivated during a fault. This paper presents a solution for detecting islanding in grid-forming converters. The authors analyze the limitations of a commonly used active islanding detection method (slip-mode frequency shift) on a grid-following inverter, whose basic assumptions are undermined if a grid-forming system is operated in parallel and propose an adapted detection procedure for grid-forming converters. The proposed ‘-Q-f-droop’ method is shown to be capable of reliable islanding detection in both standalone and interconnected operation with existing plants by non-detection zones created with a suitable setup in simulation. The results are verified through laboratory tests using grid-following and grid-forming PHIL inverters. The developed method ensures reliable shutdown of formed island grids even with the integration of grid-forming systems into the low-voltage level.

Post-processing techniques have been shown to improve the quality of the decision stream generated by classifiers used in pattern-recognition-based myoelectric control. However, these techniques have largely been tested individually and on well-behaved, stationary data, failing to fully evaluate their trade-offs between smoothing and latency during dynamic use. Correspondingly, in this work, we survey and compare 8 different post-processing and decision stream improvement schemes in the context of continuous and dynamic class transitions: majority vote, Bayesian fusion, onset locking, outlier detection, confidence-based rejection, confidence scaling, prior adjustment, and adaptive windowing. We then propose two new temporally aware post-processing schemes that use changes in the decision and confidence streams to better reject uncertain decisions. Our decision-change informed rejection (DCIR) approach outperforms existing schemes during both steady-state and transitions based on error rates and decision stream volatility whether using conventional or deep classifiers. These results suggest that added robustness can be gained by appropriately leveraging temporal context in myoelectric control.

We develop an efficient and near-optimal solution for beamforming in multi-user multiple-input-multiple-output single-hop wireless ad-hoc interference networks. Inspired by the weighted minimum mean squared error (WMMSE) method, a classical approach to solving this problem, and the principle of algorithm unfolding, we present unfolded WMMSE (UWMMSE) for MU-MIMO. This method learns a parameterized functional transformation of key WMMSE parameters using graph neural networks (GNNs), where the channel and interference components of a wireless network constitute the underlying graph. These GNNs are trained through gradient descent on a network utility metric using multiple instances of the beamforming problem. Comprehensive experimental analyses illustrate the superiority of UWMMSE over the classical WMMSE and state-of-the-art learning-based methods in terms of performance, generalizability, and robustness.

Relative time emerges from relative frequencies. It is the phase shift in relative frequencies due to infinitesimal loss in wave energy and corresponding enlargement in the wavelengths of oscillations; which occur in any clock between relative locations due to the relativistic effects or difference in gravitational potential; result error in the reading of clock time; which is wrongly presented as time dilation.

Over the last decade, deep learning applications in biomedical research have exploded, demonstrating their ability to often outperform previous machine learning approaches in various tasks. However, training deep learning models for biomedical applications requires large amounts of data annotated by experts, whose collection is often time- and cost- prohibitive. Self-Supervised Learning (SSL) has emerged as a prominent solution for such problem, as it allows to learn powerful representations from vast unlabeled data by producing supervisory signals directly from the data. The high amount of recent works employing the self-supervised learning paradigm for the analysis of biomedical signals (biosignals) can make it difficult for researchers to have a complete picture of the current research state. Therefore, this paper aims at outlining and clarifying the state-of-the-art in the domain. The article: briefly summarizes the nature and acquisition modality of the main biosignals; introduces the self-supervised learning method, focusing on the different pretraining strategies; provides a concise overview of the works employing SSL for the analysis of different types of biosignals; provides an overall analysis of critical aspects to consider when employing SSL to biosignals, also highlighting current open challenges. The analysis of the scientific literature highlights the importance of SSL, confirming its potential to improve models’ performance and robustness, and to promote the integration of deep learning into clinical tasks.