ABSTRACT. The talk presents the integration and synergy between generative AI, human augmentation, and wireless technologies. It effectively conveys advanced, innovative concepts, appealing to an audience interested in the future of AI and technology. Every participant will receive a complimentary code from NVIDIA to test the models.

ABSTRACT. Accurate short-term electrical load forecasting is crucial for efficient energy management and grid stability. This paper presents an advanced ensemble approach for enhancing short-term load forecasting accuracy by integrating Random Forest (RF) and Histogram-based Gradient Boosting Regression (HGBR) algorithms. The ensemble method leverages the strengths of each algorithm to capture complex patterns and interactions in the data. Notably, the model incorporates real-world data from the University of Brescia's electrical facility data collection system, adding practical relevance to the study. Additionally, external weather components, particularly ambient temperature, are included to improve predictive capabilities. Experimental results demonstrate that the proposed ensemble model significantly outperforms individual forecasting methods in some metrics, achieving higher accuracy and reliability in predicting electrical load. The inclusion of ambient temperature as an external variable contributes to the enhanced performance, highlighting the importance of weather factors in load forecasting.

ABSTRACT. Fabric weave type identification has many uses that adds to importance of advance classification methods; e.g fabric weave should be known in order to identify the most suitable use. It is important to retailers that they should be sure of fabric make before placing their bulk orders. Sometimes when importing raw material overseas, it is important to know the fabric make to calculate custom duty computations. The classification of weave pattern is difficult because the manual inspection of fabric takes more time which requires extensive labour. In recent years, computer vision based approaches have achieved remarkable performance in the fields of object recognition and image processing, thus opening a new door to develop an automated system for the recognition of woven fabric patterns for production of fine quality fabric according to customer needs. In this paper, we have devised an automatic weave pattern recognition mechanism based on convolutional neural networks(CNN). Initially three basic weave patterns were chosen i.e Twill, Plain and Satin. The results were analysed for its possible adaptability in the task of fabric weave pattern recognition.

ABSTRACT. Crops are one of humanity's fundamental needs. A product for which demand continually increases. A considerable amount of its production is lost owing to a lack of care or incorrect assessment of the various growth and productivity-related metrics. The wireless crop monitoring system requires a portable, user-friendly, highly accurate, automated information generator for several crucial, associated parameters. Soil testing in laboratories and networked monitoring systems can provide data on the fertility and toxicity of the soil, as well as the meteorological conditions necessary for crop growth. Crop fields cover a large area, so getting information on impacting parameters is not a problem. In this paper, temperature, humidity, soil moisture, and air quality play a significant role in agricultural growth. The parameters listed above will be monitored using wireless sensor networks (WSN). By assessing soil moisture, the issue of excessive water supply can be resolved. Monitoring temperature allows for the estimation of a plant's water requirements. Monitoring air quality allows for the maintenance of healthy air quality.

ABSTRACT. The use of wireless charging in Underwater Autonomous Vehicles (UAV) has been studied in depth during last years. However, to integrate a wireless charger in a submerged vehicle, it is important to analyse the external variables that can affect the system performance. Among them, one of the consequences of charging an UAV in a seawater environment is the appearance of parasitic capacitances between the power coils, due to the conductivity of this medium. The objective of this paper is to analyse the feasibility of neglecting these parasitic capacitances. The analyses carried out conclude that there is an optimal solution in which it is valid to neglect the effects of parasitic capacitance in the system performance, considering the variations in voltage, current, as well as efficiency studies.

ABSTRACT. Several cutting-edge modern technologies, including hologram technology, have emerged due to the tremendous advancements of our era. The science of holography is used to make holograms, which are 3D images with lifelike graphics. With its vibrant 3D views and immersive experiences, holographic technology transforms into a different form of fields like education, entertainment, advertising, etc. This study outlines an easier way to modify the 2D images into 3D holograms. Blender software and a holographic fan projector are used in this process. The process begins with, Blender, the 3D graphics suite, is opened and converts flat images into more detailed animated models. These are then sent to a mobile app which transfers them onto a holographic display easily. The mobile application acts as a convenient medium for users to customize or manage their pictures. Utilizing innovative holographic technology to produce lifelike 3D images, the inquiry focuses on obtaining images from the mobile app and projecting them onto the holographic display. The human hologram and building structure are the two areas covered in this study. In effect, this article essentially assists construction businesses with building structures, and educational institutions employ human holograms for lecture purposes. This research, also investigates how the advantages of holographic models compare to other fast-growing platforms both now and soon.

ABSTRACT. Due to its adaptability in harsh situations and resistance to signal distortion, the applications for fiber optics are virtually limitless. Through wireless broadband, wireless technology has enabled access to the internet at fast speeds. As an optimal solution Radio Over Fiber (ROF) combines electromagnetic waves and light waves for the wireless and wired environment. It is a viable option for decreasing equipment, operating, and capacity growth expenses. The objective of this study is to compare the performance of ROF in wireless environments such as LTE and WiMAX to that of conventional optical fiber. Radio technology employing optical fiber as a transmission medium may be the answer for future generation. The research is then employed to determine the viability of ROF as a backhaul solution for LTE and WiMAX. MATLAB is used to deploy an optical and ROF network to acquire simulation results that validate the theory.

ABSTRACT. A Generalized singular value decomposition (GSVD) based relay broadcast beamforming for non-regenerative relays serving the multiple MIMO users in a dual-hop network is proposed. Under the relay power constraint, the GSVD-based relay beamforming is designed to increase the sum rate of the dual-hop multiuser MIMO network. Moreover for the relay receive beamforming matched filter is used to maximize signal to noise ratio at the relay node. In this relay processing design, the ideal channel information of the first and second hop channels is considered at the relay node. The source is supposed to transmit the multiplexed data to the relay node. The simulation results demonstrate the sum-rate performance of the proposed relay processing scheme.

ABSTRACT. Wireless power transfer (WPT) is becoming an interesting solution for the electric vehicle (EV) industry. This kind of technologies presents several issues in the final implementation. Between the most important challenges are the control techniques in order to correct the deviation produced by components tolerance. In this paper, the authors propose a fuzzy logic control technique in order to correct this deviations. This control will ensure that the system works under highest efficiency conditions. To do so, the control varies the value of the operating frequency inside the SAE J2954 range. A bunch of experimental measures are performed in the laboratory in order to test the viability of the proposal.

ABSTRACT. The Air Production Unit (APU) is responsible for managing the load in the metro train by making sure that the load is evenly distributed on the wheels despite passenger congestion. The failure of APU results in a complete halt of operations. Therefore, it is pivotal to timely maintain the APU system. In this paper, a predictive maintenance (PdM) approach has been suggested using multiple state-of-the-art machine learning algorithms to predict failure in the APU. Sensor parameters that lead to failures in APUs of the electric trains were identified from the MetroPT train dataset by employing a verity of baseline models including linear regression (LR), decision tree regressor (DTR), random forest regressor (RFR), gradient boosting regressor (GBR) and XGBoost Regressor (XGBR) models. This will significantly reduce both the operational cost as well as the downtime of the trains. It will also help in the identification of faulty parts at a faster rate and reduce the failure rate of urban transportation systems.

ABSTRACT. Railway wheels are one of the most critical components of the railway infrastructure, functioning as the load carrier. They are exposed to various forms of damage, caused due to intense sliding friction and inadequate inspection procedures. This research introduces a real-time wheel fault detection system using YOLOv8, leveraging computer vision and AI-based object detection. The main contribution of this research is the utilization of a custom data acquisition setup for capturing wheel images and detecting faults with high accuracy and speed. Our results demonstrate an overall F1 score of 0.80,with precision and recall curves indicating strong wheeldetection and moderate defect detection capabilities. This approach aims the reduction of inspection costs and the improvement of railway safety through automation and digitization.

ABSTRACT. Passwords remain a critical component of authentication systems due to their ease of implementation, despite the availability of more secure methods such as biometrics and smart cards. However, password-based systems are vulnerable to various attacks due to the predictable patterns users often employ. This vulnerability necessitates the development of robust strategies to enforce strong passwords. This research focuses on modeling password strength prediction as a classification task using multiple supervised machine learning algorithms. The goal is to classify passwords into categories of weak, medium, and strong, thereby enhancing the security of systems against online and offline attacks. The result findings demonstrate that Artificial Neural Networks (ANN) and Random Forest (RF) significantly outperform their counterparts in predicting password strength, achieving accuracy rates up to 98.97%. This study aims to provide insights into effective password management and underscore the potential of machine learning in enhancing cybersecurity.

ABSTRACT. Cloud storage is a service provided by cloud computing that involves remote maintenance, management, and backup of data that can be accessed by users over a network, usually the internet. The concept of data auditing has been developed due to the increasing concern about the safety of data stored in the cloud which can be attacked externally or modified without authorization. This process involves assessing data integrity through a Third-Party Auditor (TPA). This study focuses on identifying and mitigating malicious TPAs in cloud-based data storage using privacy-preserving public auditing. Simulation results carried out using Python tool show that the proposed system has been extended to support batch auditing processes. This improvement enables TPA to handle multiple authorization audits concurrently thus enhancing the overall efficiency of the auditing system. In addition, dynamic data operations and batch auditing are also supported by the suggested system. The suggested system is both very secure and very efficient, according to a thorough evaluation of its performance and security.

ABSTRACT. Every year, thousands of people throughout the world are diagnosed with skin cancer and enrolled for treatment. Cancer of the skin is one of the forms of cancer that is responsible for the deaths of millions of people every year. The early diagnosis and treatment of newly diagnosed instances of severe skin cancer are very necessary in order to guarantee a high survival rate in addition to a low mortality rate. The custom neural network architecture that has been proposed makes use of a custom CNN for feature extraction and integrates Mobile-Net as its foundation model. For diagnosis, the CNN layers have nine separate classifications. Training and testing the algorithm, the very large ISIC Data Set, which can be accessed via Kaggle, is used. This collection of data contains a large number of images that illustrate the many phases of skin cancer. Based on the findings of this research, the validation and training loss are both rather low, reaching 0.15 and 0.14 respectively. To locate the precise positions of cancer lesions, a CNN-based feature extraction method is used along with a Mobile-Net pre-trained model. Adam optimizer has been implemented with a learning rate of 0.001%, which enables the algorithm to become more effective and learn more efficiently. The accuracy of Mobile-Net models is encouraging enough with a 99.06% training accuracy and a 98.11% of validation accuracy. For further information, this model demonstrates a considerable increase in terms of outcomes for patients and survival rate.

ABSTRACT. Chronic venous disease (CVD) affects many people worldwide, especially women, due to stress and work life. Ignoring symptoms like varicose veins can lead to serious problems. Our goal is to enable early self-diagnosis of CVD using images.We used machine learning to achieve this, specifically a convolutional neural network (CNN) to train our model. The dataset, collected from GitHub, includes five stages: normal skin, reticular skin, varicose veins, pigmentation, and venous ulcers. The CNN has various layers: Image input layer, Convolution 2D layer, Batch normalization layer, Rectified Linear Unit (ReLU) layer, MaxPooling 2D layer, Fully connected layer, and Softmax layer.The datasets are splitted into training and testing datasets. After the training, we looked at the loss and accuracy graphs to decide if the model is ready for real use. We created a graphical user interface (GUI) with buttons like Preprocess, Train/Test Split, Train Data, Analyze and Test, Save Model, Load Model, Select Image, Show Image, and Predict.Once training is done, we test the model with new, unseen data to see how well it performs. After testing, the model is saved and can be reloaded. For validation, a new image is selected, shown on the screen, and the result is displayed to show which classification the image belongs to, ensuring accurate evaluation.

ABSTRACT. Nowadays it is necessary to know the different electromagnetic field exposure standards and measurement methods for getting the information regarding power density and electric field near to us. Since mobile phone is the integral part of the personal and daily life, electromagnetic field exposure from base stations is increased. Over the past eight to ten years, there has been a notable surge in wireless data traffic and the introduction of new communications technologies, leading to a rapid development of the technologies employed in mobile phones. This study looks at radio frequency electromagnetic field exposure estimations from base stations in the Kolhapur area and compares different electromagnetic field (EMF) exposure guidelines. Exposure levels are measured in V / m and mw / m2 for 15 selected sites from urban rural areas of Kolhapur using KM-195 exposure meter. Various approaches of measuring electromagnetic field were also mentioned. Power density and electric field statistics are calculated for each measuring region in volts per millimetre and miliwatts per square metre. Finally, all EMF exposure levels are compared with global standards. After comparing it is found that the stricter country in EMF exposure is Austria, its exposure level standard is 0.001 w/m2. It is very lesser than other countries. Followed by Austria the other stricter countries are Germany, Switzerland, Italy, Poland, Paris, Hungary, Russia and Bulgaria. Their EMF exposure levels are 0.09, 0.095, 0.1 and 0.2 w /m2 respectively.

ABSTRACT. CSCC investigates conventional and advanced imaging techniques, including MRI and CT scans, highlighting their respective strengths in visualizing spinal structures. The abstract delves into the evolving role of artificial intelligence algorithms, discussing their potential to enhance diagnostic accuracy and efficiency. Furthermore, the abstract addresses the clinical significance of early detection in preventing complications and guiding. timely interventions. It underscores the importance of inter disciplinary collaboration among radiologists, neurologists, and technologists to refine diagnostic protocols. The abstract concludes by emphasizing the ongoing efforts to optimize detection methods, ultimately improving patient outcomes in the realm of cervical spinal cord compression. The methods and advancements in the detection of cervical spinal cord compression. It discusses variousimaging techniques such as MRI and CT scans, along with emerging technologies like artificial intelligence algorithms. The focus is on early and accurate diagnosis to facilitate timely intervention and prevent potential neurological complications. The abstract also highlights the significance of interdisciplinary collaboration between radiologists, neurologists, and technologists in improving patient outcomes for cervical spinal cord compression.

ABSTRACT. This research explores the implementation of common Wi-Fi attack vectors using the ESP32 microcontroller platform, aiming to assess the potential security implications of readily available, low-cost hardware in the context of wireless network vulnerabilities. The study focuses on developing a comprehensive ESP32-based Wi-Fi penetration testing tool, dubbed "Slipper Zero," which incorporates various attack methodologies including de-authentication, WPA/WPA2 handshake capture, PMKID extraction, and rogue access point creation. By leveraging the ESP32's capabilities and the Espressif IoT Development Framework (ESP-IDF), we demonstrate the feasibility of executing sophisticated attacks with minimal hardware requirements and reduced complexity. The research highlights the accessibility of these attack vectors to potential malicious actors and underscores the urgent need for enhanced Wi-Fi security measures. Our findings indicate that the ESP32 platform, with its small form factor, low power consumption, and versatile Wi-Fi interface, presents both opportunities for security research and potential risks in the hands of attackers. The study concludes with recommendations for improving Wi-Fi security practices and suggests future directions for research in wireless network protection against emerging threats posed by low-cost, portable attack platforms.

ABSTRACT. A sustainable urban transport system is essential for development, particularly in developing cities like Karachi. Achieving public acceptance of sustainable transport measures is challenging, requiring accessibility for all income groups without compromising on travel time, cost, and environmental impact. This study focuses on public attitudes towards the implementation of the transit system on Corridor III in Karachi. Using a combination of qualitative and quantitative analyses, including Cross Tabulation Test and One-way ANOVA via SPSS, the research assesses key performance variables such as time, travel mode, cost, environmental and psychological issues, and sustainability. The data collected is crucial for understanding public willingness to shift from Private Vehicles (PV) to Bus Rapid Transit (BRT) and for forecasting improvements in Karachi's transit network. The study highlights the role of Location Intelligence and smart technologies in enhancing urban mobility and sustainability.

ABSTRACT. MobileNetV2 represents a significant advancement in the design of efficient neural networks for mobile and embedded applications. Its innovative use of inverted residuals, linear bottlenecks, and depthwise separable convolutions makes it an ideal choice for developers seeking to deploy powerful AI models on resource-constrained devices. This paper presents an efficient algorithm leveraging the MobileNetV2 architecture for the classification of eye diseases, specifically targeting Normal, Cataract, Glaucoma, and Retina Disease categories. The proposed model is designed to perform with high accuracy on resource-constrained devices, making it suitable for mobile and embedded applications in telemedicine and ophthalmology. Extensive testing on a comprehensive dataset has demonstrated the model’s effectiveness, achieving precision and accuracy of 96.15% and 97.39%, respectively. This research addresses the growing need for accessible and reliable diagnostic tools in eye care, particularly in remote and underserved regions.