Spectrum of Engineering Sciences

A DEEP LEARNING APPROACH TO PCOS DIAGNOSIS: TWO-STREAM CNN WITH TRANSFORMER ATTENTION MECHANISM

Tariq Rahim — 2025-07-02

Polycystic Ovary Syndrome (PCOS) is a prevalent endocrine disorder affecting a significant portion of women worldwide, often underdiagnosed due to the complexity of its symptoms and limitations in existing diagnostic tools. With advancements in deep learning and medical imaging, automated classification systems offer the potential to revolutionize PCOS detection through precision and scalability. This study proposes a novel Two-Stream Convolutional Neural Network (CNN) architecture enhanced with Transformer-based attention mechanisms for classifying PCOS from ultrasound images. Leveraging dataset of 11,784 images, our framework splits each ultrasound image into upper and lower halves to capture anatomical variance and apply convolutional encoding separately. A Multi-Head Attention layer then integrates spatial dependencies between the two streams, enhancing feature discrimination and improving model interpretability. Experimental evaluations show that the proposed model achieves a classification accuracy of 98.96%, an F1-score of 0.99, and minimal loss on the test dataset. These results highlight the model’s robustness and potential applicability in real-world clinical settings for the early detection of PCOS.

HYBRID DEEP LEARNING MODELS FOR MULTI-CLASS CLASSIFICATION OF CHEST X-RAY IMAGES: NORMAL, PNEUMONIA, AND COVID-19

Ayesha Saddique — 2025-07-02

CXRs remain a major diagnostic method that assists in the identification of respiratory diseases such as COVID-19, pneumonia, and tuberculosis. They are helpful in the clinical setting, especially in low and middle-income countries, where they serve as the gateway to the healthcare system as a result of their affordability when compared to CT and MRI scans. In spite of these advantages, CXR scans still exhibit considerable challenges, particularly in diagnosing CXRs which remains a labor-intensive high expertise process with a large range of inter-reader variability. The problem is exacerbated by multi-class classification where there is pneumonia and COVID-19 which have overlapping radiographic features. The problem that this particular work intends to address is developing and testing the hybrid models that consists of CNNs and transformers models to increase diagnostic accuracy for classification of chest X-ray images into Normal, Pneumonia, and COVID-19 categories. The dataset used was comprised of 7,135 chest X-ray images, which after were subjected to uniform pre-processing to aid in consistency and standardization. Hybrid models were developed such that they paired CNNs with other transformer-based models like DenseNet121 + Swin Transformer and EfficientNetB0 + Vision Transformer. With these models, training was undertaken using the TensorFlow/Keras framework and evaluation was done based on accuracy, precision, recall, F1 score, and confusion matrix. The findings indicate that the DenseNet121 + Swin Transformer model achieved the highest accuracy, precision, and recall scores, outperforming all other models, which demonstrates its potential for more reliable classification compared to CNN-based techniques. The study nonetheless notes the considerable potential of such hybrid models to augment diagnostic functionalities in clinical practice, even with hurdles like dataset imbalance and the absence of real-world validation.

AI-BASED RESOURCE MANAGEMENT FRAMEWORK FOR NEXT-GENERATION WIRELESS NETWORKS

Rahat Ullah — 2025-07-02

The advent of sixth generation (6G) wireless networks brings unprecedented challenges in managing ultra-dense, dynamic, and heterogeneous environments. Classical optimization methods lack the scalability, adaptability, and selflessness required to address the challenge of resource management. This paper proposes an AI-enabled resource allocation framework specifically designed for 6G networks through the integration of state-of-the-art machine learning (ML), deep reinforcement learning (DRL), federated learning (FL), and graph neural networks (GNNs). The envisioned multi-layered architecture allows edge devices, UAVs, and base stations to perceive nearby environments, forecast traffic trends, make in-real-time decisions, and jointly train models with privacy preserved. An end-to-end global controller from GNN provides orchestration over the network topology. We review state-of-the-art AI methods and discuss their adequacy in accommodating resource allocation complexity with trade-offs between convergence, latency, and scalability. We conclude by describing current challenges—heterogeneous data, stable convergence, and limited computations—and sketch future directions of research towards reliable, explainable, and energy-efficient AI deployment in 6G systems.

REAL-TIME UNDERWATER VISIBLE LIGHT COMMUNICATION USING PULSE WIDTH MODULATION AND LIGHT DEPENDENT RESISTORS

Ali Sher — 2025-07-03

Visible Light Communication (VLC) has emerged as a groundbreaking technology, utilizing light as a medium for wireless data transmission. This paper introduces an innovative Underwater Visible Light Communication (UVLC) approach for real-time imaging transmission. The proposed system employs Pulse Width Modulation (PWM) and an integrated Light Dependent Resistor (LDR) to achieve fundamental light detection capabilities. Our system can transmit data at 160 bits per second (bps) over substantial distances underwater, utilizing Light Emitting Diodes (LEDs) as optical transceivers. This setup ensures robust and efficient communication in aquatic environments and demonstrates visible LEDs' dual functionality for illumination and optical communication.

EVALUATION OF THERMAL CONDITIONS AND STANDARDS A CASE OF SCHOOL BUILDINGS IN ORANGI TOWN, KARACHI

Fabiha Khalid — 2025-07-03

The rapid global urbanization has led to significant climatic changes, with rising temperatures being a pervasive concern across various regions. In Karachi, these shifts have particularly impacted the thermal comfort within built environments. As a result, occupants increasingly spend more time indoors to mitigate extreme weather conditions. In educational settings, where students spend a large portion of their day, the thermal environment plays a crucial role in both their comfort and learning efficiency. There is well-established evidence linking indoor environmental conditions to student performance, highlighting the importance of a suitable thermal environment for fostering optimal learning outcomes. Considering this, it is critical to establish a satisfactory indoor environment in educational buildings. This study focuses on private school buildings in Orangi Town, Karachi, which are constructed on small plots. These schools have thermal environments that are not conducive to effective learning. The poor thermal conditions have adversely affected the behavior and productivity of building occupants, particularly students. Therefore, this research aims to assess the thermal conditions and various building-related factors that contribute to the discomfort. The methodology for this study includes mapping, field surveys, and quantitative measurements of indoor and outdoor temperatures and humidity levels from December 2023 to May 2024. Additionally, the study will use an adaptive model to identify comfort temperature standards for the surveyed school buildings and provide recommendations for creating a more comfortable and productive learning environment.

ELECTROCHEMICAL TRANSFORMATION OF CAPTURED CARBON DIOXIDE INTO SUSTAINABLE FUELS AND CHEMICALS: LEVERAGING KC8 TECHNOLOGY FOR ADVANCED CO₂ CAPTURE AND UTILIZATION

Neelam Shahadat — 2025-07-03

The rapid increase in atmospheric carbon dioxide (CO₂) concentrations due to anthropogenic activities is a major contributor to climate change and environmental degradation, prompting urgent efforts to develop efficient carbon capture and utilization (CCU) technologies. This research investigates a pioneering integrated system that combines the novel use of potassium graphite (KC8) for highly effective CO₂ capture with an advanced electrochemical conversion process to synthesize sustainable fuels and valuable chemicals. KC8, with its exceptional electron-donating properties and strong chemical affinity for CO₂, enables efficient and selective carbon capture at ambient temperatures and pressures, overcoming many of the limitations faced by traditional carbon capture materials. Once captured, the CO₂ is electrochemically reduced in a specialized reactor designed to optimize reaction kinetics and product selectivity, leading to the generation of critical renewable energy carriers such as syngas, formic acid, and methanol. Experimental results demonstrate that the KC8-assisted capture significantly enhances the Faradaic efficiency of the electrochemical process, lowers the overall energy input required for CO₂ conversion, and improves the selectivity toward desired fuel and chemical products compared to conventional electrolysis systems. This dual-stage approach not only provides a viable method for mitigating greenhouse gas emissions but also promotes a circular carbon economy by transforming captured CO₂, a major environmental pollutant, into commercially viable and environmentally friendly products. The findings indicate that the integration of KC8 capture technology with electrochemical conversion presents a scalable, cost-effective, and energy-efficient pathway toward sustainable carbon management and renewable fuel production. Furthermore, this approach aligns with global efforts to achieve net-zero emissions targets and offers a promising platform for future research in the field of sustainable energy and environmental technology. The study contributes valuable insights into the design of next-generation CCU systems and highlights the critical role of advanced materials in enabling efficient electrochemical CO₂ utilization for climate mitigation and green chemistry.

DESIGN AND PERFORMANCE ANALYSIS OF 28 GHZ MIMO ANTENNA FOR 5G COMMUNICATIONS

Irfan Ali — 2025-07-03

This research examines the specifications of the 5G mobile communication network. The arrangement distinctly illustrates the two conventional semi-circular microstrip patches of the MIMO antennas operating at 28 GHz. The antenna employs embedded feeding technology, with Rogers RT/duroid 5880 as the dielectric substrate. The antenna dimensions are 18mm x 36mm x 0.8mm. The Defected Ground Structure (DGS) is groove-deflected on the ground plane to enhance impedance bandwidth and optimize isolation performance; therefore, the impedance-matching bandwidth is 5.6 GHz, substantially augmenting the antenna's frequency flexibility. Furthermore, the design incorporates a T-shaped slot in the chosen optimization, mitigating electromagnetic interference among the antenna parts. The test findings indicate that the antenna will exhibit a reflection coefficient (S11) below -10 dB within the frequency range of 24.9 GHz to 30.5 GHz, and the mutual coupling (S21) will be below -40 dB. The Envelope Correlation Coefficient (ECC) is below 0.005, the radiation efficiency is above 95, and the diversity gain (DG) surpasses 9.99 dB, therefore fully satisfying the performance criteria for 5G communication systems. The MIMO arrangement minimizes return loss and broadens operational bandwidth. Consequently, it results in tasks characterized by minimum mutual coupling, thereby substantially improving the major performance parameters of directivity, gain, and radiation efficiency.

INTEGRATING MACHINE LEARNING WITH BAROCLINIC INSTABILITY MODELS FOR ADVANCED MESOSCALE ENERGY CASCADE ANALYSIS

Engr. Syed Ishfaq Ahmad — 2025-07-04

This study combines traditional methods, like normal mode decomposition, with cutting-edge machine learning (ML) techniques to enhance the analysis of energy transfer across scales in baroclinic instability models. By leveraging high-resolution, non-hydrostatic simulations, we explore the energy distribution between geostrophic and ageostrophic modes, uncovering distinctive spectral slopes of −3.1 and −2.7, respectively, which underscore the role of inertia-gravity waves at the mesoscale. Employing Convolutional Neural Networks (CNNs), we automate the identification and classification of these modes, streamlining spectral analysis and improving accuracy, even in highly turbulent environments. This approach not only advances our understanding of mesoscale energy cascades but also highlights the transformative potential of machine learning in atmospheric dynamics, paving the way for more precise weather and climate predictions.

ADAPTIVE MULTI MODAL ANNOTATION FOR HIGH QUALITY, SCALABLE MACHINE LEARNING DATA PIPELINES

Shah Faisal — 2025-07-04

The shift to data-centric artificial intelligence emphasizes high-quality labeled data as a cornerstone of machine learning model performance. Manual annotation, however, is labor-intensive, costly, and prone to inconsistencies, limiting scalability for large datasets. This paper proposes the Adaptive Multi-Modal Annotation Framework (AMAF), a novel system integrating weak supervision, large language model-based labeling, and active learning to automate data annotation in ML pipelines. We introduce Dynamic Synthetic Data Augmentation, a technique to generate diverse, domain-specific datasets, addressing bias and scalability issues. Implemented with Snorkel and MLflow, AMAF was evaluated across healthcare (radiology image labeling), natural language processing (intent classification), and autonomous vehicles (object detection). Results demonstrate 18–20% higher label accuracy and 20–30% faster annotation cycles compared to human baselines, with downstream models achieving 7–10% F1-score improvements over tools like Label Studio and Amazon SageMaker Ground Truth. Challenges include domain-specific complexities and rule-based limitations

A SUSTAINABLE PATHWAY FOR CARBON NEUTRAL ENERGY PRODUCTION

Jan Sher Khan — 2025-07-04

The increasing global energy demand, coupled with the urgent need to mitigate environmental degradation, necessitates such innovative solutions that can address both challenges. This research presents a sustainable approach to converting carbon dioxide (CO₂), a predominant greenhouse gas, into a valuable energy resource. By implementing the Sabatier reaction, in which CO₂ is hydrogenated using renewable hydrogen produced via water electrolysis, resulting in the synthesis of methane (CH₄). The methane generated is a fuel which has the ability of producing electricity. From which we gain an effective transformation of a harmful emission such as CO₂ into a clean energy source. This entire process is modeled and simulated by using Simulink/MATLAB that is integrating all necessary chemical reactions by optimizing it through a Model Predictive Controller (MPC) which will be able to ensure efficient energy conversion and maintain system stability. This process includes CO₂ capture, hydrogen production, and methane synthesis, with a dynamic controller (MPC) to maximize its performance. The results shows a fluent methane output of 0.98 mol/s, a power generation efficiency of 39.8%, and a 35% reduction in CO₂ emissions compared to traditional combustion processes. The total efficiency of system reaches to a high value of 57.1%, which shows that how potential this method is to reduce the energy crisis while mitigating the adverse effects of CO₂ on our environment. This combined approach do not address the important issue of CO₂ emissions but it also contributes to energy sources that we are already using which enhances energy production, by transforming CO₂ into a resource for electricity generation. This study offers a viable pathway toward achieving net-zero emissions while supports the global transformation to sustainable energy systems.

INTELLIGENT BLUETOOTH-DRIVEN AUTOMATION:MULTI- SENSOR INTEGRATION FOR REAL-TIME SMART HOME CONTROL

Farhat M. Khan — 2025-07-04

The integration of IoT into smart home systems addresses the growing demand for advanced home automation by enabling adaptive, intelligent decision-making. While contemporary smart homes leverage IoT, existing systems often lack adaptability, leading to inefficiencies in user experience, energy consumption, and security. This study explores the challenges, opportunities, and implications of implementing intelligent-based smart home technology, proposing a framework that utilizes machine learning and computer vision to create systems capable of learning user behaviors, anticipating needs, and optimizing home management in real time. The study highlights the limitations of current technologies, including their inability to dynamically adjust to user preferences and on-demand accessibility. A prototype smart home system was developed, employing algorithms to analyze usage patterns and autonomously manage lighting, temperature, and security. Deployed in real-world settings, the system’s performance was evaluated through user feedback and operational data. Results demonstrated significant improvements in user satisfaction (due to personalized automation), energy efficiency (through optimized resource usage), and security (via proactive threat detection). Additionally, the system’s adaptive capabilities reduced energy waste by intelligently adjusting to occupancy and environmental conditions. This study underscores the transformative potential of AI-driven smart homes, offering a sustainable alternative to conventional systems by prioritizing user-centric design and responsiveness. The findings advocate for broader adoption of intelligent technologies to create homes that not only automate tasks but also evolve with residents’ needs, fostering a more efficient, secure, and comfortable living environment. Ultimately, this

research validates AI’s role in redefining human-home interactions, paving the way for future innovations in smart living solutions.

DEVELOPING ENVIRONMENT FRIENDLY GROWING MEDIA FOR SOILLESS CULTIVATION OF SEASONAL VEGETABLES

Sheeraz Aleem Brohi — 2025-07-04

The pursuit of sustainable agriculture, particularly in arid and semi-arid regions, necessitates alternatives to traditional soil-based cultivation. This study explores the development of environmentally friendly, cost-effective soilless growing media using locally available materials in Tandojam, Sindh, Pakistan. The objective was to identify optimal combinations of biomass, nutrient supplements, and stabilizers for cultivating seasonal vegetables, tomato (Solanum lycopersicum) and okra (Abelmoschus esculentus). Locally sourced materials were categorized into biomass ( coco peat, rice husk, wheat straw), supplements (e.g., compost, manure, biochar), and stabilizers (sands, expanded clay, styrofoam). These materials were dried, processed to 2-5 mm size, and mixed in volumetric ratios. Seventy-five media formulations were initially screened under greenhouse conditions. Based on germination rates (>70%) and seedling vigor, 15 treatments were selected for detailed analysis in a Randomized Complete Block Design (RCBD) with three replications. Comprehensive analysis of physico-chemical properties (WHC, porosity, pH, EC, nutrients) and plant growth parameters (germination, plant height, leaf and fruit number) was conducted. Results showed high WHC in banana leaves and coco peat (>82%), with high porosity in wheat straw and styrofoam. Media T32 (60% wheat straw, 20% biochar, 20% expanded clay) showed high vigor in both crops, while coco peat-based media T6 (50% coir, 25% compost, 25% pit sand) and T11 (60% coir, 20% compost, 20% pit sand) consistently achieved superior plant growth. However, bagasse and wheat straw-based media underperformed. Nutrient analysis revealed post-harvest depletion, highlighting the need for fertilization. This study concludes that coir-based media, combined with compost and structural stabilizers, offer a sustainable, low-cost solution for soilless vegetable cultivation in arid climates.

SECURING THE FUTURE, THE DUAL ROLE OF ARTIFICIAL INTELLIGENCE AS A THREAT AND A TOOL IN MODERN CYBERSECURITY

Muhammad Ali Khan — 2025-07-04

This study explores the counterintuitive nature of AI in cybersecurity today. It is looking at its potential applicability in terms of protecting the digital infrastructure and examining the dangers of mishandling the technology. AI enhances defense systems by use of predictive analytics, anomaly identification and automated response systems. AI presents an opportunity because malicious groups use it to conduct advanced attacks like deepfakes, intelligent phishing and malware that is created using AI. The research is based on a qualitative and quantitative mixed-methods technique. A review of the literature was made to study the existing AI-based tools of cybersecurity and the tactics of attacks. The results indicate that the connection between AI and cybersecurity is complicated since innovations are only followed by increased vulnerability. AI drastically enhance the speed of threat identification, the responsiveness of a system, and its resilience, dual-use capabilities impose ethical requirements that require proactive implementation, superior governance, and counter-AI strategies. It is merged with the realm between the positive impacts of the role of AI in defense. The negative impacts that could demonstrate the exploitation of this running defensive technology.

Keywords

( Artificial Intelligence, Cybersecurity, Machine Learning, Anomaly Detection, Intrusion Detection Systems, Cybercrime, Zero Trust Architecture, AI Governance, Cyber Defense, Intelligent Phishing).

DEVELOPMENT OF A DYNAMIC ONTOLOGY-BASED FRAMEWORK FOR ENHANCING THE INTELLIGENCE AND AUTONOMY OF ROBOTIC SYSTEMS

Abdulrehman Arif — 2025-07-04

In recent years, intelligent agents driven by ontologies have shown significant promise in learning from their experiences and adapting to dynamic environments. These agents have been successfully applied to a variety of domains, including autonomous systems, robotics, and decision support systems, where their ability to process and adapt to changing circumstances is crucial. However, despite their potential, these agents often encounter significant challenges when confronted with unpredictable or hostile conditions, such as high uncertainty, evolving adversarial behaviors, or environmental disturbances. In such scenarios, traditional models may struggle to maintain effective learning and decision-making, leading to suboptimal performance and failure to achieve desired outcomes. This limitation arises from the inability of many existing models to fully account for anomalies or unexpected changes that deviate from anticipated patterns. As a result, the agents' capacity for robust decision-making and adaptability is compromised, hindering their performance in real-world applications. Addressing these challenges requires models that are not only capable of learning from experience but also flexible enough to deal with unexpected events and changes in the environment. To overcome these limitations, this thesis introduces a novel dynamic, ontology-driven agent model that emphasizes continuous learning from past experiences. The model integrates an adaptive reasoning framework capable of adjusting its strategies and planning processes in response to new, unforeseen challenges. By leveraging the rich contextual information embedded in ontologies, the proposed agent model can enhance its action planning and reasoning abilities, enabling it to identify and adapt to anomalies more effectively. Furthermore, the model is designed to improve its adaptability by dynamically updating its ontological knowledge base based on real-time data, ensuring that it remains resilient in hostile and unpredictable environments. This research seeks to advance the field of intelligent agents by proposing a more robust and adaptive framework that not only learns from past experiences but also evolves and adjusts its strategies in response to novel and challenging conditions. The expected outcome is a highly adaptable agent model capable of improving decision-making, planning, and performance in dynamic, uncertain, and adversarial environments, thus expanding the potential applications of intelligent agents across various complex domains.

Keywords

( Autonomous Agent, Dynamic Ontological Model, Ontology-based Framework, Adaptability, Environment Perception, Meta-cognition, Heterogeneous Environments, Dynamic Problem Solving).

SOLAR PHOTOVOLTAIC TECHNOLOGIES AND APPLICATIONS: A CASE STUDY OF PAKISTAN

Nadir Mehran — 2025-07-04

This research paper analyzes the situation of the development of solar photovoltaic (PV) technologies and their usage, paying particular attention to the outstanding absorption of solar energy in Pakistan. It has become one of the world's fastest-growing solar markets, and solar additions contributing to about a third of the national total generating capacity were completed in 2024 alone. This paper examines the different PV technologies comprising monocrystalline, polycrystalline, and thin-film solar panels, as well as their efficiency qualities and how specific technologies can be used in the wind and solar energy sector in Pakistan. The article examines the factors that are causing the solar explosion in Pakistan: they include unreliable supplies of electricity, increasing energy prices, and government policy. This paper examines some PV technologies such as monocrystalline, polycrystalline, and thin-film solar panels, their efficient nature, and their particular application in the Pakistan energy sector. The paper discusses the factors that have fueled the solar boom in Pakistan, and they include an unstable electricity supply, an increase in the cost of electricity, and favorable government regulations. This work analyzes market data, policy guidelines, and technology applications in detail to give implications of how developing countries can adopt the use of solar PV technology to improve energy security issues. The results show that the solar market in Pakistan has been increasing since 2023 when it reached 2.9 GW of imports, 16 GW in 2024, and by 2024, it is recommended that the country becomes a global leader in solar energy consumption, as the case with Pakistan. The vast majority of solar is being added on the frontier of renewable installations; Pakistan is one of the fastest-growing solar markets in the world, with solar providing a share of about a third of the entire generating capacity added in 2024 alone. By conducting a detailed study of the market trends, policymaker regulations, and technological execution, this study gives a clear idea of how the Third World countries can use solar PV technology to meet energy security issues. The conclusion shows that the solar market of Pakistan increased during the year 2024 (16 GW of imports) in comparison to 2023 (2.9 GW of imports), making the country one of the global leaders in adopting solar energy

Keywords

Solar photovoltaic, renewable energy, Pakistan, energy policy, solar technology, sustainable development

PERFORMANCE ANALYSIS OF SOLAR PHOTOVOLTAIC SYSTEMS: A CASE STUDY

Kifayat Ullah Khan — 2025-07-04

Solar PV technology has become the central pillar of renewable energy strategies all over the world due to its rapid growth rate. The paper offers a thorough analysis of the performance features of solar PV systems by means of an overview of operation data, environmental conditions, and degradation mechanisms. As of 2024, the worldwide capacity of PV is over 1.6 TW, producing 2,136 TWh of electricity per year; it is important to understand the performance of a system in order to maximize energy production and cost-effectiveness. The case study compares various PV installations in various climatic settings and the differences in performance and degradation rates, as well as the environmental influence on the performance of PV systems. Important discoveries are that modern PV systems degrade with an average rate of 0.2-0.8 percent per year, and the environmental conditions known to affect the performance aretemperature, humidity, and dust accumulation on the modules. The research shows that with appropriate monitoring and maintenance, up to 15-25 percent of the efficiency of the system can be enhanced, whereas sophisticated tracking systems can generate up to 30 percent of energy rise. The results help in the design and operation of PV systems to optimize them and promote the global shift towards sustainable energy systems.

Keywords

Solar photovoltaic, performance analysis, energy efficiency, degradation analysis, renewable energy, system monitoring

IMPACT OF ICT ON INVENTORY MANAGEMENT IN RETAIL SUPPLY CHAINS: A STUDY OF IOT AND AI APPLICATION

Muhammad Hashim Zia — 2025-07-05

The retail sector is using information and communication technology (ICT) more and more to maximize inventory control. This study examines how ICT, especially Internet of Things (IoT) and Artificial Intelligence (AI), affects inventory management in retail supply networks. Objective: To study the use of IoT and artificial intelligence in stock management and its effect on retail supply chain efficiency. Methodology: Survey-based study using stratified random sampling for data gathering from 100 retail professionals taking part in the survey. With an emphasis on businesses that have incorporated Internet of Things and artificial intelligence in their inventory management systems, data were gathered from Retail companies in urban areas. Online survey questionnaire was used for data collection. Results: Following the deployment of IoT and artificial intelligence technology, the research discovered a 25% decrease in inventory costs, from $100,000 to $75,000. From 80% to 104%, the research showed a 30% increase in supply chain efficiency, therefore implying higher production and reduced operational costs. Better product availability and fewer stockouts increased consumer happiness by 20%, from 80% to 96%. Overstocking and stockouts were found to have fallen 15% from 20% to 5%. Conclusion: The use of artificial intelligence and IoT in inventory management could completely revolutionize retail supply chains. Retailers can optimize inventory levels, minimize costs, and improve customer satisfaction by means of these technologies, hence gaining a competitive edge in the industry.

DESIGN AND IMPLEMENTATION OF A FIREBASE-INTEGRATED IOT-BASED FLOOD DETECTION AND EARLY WARNING SYSTEM FOR PAKISTAN

Naeem Akbar Channar — 2025-07-05

Pakistan is a country that is very susceptible to floods due to its geographical and climatic conditions. Poor urban planning, weak infrastructure, and no early warning systems make it even more deadly. Destructive monsoon floods break every year, homes, agriculture, and the transport system apart not to mention the loss of human life. Climate change has already had an impact on increasing the frequent severity of flooding; hence there is a strong requirement for modern technology-based solutions. This study presents a low-cost scalable IoT-based flood detection and early warning system suitable for flood-prone areas within Pakistan utilizing ultrasonic water level sensors, rain sensors plus water flow sensors integrated with NodeMCU(ESP8266) microcontroller continuously monitoring environmental parameters realtime data being uploaded to Firebase Realtime Database where it gets visualization through a custom-built web application. These sensors continuously monitor the environmental parameters and upload the data in real time to Firebase Realtime Database where it gets visualized through a custom-built web application. Instantly, based on thresholds, alerts are sent to warn emergency responders. The system has been prototyped and tested to validate its responsiveness and accuracy. Future integration with predictive models will advance system capabilities.

ANPR FOR PAKISTANI LICENSE PLATES: A DEEP LEARNING APPROACH USING CNN AND YOLO

Abdulrehman Arif — 2025-07-05

Automatic Number Plate Recognition (ANPR) is an important technology used in various applications such as traffic monitoring, law enforcement, and parking management and used to recognize number plates in real time. . This discrepancy can be attributed to a number of difficulties Pakistan faces, such as the lack of uniform number plate sizes, the predominance of fancy number plates, a variety of fonts and a wide array of colors. These elements seriously impede the development of an effective ANPR system in the nation. The research proposal offers an effective method for Automatic Number Plate Recognition (ANPR) that is tailored to the particular features of Pakistani number plates. The method entails two main steps: Identification and localization of license plates. Using optical character recognition (OCR), convolutional neural networks (CNN), and YOLOv3 (You Only Look Once), characters from a license plate image are retrieved. In order to determine whether a license plate is genuine or fake, a special algorithm is also used. This procedure identifies the license plate precisely in an image and extracts the essential data for further processing. Utilising a dataset of 200 images of Pakistani license plates, including both bike and car plates, the effectiveness of the suggested method was assessed. The evaluation's findings demonstrated that the suggested method had a remarkable accuracy rate of 96.7%. This shows that the method is very effective at localizing and identifying Pakistani license plates.

POTENTIAL REUSE OF WASTEWATER BY UTILIZING NANO-FILTRATION TECHNIQUE-A RESEARCH PAPER

Muhammad Waqas Ahmed Saleem — 2025-07-05

The generation of wastewater from human activities is an inevitable consequence of urbanization and industrialization. In Pakistan, where water scarcity is exacerbated by aging infrastructure and increasing demand, the development of cost-effective and efficient water treatment technologies is crucial. Conventional wastewater treatment plants (CWWTPs) are often limited by high construction and maintenance costs, as well as significant land requirements. This study evaluates the performance of nanofiltration (NF) membranes for wastewater treatment as a sustainable alternative. A pilot-scale NF system using hollow fiber membranes with a pore size of 0.002 µm was employed to assess the removal efficiency of key water quality parameters, including biological oxygen demand (BOD), chemical oxygen demand (COD), total suspended solids (TSS), total dissolved solids (TDS), turbidity, hardness, and pH. Experiments were conducted at pressures ranging from 4 to 10 bars, with optimal performance observed at a filtration time of 15–20 minutes and a flux rate of 20 L/m²·h using undiluted wastewater. Results indicated a marked improvement in treatment efficiency with increasing operating pressure, demonstrating the potential of direct NF treatment in reducing contaminants effectively while minimizing infrastructure demands.

INTELLIGENT BEARING FAULT DIAGNOSTICS WITH MACHINE LEARNING AND DEEP LEARNING FRAMEWORKS: A COMPREHENSIVE AND COMPARATIVE ANALYSIS

Arham Memon — 2025-07-05

In this research work, we systematically investigate the deep learning and classical machine learning approaches for bearing fault diagnostics. Bearings are the most critical mechanical components as their operating conditions have a direct effect on the safety and working of the mechanical equipment. As per the demand of industry users, researchers have always paid great attention to the quality, durability, and service life of bearings. Recent developments in machine learning and especially in deep learning have framed new research areas that have developed an increased interest in both industry experts and academic researchers. In this research, we first investigate the working, characteristics, and limitations of available machine learning and deep learning methods in bearing fault diagnostics applications such as artificial neural networks (ANNs), deep belief networks (DBNs), support vector machine method, etc. Apart from current methods in available literature, the new methods and functionalities are also analyzed and a detailed section on potential methods along with data sets is also dedicated so that it helps other researchers to extend their research. In last, a detailed comprehensive and comparative analysis between the machine learning and deep learning methods is also provided along with discussion section which is intended to facilitate in applying these algorithms for specific applications. The future research section is also added to discuss the current research limitations and potential research areas.