As greenhouse gas emissions increasingly become a central issue in global sustainability, it is imperative to develop solutions that effectively reduce avoidable emissions. One significant contributor to these emissions is the high percentage of empty trucks on the road at any given time. This paper aims to investigate a full truckload marketplace, specifically designed to utilize the empty cargo space of trucks, thereby minimizing empty backhauls by facilitating container transportation between two terminals. The study examines two levels of planning: operational and tactical. At the operational level, we present a lower bound for the marketplace's costs and propose an online just-in-time policy. This policy involves the marketplace's decision to accept or reject a truck offer, aiming to synchronize the incoming rate of truck offers with that of shipment requests, ultimately maximizing profitability. Moving to the tactical planning problem, our objective is to identify optimal shipping prices and truck payments that maximize the objective function of the marketplace. Two scenarios determine this objective function: the marketplace can operate as a profit-maximizing firm or as a social planner. In the latter scenario, the goal is to maximize the volume of transportation services provided, consequently minimizing the additional non-productive miles driven by empty trucks. To address the tactical planning problem, we propose algorithms that leverage the problem structure. Furthermore, we conduct simulation experiments for both planning levels. The results reveal several key findings, including a pooling effect triggered by higher rates of drivers logging into the marketplace. Additionally, we observe an increase in optimal truck costs and shipping prices in response to an escalation in the delay penalty. Furthermore, we demonstrate the greater social value generated by the social planner approach. Overall, the experimental results offer valuable insights and practical guidelines for decision-makers regarding market design and pricing decisions.

Generative Adversarial Networks (GANs) are increasingly becoming essential tools for improving the perception of autonomous vehicles in adverse environmental conditions, like low light, fog, rain, and snow. This survey discusses recent trends in applying GANs to tackle these problems and classifies the available work based on environmental conditions and GAN architectures. Studies showcase the impact of different GAN models such as CycleGAN, Super-Resolution GANs, and DeblurGAN in enhancing image quality, visibility, and detection accuracy of objects under challenging conditions. For instance, CycleGAN has been remarkably effective in nighttime image enhancement; DeblurGANs significantly reduce motion blur at capture. Furthermore, multi-modal fusion approaches have been applied very successfully with RGB and infrared (IR) data creating strong perception systems that work well in diverse operational conditions. However, practically deployed driving solutions must process information quickly while remaining computationally efficient. This survey emphasizes how GAN technology has made autonomous vehicle research more adaptable and safer; it also points out potential graduate study directions involving lighter-weight models of GANs or domain-adaptive architectures to make them more applicable. We hope our comprehensive review will help researchers utilize GAN technology to enhance perception for autonomous vehicles.

The next generation of wireless communication networks is expected to utilise unmanned aerial vehicles (UAVs) and reconfigurable intelligent surfaces (RISs) to enhance spectrum and energy efficiency. This work establishes a theoretical foundation for RIS-assisted UAV implementation, capitalising on the passive beamforming capabilities of RIS alongside the adaptable deployment and dynamic mobility of UAVs to enhance internet-of-things (IoT) networks performance. A comprehensive framework for RIS-assisted UAV IoT data collection is represented and optimised to enhance critical performance metrics, including the quantity of served IoT devices and achievable data rates. The optimisation strategy deploys a deep reinforcement learning (DRL) algorithm to fine-tune UAV trajectories and IoT device scheduling decisions, complemented by a codebook for RIS beamforming to optimise the RIS phase shift matrix. This integrated approach addresses the ever-increasing demand for efficient data collection in wireless IoT networks. Simulation results show substantial improvements in system performance, demonstrating the efficiency of the proposed algorithm. By coordinating the RIS phase shift matrix and UAV trajectory planning, the proposed framework achieves improvements in terms of the number of served IoT devices and achievable data rates. For example, compared to baseline methods, our approach outperforms benchmark scenarios by over 50% in terms of the number of served devices. The results reveal the potential of RIS-assisted UAV solutions in meeting the increasing demands of wireless IoT networks.

The ongoing evolution of cyber threats has demanded increasingly sophisticated detection methods to counteract ransomware, which continues to impose severe risks on global systems. The Behavioral Entropy Detection Protocol (BEDP) introduces a real-time ransomware detection framework that leverages entropy-based metrics and behavioral analysis to identify ransomware activities before significant harm occurs. BEDP's design incorporates a dual-layered architecture, combining entropy thresholding with machine learning classifications to detect ransomware-linked file manipulations and encryption patterns, enabling it to effectively handle both known and unknown ransomware variants without reliance on pre-existing signatures. Extensive testing revealed BEDP's high accuracy and low false positive rate across various datasets, achieving significant reductions in computational resource usage while ensuring scalable performance in high-frequency file environments. The protocol's hierarchical resource allocation further optimizes entropy calculations, demonstrating low latency and reliability in real-time applications. Findings indicate BEDP's robust adaptability within enterprise-level cybersecurity infrastructures, where rapid detection of ransomware is essential, showing its suitability for deployment in both decentralized and large-scale networks. BEDP's innovative use of entropy metrics presents an adaptable, efficient approach to ransomware defense, establishing a foundation for the enhancement of proactive security measures.

Transcranial Magnetic Stimulation (TMS) efficacy relies on accurate determination of the resting motor threshold (RMT), which is currently determined manually. While previous studies have identified various anatomical and physiological factors influencing RMT, their analyses remained primarily descriptive. This study introduces a machine learning approach to predict RMT levels using anatomical MRI, EEG, and simulated E-field features. We analyzed data from 17 subjects with mild-to-moderate traumatic brain injury, implementing a three-phase analytical approach: independent analysis of feature groups, evaluation of feature importance, and development of a final predictive model. Our results confirmed previously identified RMT descriptors (gray matter percentage, brain-scalp distance, maximum E-field intensity) and revealed novel anatomical factors, including volume percentages of skull, white matter, brain, and skin. EEG analysis showed that beta band activity and delta band connectivity at specific channels were predictive of RMT levels. The optimal model, an SVM with radial basis function kernel, achieved the highest stability in crossvalidation and 100% balanced accuracy score on the test set. This work represents the first machine learning model for RMT prediction, contributing to the development of more efficient, personalized TMS protocols.

We present the design of a 9.6 µm pitch eventbased pixel with a configurable antagonistic center surround. Using this pixel with a customised address-event representation encoding system, we implement a dual-line event-based sensor with a resolution of 512x2. In simulation, our sensor achieves a dynamic range of >120 dB and operates at <250 mW, surpassing modern commercial line sensors. The center surround network implements a configurable spatial high pass filter, mitigating output redundancy while maintaining 12.00 µs event latency. This novel sensor will enable faster and more data efficient industrial machine vision, with the configurable center surround helping filter out problematic lighting fluctuations.

This review explores the potential of telepsychiatry (TP) and artificial intelligence (AI) in addressing mental health issues in India, showing the opportunities and challenges of implementing digital health solutions to address the indirect effects of the COVID-19 pandemic on mental health. The paper analyzes the system of VOS viewer, co-citation maps, and keyword hotspots. 2019 led to the widespread introduction of lockdowns and social separation as public health measures. These actions, however, also carry hazards to mental health, such as a rise in loneliness and a decrease in access to psychiatric care. Methods to close this gap include telepsychiatry (TP) and other advanced technologies that enable access to care and remote mental health support. The combination of technology such as virtual reality, artificial intelligence, and mobile apps with TP has great promise for improving mental health treatments. Using these digital technologies could transform the way mental healthcare is delivered and provide improved support for patients, families, and healthcare providers as the world works through the COVID-19 pandemic’s issues.

Extending the perception of cars and other users to prevent potential accidents is one of the main functions of the Cooperative Intelligent Transportation System (C-ITS). Critical message exchanges, which are frequently resource-intensive, between vehicular nodes constitute the foundation of such an extended perspective. Therefore, it is necessary to distribute this information effectively so that road safety application criteria are met and resource usage is maximized. To do this, the communication must be context-aware and capable of dynamically adjusting communication parameters to shifting traffic conditions while taking application requirements, the accessibility of radio resources and communication technologies, and external circumstances (such as road layout, traffic density, and the presence of roadside units) into account. Our goal in this research is to examine the current level of context-aware communication for applications related to road safety. First, we go over current research on context awareness in the context of the C-ITS domain. Subsequently, we suggest grouping contexts into three categories: application requirements, resource availability, and environmental considerations. Finally, we address several open research directions to be taken into consideration in future work and shed light on context-aware architecture that should be used to alter communication parameters.

Various applications in maritime scenarios, including offshore operation, navigation, and search and rescue, require reliable high-speed communication. However, wireless maritime communication remains challenging due to limited available radio frequency (RF) bandwidth. Additional challenges are imposed by ducting and water surface reflections that lead to high packet loss rates. This work aims to provide a freespace optical (FSO) communication-based solution to address the limited bandwidth issues. An unlicensed 6 GHz RF link is used as a backup for the FSO link during outage events. The hybrid link deployment was conducted on an island located 19 km offshore at the Red Sea shore. The location was chosen based on the need for construction operations led by Red Sea Global (RSG). In addition, the deployment area is considered a challenging environment for FSO due to its high temperatures, humidity, wind speed, and sand/dust storm events. Our findings indicate that the FSO solution successfully achieves a 20 Gbps full duplex with an availability of 98.71%. The overall connectivity remained at 99.999% with the RF as a backup mode. These results highlight the potential of using FSO for high-speed maritime communication and thus connecting remote and unconnected areas.

An important part of training simple neural networks for students and researchers alike is choosing an adequate activation function. Activation functions serve a number of important purposes, including but not limited to: preventing dead neurons, eliminating the vanishing gradient problem, and allowing networks to learn complex patterns. Currently, students are taught to use the "guess-and-check" method in choosing the best activation functions for their neural networks. Although literature and knowns exist for choosing activation functions for specific tasks, not all literature applies. The purpose of this presentation is to report an algorithm that can help choose the best activation function by eliminating human-handed checking and increasing time efficiency.

This research investigates predictive maintenance strategies using the AI4I 2020 dataset, a synthetic dataset reflecting real-world industrial environments. The study identifies failure trends, explores their root causes, and evaluates cost impacts associated with different failure types. Advanced analytics reveal heat dissipation failures (HDF) as the most significant contributor to operational disruptions, accounting for over 74% of total maintenance costs. Through a detailed examination of machine operational parameters and failure distribution patterns, this work proposes tailored maintenance solutions, such as enhanced cooling systems, load optimization, and power stabilizers, to mitigate failures effectively. The findings underscore the importance of prioritizing high-impact failures, leveraging predictive maintenance, and adopting proactive strategies to minimize downtime and optimize costs in industrial settings. This research serves as a comprehensive guide for industries aiming to enhance operational reliability and sustainability.

If touching is not a single perception but a plural, then its objects are a plurality, too." (Aristotle) For investigating steering feel, a very important part is how to measure what people feel. The hypothesis in the present research work is that steering feel, as perceived by human beings, can be allotted in dimensions. The steering feel of a certain vehicle can then be seen as a point in a space with multiple orthogonal dimensions. The aim is to find the dimensions that people use to perceive and describe steering feel, in order to define this non-instrumental space. This article describes two modes of evaluation, a manual mode and a mathematical mode using the statistical method of multidimensional scaling. Applying these modes, it has been possible to extract nine dimensions describing the steering feel of road vehicles.

Suspended microstrip antenna (SMSA) suspended in air offers higher gain and wider bandwidth (BW). However, cross-polarization level (CPL) increases and impedance becomes inductive resulting in mismatch due to increased probe feed length. Slots in SMSA (E-SMSA) decrease the inductive reactance, increase the capacitive reactance and thus improve impedance BW. Orthogonal surface currents in opposite directions around two rectangular slots in E-SMSA cancel the cross-polar radiations and reduce CPL. Meta-surfaces consisting of an array of square patches of dimensions and periodicity <0.1λ0, fabricated around slotted SMSA and on the other side of the substrate, around a rectangular parasitic patch (PP) designed on a superstrate and placed above slotted SMSA, increase the inductance of structure. To decrease the inductive reactance, substrate height is decreased which decreases CPL and improves radiation characteristics. The optimized structure consisting of SMSA with slots, PP and meta-surfaces offers 8.8dBi peak gain with side lobe level (SLL) and CPL 20dB over the entire 3.3-3.8 GHz, sub-6GHz band used for 5G applications. The measured results of the prototype structure agree closely with the simulation results.

The Indian military healthcare system plays a crucial role in safeguarding the health and well-being of those who serve in the military, including active personnel, retired veterans, and their dependents. This extensive network of hospitals and medical facilities spans across all branches of the armed forces—Army, Navy, and Air Force—providing specialized medical care. Despite its broad reach, the system faces challenges, particularly within Outpatient Departments (OPDs), which handle a substantial volume of daily visits. This workload tends to spike during health crises, such as pandemics. Managing patient flow effectively remains a persistent challenge for military hospitals. Traditional queuing strategies, such as the first-come, first-served and token-based systems, fall short of handling the complexity of patient prioritization, doctor efficiency, and real-time workload adjustments. These methods often lead to inefficiencies like extended waiting periods, uneven workload distribution among medical personnel, and suboptimal resource use. The manual system’s inability to adapt to changing patient demands and the urgency of medical needs further exacerbates these issues. There is an urgent need for improved queue management solutions in military hospital OPDs. A dynamic queuing algorithm that can adapt to real-time data—such as patient arrivals, doctor availability, and consultation times—could significantly enhance patient flow, reduce waiting times, and improve overall system efficiency. By implementing a flexible queuing approach, military hospitals can better address the needs of their beneficiaries, thereby enhancing both operational outcomes and patient satisfaction.

This study presents two main aims: (i) to assess functionally-calibrated musculoskeletal models (FCMs) in both tracking and predictive simulations of human motion, against non-linearly scaled models (NSMs), and (ii) to examine the effects of three different variations of our baseline functional calibration approach on the results of tracking and predictive simulations. Methods: Motion capture experiments of six functional activities were performed with three healthy subjects. The musculotendon (MT) parameters of 18 muscles per leg were estimated using an optimal control problem. A baseline problem formulation and three variations were developed to generate four different FCMs per subject. Then, the FCMs were compared against NSMs in tracking simulations of the motions excluded from the calibration and fully-predictive simulations of gait. Results: In the tracking simulations, the FCMs led to more accurate joint torques estimations. Including gait in the calibration problems improved the knee torques accuracy (normalised root mean square error: 0.31+/-0.11), compared to the baseline calibration (normalised root mean square error: 0.70+/-0.21). Regarding the gait predictive simulations, the NSMs consistently yielded more accurate subtalar inversion/eversion torques and knee flexion angles, compared to the FCMs. The accuracy of the predicted muscle excitations was generally consistent between NSMs and FCMs. Conclusion: The results suggest that, while the FCMs led to more accurate joint torques estimations in the tracking simulations, they did not outperform the NSMs in the fully-predictive gait simulations.

Despite advancements in deep learning, conversational translation still faces challenges, including ambiguity in short sentences and limited data for colloquial language, as well as the need for culturally-situated and commonsense-based translations. However, Large Language Models (LLMs), using their vast background knowledge, can bring about a fundamental change in conversational translation. Therefore, in this paper, we demonstrate that LLMs, with appropriate prompts, could potentially solve most challenges identified in the last two decades of conversational translation research.

In this work, a high-bandwidth very-low input capacitance original topology buffer was designed. The proposed buffer circuitry is conceived as part of a novel non-invasive non-immersed real-time dielectric spectroscopy biomass sensor. Both a circuital-based simulation model and an experimental PCB prototype were implemented and successfully tested. The buffer effective parasitic input capacitance was lowered to a 2% of its original value by means of power supply bootstrapping. This improvement will allow performing differential voltage measurements with the required accuracy and without compromising the necessary bandwidth.

Data availability can limit machine learning model generalizability, while privacy concerns arise when sharing data for collaborative learning. Additionally, limited device capabilities and increased communication power can hinder real-world deployment. Differential-Private-Federated-Learning (DPFL) addresses these challenges by sharing encrypted gradients to enhance privacy and minimize data vulnerability while supporting collaborative model training in privacy-sensitive environments. The integration of DPFL with low-power FP-GAs is beneficial in overcoming edge-device limitations. FP-GAs' parallel processing, reconfigurability, quantization support, and hardware-software co-design, enhance DPFL's efficiency and adaptability. This combined approach optimizes training, ensuring secure and efficient collaboration across applications. Importantly, as ML algorithms evolve, FPGA implementations promote sustainability by reducing energy consumption, extending the hardware lifespan, and reducing electronic waste. Moreover, this framework can be used for emulation on FPGAs to explore and develop novel ASIC architectures for DPFL machine learning accelerators. We propose a novel open-source framework-"DPFL-FPGA-Accel" for carrying out design-space exploration of FPGA-based DPFL, source code is made available on https://github.com/shakeelakram00/DPFL-FPGA-Accel-Framework.git for wider adoption. To the best of our knowledge, this marks the first implementation of DPFL with FPGAs that enables users to design optimized FPGA accelerators for ML tasks implementing privacy-preserving federated learning with adjustable epsilon values. We illustrate the application of this framework in processing healthcare data, designing a DPFLenabled FPGA system for cardiac arrhythmia classification within a typical hospital setting, where electrocardiograms are not shared among hospitals. The FPGA-Accel achieves significant inference speedups, providing up to 605.2x and 107.2x faster performance compared to Arm Cortex-A53 and Intel Core i7 processors, respectively.