CSE Portal — Boot Sequence
?[[]1&— B/D4&3CD 9]6 /\7!!—
······· ········ ·······
········· ·········· ·······
········· ······· ·····
-%E&2470 46E495E 1! 52[#
······· ·········· ··· ····
········ ····· ·· ········
········· ······· ·······
Project ID = F25SDP 01 CE F
Supervisor: Prof. Qutaibah Malluhi
Amna Al-Obaidli, Aldana Al-Tamimi, Latifa Albuenain, Muneera Alkubaisi
This project presents the design, implementation, and evaluation of a Secure Wireless Vehicle Access System that addresses vulnerabilities in Remote Keyless Entry (RKE) systems. While such systems offer convenience, many are vulnerable to replay and relay attacks due to the use of static or predictable authentication mechanisms. This project aims to demonstrate these vulnerabilities and develop a practical security solution suitable for embedded environments. The system follows a three-stage approach. In Stage 1, a baseline system is implemented using static RF communication to perform lock and unlock operations, with LED and buzzer feedback indicating system states. In Stage 2, system vulnerabilities are demonstrated through replay and relay attacks, where an attacker device captures and retransmits signals to gain unauthorized access, highlighting the weaknesses of static communication. To address these issues, Stage 3 introduces a rolling code–based authentication mechanism. Each transmission incorporates a dynamically changing counter combined with a non-linear polynomial transformation and XOR operation using a shared secret key. This ensures that every transmitted signal is unique and prevents reuse of previously captured signals. The receiver independently verifies incoming messages while enforcing replay protection and synchronization to maintain reliability. The system integrates embedded hardware and software components and was validated through comprehensive testing. Results confirm that the baseline system is vulnerable, while the rolling code mechanism effectively mitigates replay attacks and improves overall system security with minimal impact on performance. The main contribution of this work is a low-cost and modular platform that demonstrates both wireless security vulnerabilities and practical countermeasures within a single system. The structured design provides a clear transition from insecure to secure communication, offering both educational value and a foundation for further development in secure embedded systems.
Project ID = F25SDP 02 CE F
Supervisor: Prof. Junaid Qadir
Dana Al-Janahi, Fay Al-Marri, Asmaa Al-Moaden, Leen Alsatoof
With the rapid growth of e-commerce and online shopping, the need for secure, efficient, and user- friendly delivery solutions has become increasingly important. Traditional package delivery systems often leave packages unattended outside homes, exposing them to risks such as theft, damage from weather, and delivery errors. This project addresses the problem by designing and implementing a Smart Secure Delivery Reception System that provides a safe, automated, and technology-driven method for receiving and storing packages. The main objective of the project is to create a prototype that ensures packages are securely delivered and stored until retrieved by the rightful owner. The system is designed with two gates: an outer gate for initial package drop-off and an inner gate that transfers the package into a secured storage compartment. Access to the inner gate is controlled by sensors and microcontroller-based logic, ensuring that packages are only accepted when space is available and securely locked afterward. To achieve this, the system integrates ESP32 microcontrollers, ultrasonic sensors, and magnetic door switches, and servo motors to manage gate control and package detection. A notification mechanism is also included to alert the user via their smartphone when a package has been successfully received. The design prioritizes modularity and expandability, such as camera monitoring, mobile application support, and environmental control (temperature and humidity monitoring). This project highlights the multidisciplinary application of computer engineering principles, combining hardware interfacing, embedded systems programming, networking, and IoT communication. By implementing real-time control of actuators and sensors, along with potential networking capabilities, the system demonstrates a practical solution to a real-world problem that aligns with modern smart-home trends.
Project ID = F25SDP 03 CE F
Supervisor: Dr. Armstrong Nhlabatsi
Fatima Al-Bader, Jannatul Hossain, Shima Salemzadeh, Aisha Shams
In a world where anyone can take photos or videos with their smartphones or cameras at any moment, privacy threats have become extremely serious. Numerous people are photographed without consent and even though there is anonymization software that we could apply in securing data, it is inefficient and still vulnerable to misuse by app developers or the phone’s operating system. On recognizing the prevalence and seriousness of this problem, the State of Qatar has recently increased the penalties for taking pictures of individuals without their consent in public places, as a measure to safeguard personal privacy and dignity, pursuant to Law No. 11 (2025). This has made many individuals more cautious about taking photographs in crowded areas due to the fear of potential legal consequences. Despite the existence of the law, its efficiency relies on the consequences faced by offenders only. Thus, legal measures are insufficient alone. Our project suggests a technical solution, the HemayaWajh, that corresponds to the law by helping in the enforcement process and preventing the irresponsible, disrespectful or nonconsensual photography in the open areas. This solution will allow people to capture images and videos without encroaching on the privacy of other individuals or causing them embarrassment in a social setting. The proposed device consists of a camera connected to a Raspberry Pi, which processes captured images using facial detection and recognition. The system receives registered faces from the Flutter app database through the website and selectively pixelates the registered faces from the images. This ensures that individuals who do not wish to be photographed remain unidentifiable before the media is stored or shared. HemayaWajh provides peace of mind and convenience when taking photos or videos in a public event or setting. By processing all visual data locally on the hardware, it ensures that no imagery is exposed to or shared with any third-party systems.
Project ID = F25SDP 04 CE F
Supervisor: Dr. Khalid Abualsaud
Dema Lafi, Lojain Najjar, Asma Sakr, Areen Hayajneh
Anxiety and panic disorders are becoming a serious mental health concern in Qatar and across the Middle East. This rise is linked not only to global trends but also to local factors such as fast-paced social and economic change, stressful work environments, and the stigma that still surrounds seeking mental health care. A national survey conducted between 2019 and 2022 found that 28.0% of the population had a lifetime diagnosis of a mood or anxiety disorder, and 21.1% experienced such conditions within 12 months (Al-Marri et al., 2024). This project presents the design and development of a smart wearable system for the real-time detection and relief of panic attacks. The proposed solution consists of a wristband equipped with physiological sensors to continuously monitor heart rate, skin conductance, and movement patterns. When abnormal signals associated with the onset of a panic attack are detected, the device activates a vibration-based calming mechanism to help reduce symptom severity and provide immediate support. An integrated mobile application enables users to track and record panic-attack frequency and intensity, customize vibration patterns, access relaxation tools such as guided breathing and soothing sounds, and share data with healthcare professionals for clinical follow-up. The system also enhances user safety by optionally notifying trusted contacts during extended events. By integrating hardware-based monitoring with software-driven support, this project provides a comprehensive approach to panic- attack management, seeks to improve quality of life, and contributes valuable continuous data for digital-health research and practice.
Project ID = F25SDP 05 CE F
Supervisor: Dr. Khalid Abualsaud
Moudhi Al-Mohanadi, Muneera Al-Mulla, Aljazi Al-Zeyara, Shamma Al Kaabi
Atrial fibrillation (AFib) is one of the most common types of cardiac arrhythmia and is associated with serious complications such as stroke, heart failure, and sudden cardiac events. Early detection and continuous monitoring are essential, particularly for individuals at high risk, including patients with prior cardiac conditions or a family history of arrhythmia. This project presents a wearable smart wristband designed for continuous real-time monitoring of physiological signals related to AFib. The system integrates a photoplethysmography (PPG) sensor (MAX30101) to measure heart rate (HR) and blood oxygen saturation (SpO₂), and applies coefficient of variation (CV) analysis of heartbeat intervals to detect irregular heart rhythms. The coefficient of variation (CV) is a statistical measure that represents the variability of heartbeat intervals relative to their average, enabling reliable identification of irregular patterns. The system is implemented using an ESP32-S3 microcontroller for data acquisition, processing, and wireless communication. When abnormal conditions are detected, the device generates alerts through a buzzer and mobile notifications via the Blynk IoT platform. A single tap on the touchscreen allows the user to manually trigger an emergency alert. Additionally, a NEO-6M GPS module is integrated to provide real-time location tracking, which is shared through the mobile application during emergency events. All physiological data, including HR, SpO₂, and CV, are displayed on the integrated screen and transmitted in real time to the mobile application for monitoring and logging. The proposed system provides a low-cost, non-invasive, and portable solution for continuous cardiac monitoring, supporting early detection, improved patient safety, and remote healthcare applications.
Project ID = F25SDP 06 CE F
Supervisor: Prof. Abdelaziz Bouras
Shamma Al Hajri, Najd Al-Turki, Alghalya Alyafei, Shaikha Alyafei
This project presents the design and implementation of a Digital Twin–based robotic sorting system for integrated robotic applications. Traditional robotic systems often lack real-time monitoring, safe testing environments, and efficient methods for performance evaluation, which limits system optimization and increases operational risks. The primary objective is to develop a virtual replica of a physical robotic system that accurately reflects real-time operations and enables efficient monitoring, control, and performance optimization. The proposed system establishes a seamless connection between the physical and virtual environments, allowing users to simulate, analyze, and improve system behavior in a safe and controlled manner. The developed system integrates both hardware and software components within a unified framework. The physical system performs pick-and-place operations using sensors, actuators, and an embedded controller that processes real-time data. A camera module is used for object detection and color classification, while the embedded controller gathers sensor inputs and transmits them to the Digital Twin environment. The virtual model, implemented using Unity, mirrors the behavior of the physical system by updating its state dynamically based on incoming data, ensuring accurate synchronization between the two environments. The integration between the embedded system and the Digital Twin is achieved through continuous data exchange and control feedback, enabling real-time tracking and improved system coordination. This approach enhances system visibility, reduces reliance on physical testing, and supports safer experimentation and validation of control logic. Key achievements of this project include reliable synchronization between the physical and virtual systems, improved monitoring capabilities, and a user-friendly visualization interface that supports system analysis and testing. The system also demonstrates the ability to evaluate performance and identify potential issues before deployment in real-world scenarios. Overall, the proposed solution provides a practical and scalable Digital Twin implementation that enhances system efficiency and supports real-time decision-making. Unlike conventional approaches, the system integrates real-time bidirectional communication between a physical robotic platform and a Unity-based virtual model using continuous data synchronization, enabling accurate system replication and interactive control. This novelty allows safer testing, improved system analysis, and reduced dependency on physical trials, contributing to advancements in automation, predictive maintenance, and robotics education.
Project ID = F25SDP 07 CE F
Supervisor: Dr. Mohamed Al-Meer
Maryam Almarri, Sara Alsulaiti, Samar Mohamed, Aisha Alkhulaifi
The rapid growth of Internet of Things (IoT) technologies has increased the demand for reliable and real-time location tracking systems. This project presents a low-cost, real-time GPS tracking solution based on the ESP32 microcontroller and the LilyGO SIM7600G LTE development board. The system acquires GPS coordinates and transmits them over a cellular LTE network using the MQTT protocol, enabling continuous tracking without reliance on Wi-Fi. This ensures stable connectivity and real-time data delivery across different locations. The collected data is visualized through a custom-developed mobile application, allowing users to monitor location and movement dynamically. To enhance safety, the system incorporates a geofencing mechanism that triggers automatic alerts, including a Google Maps link, when the tracked object exits predefined boundaries. This feature enables immediate user awareness and response in critical situations. Overall, the project demonstrates a scalable and cost-effective IoT-based solution that integrates embedded systems, cellular communication, and mobile application technologies. The system is particularly suited for applications such as family driver monitoring and vehicle tracking, providing a practical and deployable approach to real-time safety and supervision.
Project ID = F25SDP 09 CE F
Supervisor: Dr. Mohammed Al-Sada
Mirna Abou Tafish, Enas Ghilan, Baraah Qafisheh, Leen Sukik
The increasing number of misplaced personal items in busy indoor environments such as airports, shopping malls, universities, and offices presents significant challenges for efficient item recovery and management. Traditional lost-and-found systems rely heavily on manual reporting, human intervention, and delayed retrieval processes, which can result in inefficiencies and reduced service quality. This project addresses these limitations by developing an autonomous Lost and Found Robot that integrates navigation, perception, manipulation, and secure user interaction into a unified system. The objective of this project is to design and implement a robotic system capable of autonomously patrolling indoor environments, detecting unattended objects, retrieving them using a robotic arm, and securely storing them for future collection. The system further extends beyond conventional robotic retrieval by incorporating a user authentication mechanism to ensure secure and accurate item return. The developed system employs autonomous navigation to move through predefined waypoints while continuously scanning the environment for objects using a vision-based detection module. Once an object is identified, the robot approaches it and executes a grasping sequence using a robotic manipulator. The retrieved item is then placed in a locked internal storage compartment, where it remains securely stored until claimed by its rightful owner. For item retrieval, a chatbot-based interface is used to interact with users and verify ownership through descriptive matching with previously recorded detection data. Upon successful verification, the system generates a unique PIN code. The user is required to visit the robot’s home position and enter the PIN on a keypad interface to unlock the storage compartment and retrieve the item. After completion of the retrieval process, the robot resumes its patrol and monitoring operation. Experimental testing conducted in controlled indoor environments demonstrated that the system successfully performs autonomous navigation, reliable object detection, stable manipulation, and secure retrieval operations. The system achieved consistent performance across multiple trials, confirming its ability to operate effectively in structured indoor scenarios. The novelty of this design lies in the integration of an autonomous mobile robotic platform with a real-time object detection system, a robotic manipulation unit, and a secure authentication based retrieval mechanism. Unlike traditional lost and found approaches that depend solely on human intervention, this system provides an end-to-end automated workflow that not only identifies and collects objects but also ensures controlled and verified access during item retrieval. This integration enhances operational efficiency, improves security, and reduces human workload, making the system a scalable solution for modern indoor environments.
Project ID = F25SDP 10 CE F
Supervisor: Dr. Hela Chamkhia
Wadha Al-Mefqaei, Hemayan Al-Mohamed, Amna Al-Thani, Aldana Almarri
This project presents the design and implementation of HomePulse, an intelligent smart energy monitoring and control system that integrates real-time sensing, automation, and artificial intelligence to improve energy efficiency, safety, and user awareness. The system addresses the growing need for real-time energy management in residential environments, where users often lack visibility into power consumption patterns, anomalies, and environmental conditions. The proposed solution combines hardware and software components into a unified platform. An ESP32-based system collects real-time electrical parameters such as voltage, current, power, and energy consumption, along with environmental data including temperature and humidity. This data is transmitted to a Raspberry Pi server running a FastAPI backend, which processes, stores, and visualizes the information through a web-based dashboard. To enhance system intelligence, anomaly detection techniques are implemented using machine learning models, including Isolation Forest and Autoencoder-based approaches. These models analyze sensor data to detect abnormal patterns such as unusual power spikes or irregular consumption behavior. In addition, a rule-based spike detection mechanism identifies sudden increases in power usage and logs them with precise timestamps and durations. The system also includes an AI-powered chatbot that allows users to query system data, receive insights, and trigger emergency alerts. A key feature of the system is its automated control capability. The system can automatically activate or restrict device operation based on environmental conditions, such as extreme temperatures or occupancy detection using a camera-based person detection module. Safety mechanisms, including relay locking and email alert notifications, ensure that the system responds proactively to critical situations. The results demonstrate that HomePulse effectively provides real-time monitoring, anomaly detection, and automated control within a single integrated platform. The system improves user awareness of energy consumption patterns and enhances safety through intelligent alerts and automation. The novelty of this project lies in the integration of real-time IoT monitoring, AI-based anomaly detection, occupancy-aware automation, and an interactive chatbot within a single, scalable system. This approach offers a practical and cost-effective solution for smart home energy management, with potential applications in residential, commercial, and smart city environments.
Project ID = F25SDP 11 CE F
Supervisor: Dr. Muhammed Azeem
Raghad Abed, Dana Al-Mushiri, Rawan Alsati, Jaman Eltatr
Disasters and humanitarian crises create hazardous environments that endanger civilians and first responders. This project presents the design and development of a multi-functional drone for use in disaster situations, providing real-time situational information and enabling early hazard detection in complex and insecure environments. The system aims to support emergency response teams by enabling remote monitoring and rapid identification of critical threats. The proposed system utilizes the Jetson Nano as the central processing unit for edge-based computation. It is integrated with the OAK-D Pro wide, which combines multiple vision capabilities, including real-time human detection, depth perception, and night vision within a single module. In addition, the system incorporates an MQ-9 gas sensor for hazardous environment monitoring, enabling the detection of gases such as carbon monoxide and combustible gases in post-disaster scenarios. The system is further supported by a Pixhawk 6C flight controller for stable autonomous flight and an M9N GPS module for accurate positioning and navigation. All sensor and imaging data are processed locally and transmitted to a remote base station via a wireless Wi-Fi connection and telemetry link ensuring low-latency and reliable data streaming. The key novelty of this design lies in its optimized and unified architecture, which replaces a multi- sensor, multi-camera setup with a single intelligent vision system supported by integrated sensing and navigation modules. This reduces system complexity, improves reliability, and enhances real- time performance. By combining advanced AI-based perception with environmental sensing, autonomous flight control, and efficient communication capabilities, the proposed drone offers a scalable and practical solution for post-disaster assessment, ultimately contributing to improved safety and operational efficiency in emergency response missions.
Project ID = F25SDP 12 CE F
Supervisor: Dr. Loay Ismail
Tahani Almarri, Rowaa Khaled, Abir Sidilemine, Abeer Tag
Elders commonly face challenges adhering to prescribed medication regimens, making noncompliance a critical health concern that leads to severe health consequences and diminished quality of life. Noncompliance patterns vary, including missing scheduled doses, prematurely discontinuing medication when immediate results are absent, substituting drugs with folk remedies, or dangerously ingesting excessive doses These issues are frequently caused by physical limits, memory deterioration, and the complexities of handling many medicines without assistance. As a result, many older people struggle to manage and schedule their prescriptions on their own, emphasizing the importance of dependable support systems to guarantee regular dosage administration. This project aims to solve these problems by creating a dependable technology solution that helps older people manage their daily drugs safely and effectively. The suggested system combines hardware and software components to enable automatic dose scheduling, timely reminders, and precise drug distribution. Moreover, the system utilizes a microcontroller-based automated dispenser integrated with a sensor suite and an authentication system to recognize the user, ensuring that each dose is given at the exact time and quantity. This precise process, further managed by a mobile robot, eliminates human error and significantly increases adherence to medical prescriptions. The project's goal with this invention is to improve medication compliance, protect the health of the elderly, and promote independence. By merging current assistive technology with user-friendly design, the solution provides a realistic way to enhance aged care, lower hospital readmissions, and lessen the strain on caregivers and healthcare systems.
Project ID = F25SDP 13 CE F
Supervisor: Prof. Abdelaziz Bouras
Alhanouf Al-Mesaifri, Hissa Al-Qahtani, Amna Almansoori, Noora Alyafei
This project presents the design and implementation of an integrated system for the acquisition and three-dimensional visualisation of historical coins using a combination of photogrammetry, embedded control, and holographic display technologies. The main problem addressed is the difficulty of accurately capturing fine surface details of historical coins, such as inscriptions and textures, using traditional 2D imaging methods, while minimizing physical handling of fragile artifacts. To address this challenge, a complete end-to-end solution was developed. The system employs a camera-based acquisition setup in which coins are placed on a microcontroller-controlled rotating turntable to ensure consistent and repeatable image capture from multiple angles. High-resolution images covering a full 360° view are collected and processed using photogrammetry software to generate accurate 3D models that preserve the geometric and visual characteristics of the coins. The developed system was implemented and evaluated in SDP II, demonstrating its ability to produce high-quality 3D reconstructions suitable for cultural heritage applications. The results show that the system can effectively capture fine details and achieve a level of accuracy sufficient for documentation and analysis. In addition, different photogrammetry tools were compared in terms of processing time, reconstruction quality, and model fidelity. The generated 3D models are presented through an interactive holographic display, allowing users to explore the coins from different perspectives using rotation and zoom functions. Furthermore, a web-based platform was developed to provide access to coin information, 3D models, and an AI- assisted interface for user interaction, enhancing accessibility and engagement. Overall, the project delivers a cost-effective, scalable, and integrated solution that combines image acquisition, 3D reconstruction, visualisation, and user interaction in a unified system. The proposed approach contributes to the digital preservation of cultural heritage and supports educational, research, and public engagement applications. Keywords: 3D Acquisition, Photogrammetry, 3D Modelling, Historical Coins, Cultural Heritage, Holographic Visualisation, Numismatics
Project ID = F25SDP 14 CS F
Supervisor: Dr. Mahmoud Barhamgi
Arwa Alobeid, Roqayah Ata omar, Hala Hamdoun, Haya Hamdoun
In today’s digital age, scam and phishing calls have become one of the most common methods used by cybercriminals to trick individuals into sharing personal or financial information. Many victims, especially children and elderly users, fall prey to social engineering tactics such as fake prize announcements, impersonation scams, or urgent money requests. Despite the availability of basic spam filters and caller ID systems, most existing solutions rely on static databases and fail to analyze real-time behavior or alert users effectively. This gap highlights the need for a smarter, more responsive system that can detect and warn users about potential scam calls before any harm occurs. To address this problem, our project introduces CallGuard, a mobile application designed to identify, analyze, and alert users about suspicious or fraudulent phone calls. The system combines three main layers of protection: crowdsourced reporting, automated backend validation, and real-time call analysis. Users can report suspicious numbers, and these reports are validated through a backend system that updates a central Threat Intelligence Database. The app also supports a parental alert feature, where parents are notified if a scam call targets their child. Additionally, an audio analysis module is designed to capture short call snippets and detect risky language or emotional pressure cues, enhancing the system’s ability to identify social engineering attempts in real time. During the first phase of the project (SDP1), our main focus was on system design and partial implementation. The team successfully produced all major design deliverables, including use case, class, sequence, activity, and state diagrams, a relational database schema, and a user interface prototype. About 30% of the system’s functional use cases were also implemented, covering key backend processes such as number reporting, report validation, and database updates. These implementations helped verify that the chosen client–server architecture and relational model function effectively as the foundation of the complete system. The CallGuard architecture follows a layered modular design, ensuring clear separation between the mobile client, backend services, and data layer. This design enhances scalability, maintainability, and security. The system leverages modern frameworks such as Flutter for cross-platform development, Visual Paradigm for UML modeling, and dbdiagram.io for database schema design. Through these tools, the team was able to visualize the system’s internal logic and ensure consistency across all design components. In conclusion, the CallGuard project addresses a pressing cybersecurity problem by proposing a practical, educational, and socially aware solution for detecting scam calls. The outcomes of SDP1 demonstrate strong progress toward achieving the project’s objectives, with a complete architectural design and a functional prototype of its core reporting module. The next development phase (SDP2) will focus on completing the remaining use cases, integrating real-time audio analysis, enhancing user interaction, and performing full system testing. Once completed, CallGuard aims to provide a reliable and intelligent defense against voice-based scams, helping protect users from manipulation and fraud in everyday communication.
Project ID = F25SDP 15 CS F
Supervisor: Prof. Khaled Khan
Hala Al-Dosari, Fatima Zahra Brahamia, Sara Ibrahim, Habiba Khattab
Cybersecurity threats such as malware, phishing, ransomware, worms, and other forms of malicious software continue to increase in frequency and sophistication, while many users still lack the awareness and practical skills needed to recognize and respond to them effectively. This problem is especially significant for students, public, and non-technical users, and it becomes even more pronounced in Arabic-speaking contexts, where much of the available cybersecurity education is either English-only, highly technical, or delivered through passive formats that do not adequately engage learners. In response to this gap, Cyber Majlis was developed as a bilingual, culturally relevant cybersecurity learning platform designed to make cybersecurity awareness more accessible, interactive, and effective for both Arabic and English speakers. Cyber Majlis combines a structured, lesson-based dashboard with interactive, gamified learning experiences to teach cybersecurity fundamentals in an engaging way. The platform organizes content into learning modules such as Basic, Intermediate, and AI. Each lesson follows a clear instructional flow: the learner first watches an animated short story to understand the concept, then views a short malware demonstration showing the real effect on a computer in a virtual machine environment, then reviews a poster that summarizes the lesson and provides safety advice and finally completes a quiz to check understanding. In addition to these core lessons, the platform extends learning through mini-games, malware simulations, a chatbot for guidance, and a live SOC-inspired environment that introduces users to real-world cybersecurity monitoring and response concepts. The system also includes bilingual navigation, profile management, progress tracking, and gamification features such as XP and rankings to support motivation and learner engagement. The project’s key accomplishments include the implementation of a functional prototype that supports user registration and login, bilingual content delivery, progressive lesson completion, quiz assessment, profile customization, and learning progress tracking. It also integrates interactive cybersecurity games, simulation-based activities, and a live Security Operations Center (SOC) experience that strengthens practical understanding through scenario-driven learning. Technically, the project is built as a modular, component-based web application with secure authentication and database-backed user data management, supporting maintainability and future expansion. Overall, Cyber Majlis demonstrates that cybersecurity education can be made more effective when it is bilingual, interactive, structured, and culturally relevant. The most important conclusion is that combining animated storytelling, practical demonstrations, gamification, and simulation-based learning can significantly improve learner engagement and help users develop stronger cybersecurity awareness and safer digital behavior.
Project ID = F25SDP 16 CS F
Supervisor: Dr. Saleh Al-Hazbi
Nafisa Nirjhor, Tasnova Tabassum, Maimouna Tonny, Shaikha Alkatheri
The campus transportation system of Qatar University faces inefficiencies like student delays, bus bunching, prolonged waiting times, and general dissatisfaction. This project addresses the need for a comprehensive, reliable, and accessible Campus Transport Management System. The proposed system aims to increase the efficiency of daily student commutes, enhance overall campus productivity, reduce waiting time, and promote sustainability by reducing idle travel. Thus, it contributes to better time management for students. The solution integrates an advanced tracking mechanism for both buses and routes using GPS-based monitoring to improve coordination across shared stations. Two interconnected mobile applications will be developed, one for students and another for drivers, along with a web application for the administration to monitor and manage all bus activities through a dashboard. The mobile applications will allow users to track buses, provide drivers with instructions, notifications and visually display the bunched path to prevent bus bunching. Students can also select a route to view and receive relevant notifications about the details of the specific route. Key achievements of the project will include an algorithm to detect and mitigate bus bunching and a decision support feature that will instruct drivers when to take action based on real-time conditions. This project not only provides an effective solution to Qatar University’s current transportation challenges but also offers a model that can be adapted by other organizations seeking to enhance the efficiency and reliability of their transportation systems.
Project ID = F25SDP 17 CS F
Supervisor: Dr. Saleh Al-Hazbi
Sadien Abu El-rub, Atrab Ali, Non Alkhidir, Arwa Elaradi
Effective teacher evaluation is essential for improving educational outcomes, especially for pre- service teachers. Yet traditional evaluation systems that rely on a mentor's physical attendance often face challenges such as subjectivity, delayed feedback, and limited insights into teaching effectiveness. These limitations hinder schools and institutions from identifying areas of improvement for educators in a timely and data-driven manner. Addressing this problem is critical, as teacher performance directly impacts student learning, engagement, and overall educational quality. To tackle these challenges, our project proposes a smart coaching system for enhancing teachers’ practices, designed to provide comprehensive, real-time, and objective assessments of teaching performance. The platform leverages Artificial Intelligence (AI) elements, specifically Large Language Models (LLM) and Computer Vision (CV), to analyze classroom observations, hence extracting and calculating performance metrics and actionable insights for teachers. The metrics used for generating feedback were carefully designed and validated by experts from reputable educational institutions such as the College of Education, ensuring their relevance, credibility, and alignment with established teaching standards. By combining advanced analytics with a user-friendly interface, the platform ensures feedback is not only accurate but also practical for continuous professional development. Key achievements of the project include the design and implementation of an adaptive evaluation system, that accommodates different teaching contexts and subjects from elementary up to university level by utilizing AI tools such as Soniox and Gemini. Additionally, it provides an intuitive dashboard for presenting insights in a clear and actionable format. Beyond its technical capabilities, a key achievement lies in its strong pedagogical foundation, as the system is designed to support reflective teaching practices, foster continuous learning and align feedback with effective instructional strategies. The platform successfully demonstrates how AI can provide timely feedback, and support teachers in improving their instructional practices. In conclusion, the project represents a significant advancement in the field of teacher professional development by offering a scalable, objective, and actionable guidance. The platform not only enhances the accuracy and efficiency of evaluations but also empowers educators to reflect on their practices, make informed improvements, and ultimately elevate the quality of education. This project highlights the potential of AI technologies in transforming traditional coaching processes and promoting continuous professional growth among teachers.
Project ID = F25SDP 18 CS F
Supervisor: Dr. Mucahid Kutlu
Aisha Al-Lenjawi, Bashayer Alnasser, Maryam Bawazir, Tamara Elasmar
Our project offers an educational platform aimed at helping students and non-students build their learning levels and enjoy the learning experience. The platform enhances learning through competitions, questions, and exams, covering a wide range of educational specializations to meet diverse user needs. The platform provides educational resources, including videos and reliable books, to allow broad exploration of topics. Users can answer questions of varying difficulty: easy, medium, and hard, to assess their learning level, identify weaknesses, and reinforce knowledge. Incorrect answers are followed by clear explanations to maintain engagement and understanding. We offer educational competitions with time limits to motivate users, develop quick thinking, decision making, cooperation, and retention skills. For engagement, two gamified features are included: a museum showing the user's academic level across subjects and a castle building game that reflects progress through earned points. The subscription model allows free access to the first easy level stage, with subsequent stages requiring a monthly subscription. A trial period of at least seven days is provided to familiarize users with the platform. Additional features include a chatbot to answer questions and a calendar for planning important dates like tests and exams. Challenges faced include adjusting questions based on student performance, providing reliable resources, avoiding question duplication, guaranteeing that the answers of the system are correct, ensuring platform uniqueness, protecting personal data, offering short explanatory videos, and supporting multiple languages (Arabic and English). Solutions include using a question tracking system to avoid repetition and ensure unique questions per user session, providing features that distinguish our website from others, encrypting sensitive user information, employing AI such as DeeVid ai [1] to convert text to speech and make videos short with concise explanations to maintain engagement. We aim for the platform to serve as a comprehensive educational tool, helping students, employees, and others learn effectively, practice, and achieve their goals.
Project ID = F25SDP 19 CS F
Supervisor: Dr. Mucahid Kutlu
Dana Al-Kaabi, Bashayer Al-Marri, Sara Almuraikhi, Marwa Attia, Safya Darfan
Qatar University students currently rely on fragmented and informal tools, such as WhatsApp groups, paper-based bus logbooks, and manual schedules, to coordinate academic activities, campus transportation, and social events. This fragmentation leads to communication gaps, uncertainty around bus arrival times, and inefficient management of study groups and calendars. No single digital platform exists that unify these services, creating a disjointed campus experience that affects student productivity and engagement. The Qconnect App was developed to address this gap by providing a cross-platform mobile application that integrates real-time bus tracking, course and study-group management, event scheduling, and a unified academic calendar into a single platform. Built using Flutter for the frontend and a Django REST Framework backend with PostgreSQL database, the system serves three user roles, students, bus drivers, and administrators, each with role-based dashboards tailored to their specific needs. Students can browse and create courses, form study groups, share materials, communicate through in-app chat, and track campus buses in real time on an interactive map. Bus drivers broadcast their GPS location and update occupancy levels through a simplified mobile interface, while administrators manage system content and monitor platform activity. The project followed the Agile-Scrum methodology across two semesters, progressing through iterative sprints covering requirements gathering, system design, implementation, and comprehensive testing. The team used an Excel-based Kanban board to track 22 use cases across eight development milestones, ensuring structured sprint planning and continuous refinement based on supervisor feedback. Functional test cases were executed across all modules, authentication, course management, calendar, and bus tracking, with a 100% pass rate. Performance testing confirmed that the average response time for critical operations (login, data retrieval) was 1.2 seconds, which is 40% faster than the 2-second target. Bus location updates were delivered to students within 3.2 seconds, well within the 5-second threshold. Reliability testing yielded a 99.5% crash-free session rate, exceeding the 99% target. All network traffic was verified to use HTTPS/TLS encryption, and role-based access control correctly restricted functionality based on user type. Usability evaluation with 10 Qatar University students produced a System Usability Scale (SUS) score of 82.5 out of 100, surpassing the 80-point goal, with a 94% task completion rate and an average navigation efficiency of 2.4 clicks per task. Accessibility testing confirmed a color contrast ratio of 5.2:1 (exceeding WCAG 2.1 AA requirements) and 97% screen reader label coverage. By consolidating essential campus services into one cohesive mobile ecosystem, the Qconnect App enhances communication, reduces transportation uncertainty, supports peer-driven academic collaboration, and contributes to Qatar University's broader digital transformation initiatives. The modular architecture ensures the platform can be extended with additional features or adapted for use by other universities.
Project ID = F25SDP 21 CS F
Supervisor: Prof. Tamer Elsayed
Sumaya Alawad, Carmela Chavez, Ganna Soltan, Bsmalla Mohamed
Autism Spectrum Disorder (ASD) can affect how children understand and respond to social and emotional cues, making emotion recognition a common area of difficulty. Traditional therapeutic approaches that address these skills are often repetitive and struggle to sustain engagement. To offer a more motivating and effective learning experience, this project presents Feelo, a gamified emotional-learning application that uses interactive play, exploration, and immediate feedback to support children’s development of emotional understanding. Feelo is an open world1 game that allows children to explore environments that replicate real-world situations and navigate various emotional interactions. The game enables them to learn to recognize emotions and apply this understanding in daily life. This learning is facilitated through challenges and quests, with rewards and feedback provided to maintain motivation and engagement. By combining technology with therapeutic objectives, the application aims to enhance emotional recognition skills, build social confidence, and provide caregivers and therapists with valuable insights into a child’s progress. Feelo primarily supports children with ASD levels one and two who are aged eight and above. The game demonstrates how gamified learning can address the limitations of traditional therapy by keeping children engaged while promoting essential social and emotional skills. Unlike existing tools, Feelo integrates therapist-approved emotional scenarios, coin-based reward system and story-driven quest, which further enhance motivation and encourage consistent skill practice. This approach offers a promising solution to improving emotional development in children with ASD and provides a foundation for further research and application in therapeutic technology.
Project ID = F25SDP 22 CS F
Supervisor: Prof. Khaled Shaban
Reem Aladbi, Layan Alwattar, Khawlah Daraan, Amna Osman
The organization and management of booth-based events, including conferences, exhibitions, and community events, present several challenges such as maintaining attendee engagement, tracking attendance and trends, and ensuring user privacy, while still providing organizers with effective control. Existing event management platforms typically fall at two extremes—either rigid and inflexible, or highly customizable but complex and difficult to use—creating a gap for solutions that balance engagement and flexibility. Eventura addresses this gap by offering a modular platform that combines intuitive design with flexible event management. It provides a dedicated mobile application for visitors and a web platform for organizers and exhibitors, ensuring a clear separation of roles while maintaining a seamless overall experience. Organizers can customize event structures, implement interactive features such as QR-based attendance tracking, booth challenges, and competitions, and monitor real-time activity through dashboards. Visitors benefit from features such as progress tracking and gamified participation, which encourage exploration and increase engagement. The system demonstrated strong performance, with QR code scans responding within 3 seconds and supporting up to 500 concurrent users without noticeable degradation. Usability testing also showed that new users were able to complete core tasks, such as event registration and check-in, within minutes without prior training. Key achievements include the successful development of a functional prototype supporting event customization, QR-based attendance, and interactive competitions. Eventura provides a scalable and user-centered solution that improves event engagement and management efficiency, demonstrating its potential for real-world adoption in modern event environments.
Project ID = F25SDP 23 CE-CS F
Supervisor: Prof. Sumaya Al-Maadeed
Wadha Al-Hemaidi, Alanoud Al-Thani, Fatima Al-Thani, Nouf Ali
This project presents the design, development, and evaluation of a multimodal hand gesture recognition system that integrates surface electromyography (sEMG), Leap Motion Controller (LMC) tracking, and mmWave Radar to achieve robust, real-time recognition of Arabic air-writing and Qatari Sign Language (QSL) gestures. The system addresses key limitations of existing gesture recognition approaches, which typically rely on a single sensing modality, such as vision-only or sensor-only input. Moreover, it fills a critical gap in Human-Computer Interaction (HCI) research by introducing a recognition system that supports Arabic air-writing and QSL, both of which remain underrepresented in hand gesture recognition technologies. The proposed system employs an 8-channel Mindrove EMG armband to capture electrical muscle activity from the forearm and a Leap Motion Controller to record 3D hand and finger coordinates. Data streams from three sensors that are synchronized before feature extraction. Time-domain features are extracted from EMG signals, spatial features are derived from LMC hand joint coordinates, and radar feature representation obtained from MmWave Radar. This pairing enhances assistive communication and HCI. To support multimodal data acquisition, a data collection application was developed using Python to record synchronized EMG and LMC signals in addition to, utilizing MmWave Studio to capture Analog- to-Digital Converter (ADC) raw digital data from the MmWave radar. The applications enables controlled data collection from multiple users with precise alignment between both inputs, as well as real-time visual feedback to ensure data quality and consistency. This enabled the creation of three multimodal datasets, one for Arabic air-writing and another for Qatari Sign Language for supervised training and testing. The application ensured consistent synchronization accuracy, forming the foundation for subsequent model development. These applications serve as a reusable solution for future multimodal dataset creation and research on gesture-based interfaces. A feature-level fusion method is then used to combine all modalities into a unified dataset. Various machine learning classifiers including Support Vector Machines (SVM), Random Forest (RF), and Artificial Neural Networks (ANN) are tested and evaluated. The recognition application was implemented in Python, using modules for real-time data acquisition, preprocessing, classification, and a graphical user interface (GUI) that displays EMG channel activity, 3D hand trajectories, and recognized gesture outputs. Moreover, range-Doppler maps, micro-Doppler spectrograms, range- angle heatmaps, and 3D point cloud trajectories displayed at PostProc of MmWave Studio. The system was tested using evaluation metrics such as accuracy, precision, recall, and F1-score, to assess system performance. For the data collection process, data were collected from 18 participants, each performing 28 static QSL gestures and 3 dynamic Arabic air-writing characters, each repeated 5 times per sample. Results showed that the fusion of combined modalities showed results of consistently improving performance, achieving an accuracy up to 89% for QSL and 86% for Air-writing . This work supports social inclusion and accessibility by enabling Arabic-speaking and hearing- impaired users to interact naturally with digital systems. The novelty of the project is the fusion of physiological (EMG), spatial (LMC), and electromagnetic sensing (MmWave radar) modalities, its localization to Arabic and Qatari contexts, and its implementation of a real-time recognition system. The resulting system establishes a foundation for future research in Arabic air-writing recognition and QSL recognition, as well as for developing culturally aligned assistive technologies aligned with Qatar’s digital transformation goals.
Project ID = F25SDP 24 CE-CS F
Supervisor: Prof. Elias Yaccoub
Latifa Al-Emadi, Ahya Hamamah AlWattar, Reem Mohammed, Sara Namshan
Health tracking in various fields is increasing in demand, especially where the core of the daily tasks includes regular and consistent strenuous activity. Construction work is such a field, requiring frequent exposure to the sun due to the outdoor nature of the job, as well as demanding physical work. This leads to increasing risks of heat stress, fatigue, and muscular overloading. As such, accurate health monitoring and reducing muscular exertion are vital to ensuring worker safety and steady productivity, particularly in regions such as Qatar, known for high-heat summers. To address these challenges, this design project presents a lightweight, portable, upper-body lightweight exoskeleton with health monitoring capabilities as well as a lifting-assistance system for construction workers. The design monitors health parameters, offers muscular load support, and predicts heat stress using a deep learning model by extracting key physiological data. Sensor data from multiple sensors including a heart rate sensor (MAX30102), body temperature sensor (MAX30105), environmental temperature and humidity sensor (DHT22), and an inertial measurement unit (IMU) are used to give an overview of the worker’s current health and activity levels. The data from these sensors are transmitted to a Flutter mobile application, displaying each worker’s personal health statistics and reports, while also allowing site supervisors to monitor worker status and respond quickly in case of emergency, abnormal readings, extreme overexertion, or early signs of heat stress. A predictive convolutional neural network (CNN) and bidirectional long short term memory (BiLSTM) deep learning algorithm is also used to predict heat stress based on derived heart rate, ambient temperature, and motion readings, achieving an accuracy of 91.8%, allowing early detection and prevention of possible risks. The system displays warnings in the form of alert notifications for immediate action when required by both the worker and the supervisor. In order to establish connectivity between the sensors and the application, Wi-Fi is used to transmit data from the microcontroller (ESP32 WROOM-DA Module) directly to the mobile application, where data is stored and retrieved from a Firebase database. Additionally, the system includes a lifting-assist mechanism designed to reduce the load on arm muscles during heavy and repetitive lifting tasks, minimizing fatigue and improving productivity. The developed subsystems demonstrated successful operation during testing, forming the foundation for the complete integrated system. Overall, this proposed system combines wearable biomechanical support, sensor integration, and mobile application development into one cohesive system aimed at enhancing safety on construction sites, improving health monitoring and early prevention, and providing muscular assistance to prevent fatigue in extreme climate conditions.
Project ID = F25SDP 25 CE-CS F
Supervisor: Dr. Devrim Unal
Amna Adam, Fatima Ahmed, Dimah Alsaodi, Nooralhuda Soueid
The increasing adoption of autonomous drones in logistics, surveillance, disaster response, and environmental monitoring has highlighted the critical need for secure and resilient communication systems. As drones evolve from remote-controlled devices to pre-programmed autonomous systems, reliable and secure coordination becomes essential, particularly in mission-critical and adversarial environments. Traditional cryptographic schemes, such as RSA and ECC, face growing vulnerabilities due to advances in quantum computing, posing significant risks to drone swarms. Post-Quantum Cryptography (PQC) has emerged as a quantum-resilient solution, with algorithms such as CRYSTALS-Kyber and Dilithium demonstrating efficiency on resource-constrained embedded platforms. Our project proposes combining a standardized, lightweight PQC scheme with a secure protocol to provide quantum-resilient, authenticated communication across drone swarm topologies, aiming to reduce drone swarm accidents and attacks. The design does not assume elimination of inherent topology-level vulnerabilities; instead, it focuses on strengthening communication-layer security within realistic operational constraints. This project addresses the gap between secure communication and practical swarm deployment by developing a fully functional application for simulating and managing secure drone swarm operations. The system supports the import and execution of mission plans from QGroundControl (QGC), enabling realistic and flexible mission configuration. A Proof-of-Concept (PoC) was implemented in which multiple drones communicate securely using standardized PQC algorithms alongside a lightweight protocol optimized for low-latency swarm communication under strict size, weight, and power constraints. The platform additionally incorporates real-time object detection through an onboard camera module, with detection data transmitted securely over the encrypted communication channel. The platform integrates a comprehensive simulation environment capable of modeling drone swarms across multiple agents using the PX4 SITL simulator. In addition to normal operation, the system enables controlled experimentation under adversarial conditions, including man-in-the- middle attacks, eavesdropping, replay attacks, distributed denial-of-service (DDoS) attacks, algorithm downgrade attacks, and session theft scenarios. This allows for systematic evaluation of swarm coordination, navigation, and secure data exchange in the presence of real-world attack vectors across five swarm topology modes. Performance is assessed through metrics such as attack block rate, trajectory deviation, session establishment latency, and communication reliability, providing insight into the practical impact of PQC on autonomous swarm systems. Results demonstrate that the system achieves a 100% attack block rate under PQC protection with no measurable impact on mission delivery success. The result is a reusable and extensible platform for secure drone experimentation, bridging the gap between theoretical post-quantum security and real-world swarm deployment. This work contributes toward the development of resilient, real-time, and mission-ready autonomous aerial systems for academic, industrial, and applied applications.
Project ID = F25SDP 26 CE-CS F
Supervisor: Prof. Sumaya Al-Maadeed
Nagham Al-Ajmi, Lama Al-Dosari, Maryam Al-Kaabi, Noor Al-Kaabi
One of the most common types of tumors that many women suffer from worldwide is breast cancer. Early detection plays a vital role in survival rates and assists in decreasing the progression of cancer. Conventional methods like mammography, ultrasound, and MRI are effective and have been popular for a while now, although they do have limitations. Mammography is one of the most used methods of screening; however, frequent screenings could become unsafe as it is radioactive. Ultrasound is another screening tool used for breast cancer detection, but results can vary depending on the operator. While MRI is effective and safe, it is quite costly and not always accessible. These limitations necessitate the need to develop a safe, affordable, and non-invasive prototype. Unlike the current screening methods, our device uses a millimeter-wave (mmWave) radar as a non- ionizing and non-invasive alternative. The mmWave radar can detect abnormalities safely and comfortably. Our project will involve both the use of mmWave radar and deep learning classification techniques. We used two types of mmWave radar modules including the Texas Instruments IWR1843 and Infineon’s BGT60TR13C Development Kit. We aimed to assess the capabilities of the mmWave radar in detecting any abnormalities in breast tissue. We produced a mechanism that would allow the mmWave radar to move along the x-axis and y-axis. We tested these movements on artificial breast tissue to evaluate the radars’ consistency and accuracy. Moreover, we applied deep learning and machine learning to enhance detection accuracy. We trained both deep learning and machine learning models to utilize multimodal data for accurate breast tumor detection and diagnostic support. The mmWave radar is a non-invasive and safe system used for breast cancer detection, and the machine and deep learning models are trained on multimodal data to support automated tumor detection. It’s a safe device that makes frequent screenings safe. This project provides a safe, non- invasive, and fast approach to detect breast cancer.
Project ID = F25SDP 27 CE M
Supervisor: Prof. Elias Yaccoub
Mohamed Farag, Khalid Haji, Rufus Kadalikandathil, Faris Khalil Ur Rehman
This project presents an autonomous smart shopping cart created to make shopping easier, safer and more efficient. Traditional carts are passive, they do not track items, assist users, or provide any real automation. This results in slow checkouts, crowded aisles, and difficulties for elderly shoppers or those with mobility challenges. To address these issues, a system was developed that integrates autonomous navigation, RFID-based product tracking, and real-time communication between the cart’s hardware and an interactive, informative mobile app. At the core of the cart is Raspberry Pi 5, which handles all primary processing tasks. For navigation and obstacle avoidance, A RPLIDAR A1M8 is used, enabling the cart to use a built-in map of its environment to navigate smartly. The cart moves using high-torque 12V DC gear motors, controlled by an L298N motor driver, and powered by a quad-battery setup regulated by a buck converter. The cart automatically takes the user to their desired location given a command via the app. Simultaneously, an RFID reader records every item placed in the cart, eliminating the need for manual scanning at checkout. All data syncs to a companion mobile app, where users can view their selected items and receive navigation guidance. Ultimately, this project stands at the intersection of robotics, embedded systems, and wireless technology. It is a clear example of how thoughtful engineering can make a real impact. Transforming something as everyday as shopping into an experience that is smarter, more accessible, and more convenient.
Project ID = F25SDP 28 CE M
Supervisor: Dr. Ahmed Badawy
Abdallah Alkanani, Ali Ghazi, Khaled Qarawi, Abdulrahman Shabban
People with upper-limb disabilities often face limited, costly options for eating independently, many of which can challenge user dignity. Existing assistive feeding solutions commonly depend on caregiver assistance, manual user input, or predefined robotic motions, limiting independence and adaptability during real meals. This Senior Design Project addresses this problem by developing ITA’AM, a low-cost autonomous feeding assistant designed to support independent eating while prioritizing safety, usability, hygiene, and user dignity. The proposed system integrates a robotic arm with a fork end-effector, computer vision, facial-gesture interaction, motor-current-based contact feedback, ROS 2 communication, and reinforcement-learning- based control. The perception subsystem detects the plate, food pieces, and user facial cues such as mouth opening, blinking, and eyebrow movement. These signals support hands-free interaction and provide target information for the robotic control system. The system architecture was developed around affordable and modular components, including the SO-101 robotic arm platform, camera-based sensing, 3D-printed utensil mounting, and open-source robotics and AI frameworks. During development, the project validated several important parts of the feeding pipeline. The facial- gesture detection module was improved through calibration, the food detection module was designed to identify plate regions and food-piece centroids, and the calibrated vision pipeline converts image detections into physical plate-plane coordinates that can be used by the arm controller. Reinforcement learning policies were trained and evaluated in simulation for pick-and-feed behavior, showing improved motion behavior, safer approach patterns, and progress toward autonomous bite acquisition and delivery. In addition, simulation logs and policy-output traces provide reference trajectories that can be mapped into joint- level commands for the SO-101 arm, supporting the ongoing sim-to-real integration process while direct deployment of the learned model on the physical arm is still in progress. The main contribution of this project is an end-to-end proof-of-concept architecture for an affordable AI-assisted feeding robot that combines real-time perception, natural hands-free user triggering, adaptive robotic control, and safety-aware design. The current prototype demonstrates the feasibility of the proposed approach through validated subsystem-level results, simulation-based reinforcement learning, and an integrated ROS 2 architecture, establishing a strong foundation for continued physical validation, sim-to-real refinement, broader food handling, and future supervised testing in realistic assistive- feeding environments.
Project ID = F25SDP 29 CE M
Supervisor: Prof. Amr Mohamed
Rami Abendeh, Abdulla Al-Hussaini, Mohammed Alharbi, Abdelrahman Hasan
Natural disasters such as earthquakes and building collapses often leave survivors trapped beneath debris, where rapid detection is critical for saving lives. Traditional search-and-rescue operations face major limitations, including restricted access, safety hazards for personnel, and significant delays in locating victims. This project introduces D-SCAN, a multi-drone system designed to improve the speed and reliability of survivor detection in post-disaster environments. The system employs Frequency-Modulated Continuous Wave (FMCW) radar at 24 GHz as the primary sensing tool, with drones operating independently in coordinated pairs to scan assigned building sections and transmit findings to a central monitoring hub for visualization and decision-making. To validate the approach, the project combines simulation using the NVIDIA Sionna RT framework with hardware prototyping of a drone equipped with radar. Our experiments yielded several key takeaways. First, in simulation the dual-frequency scanning strategy (6.1 GHz and 6.3 GHz) enabled parallel data collection without interference, reducing total scan time by approximately 50% while maintaining full 360° coverage. Second, the ray-based labeling strategy, where human presence was defined by more than 10% of traced rays interacting with the human model, achieved the highest test accuracy of 84% using XGBoost, demonstrating that strong multipath density reliably indicates human presence. Third, the CIR difference-based labels (MAD and correlation) achieved recall values of 0.65–0.68 at low decision thresholds, confirming that the system can prioritize sensitivity to minimize missed detections—the most critical error in rescue operations. Fourth, on hardware, the Gradient Boosting classifier achieved 89% accuracy on data collected from a two-fl or box structure with an average inference time of 10.2 seconds per 40-frame sequence. Fifth, comparison between simulation and hardware results revealed that while simulation provides a robust foundation for model training, real-world performance depends on additional factors such as debris density and environmental interference.
Project ID = F25SDP 31 CE M
Supervisor: Dr. Ahmed Badawy
Ahmad Abdalla, Moustafa Abdelwahab, Marwan Amin, Mohamed Kechaou
Gas leaks in industrial environments such as oil and gas facilities pose serious threats to safety, the environment, and economic operations. Traditional detection systems, typically fixed in place, can identify leaks only within their sensor range and require manual intervention for containment. Meanwhile, mobile robotic inspection systems like drones or quadrupeds primarily focus on observation rather than action, lacking the ability to physically respond to emergencies. To address these limitations, this project developed a Hazardous Autonomous Robotic Intervention System (HARIS), an autonomous quadruped platform equipped with a robotic arm, integrated sensor suite, and computer vision and deep reinforcement learning (DRL) based perception system for end-to-end leak detection and mitigation. The implemented system uses a Unitree Go2 quadruped as the mobile base and a custom-built 3D- printed robotic arm for manipulation tasks. Raspberry Pi and Arduino microcontrollers handle real- time sensing and control, supported by a Raspberry Pi Camera Module 3 for visual detection. The onboard perception model is based on YOLOv11, trained on a custom dataset of about 1,000 valve images (open and closed states) to enable accurate recognition of industrial lever valves under varying lighting and orientations. Complementary gas sensors (MQ series) and environmental sensors provide multimodal data for leak detection and confirmation. Once a leak is detected, the robot navigates toward the source and uses the arm to autonomously close the main valve, thereby isolating the hazard. For autonomous decision-making and control, the project employs Deep Reinforcement Learning (DRL) algorithms. The Soft Actor-Critic (SAC) and Proximal Policy Optimization (PPO) methods were evaluated for arm manipulation, decision-making, and navigations. Both algorithms were trained and validated in simulation using the exported URDF model of the arm and then deployed for real-world testing. Simulation environments were developed in Mujoco, enabling iterative training and performance comparison under controlled conditions. Key evaluation metrics such as success rate, detection accuracy, and control latency were used to assess system performance. The project successfully achieved autonomous leak detection, localization, and physical valve closure in a controlled setup. The sensor suite and vision models demonstrated reliable real-time performance on edge hardware, while the trained DRL models exhibited stable learning and adaptability across multiple tasks. The integration of vision, sensing, and reinforcement learning within a legged robotic platform represents a novel step toward fully autonomous industrial safety systems. Unlike existing stationary detectors or inspection robots, HARIS provides a complete response solution, capable of perceiving, deciding, and acting in dynamic environments. This innovation highlights the potential of intelligent, mobile robotic systems to enhance safety, reduce human exposure to hazardous zones, and set new standards for autonomous maintenance in the oil and gas sector.
Project ID = F25SDP 32 CE M
Supervisor: Prof. Uvais Qidwai
Alabass Abusbiha, Faleh Al-Harami, Abdelrahman Mohamed, Abdelhafid Smatti
To stop diseases produced by germs that live on surfaces, hospital surgical suites must follow strict disinfection rules. Because of the shadows cast by furniture and equipment, standard Ultraviolet-C ceiling lights don’t completely disinfect rooms. This project solves this problem by developing and building an autonomous Ultraviolet-C (UV-C) Light Disinfection Robot that can find surfaces that haven’t been exposed and disinfect them safely for healthcare workers. The suggested system combines autonomous navigation, UV-C dosage control, human detection, and energy optimization into a small robotic platform. The robot has a movable base and a linear actuator with attached gimbal the holds a UV-C lamp so it can get to hard-to-reach or shadowed locations. It employs cameras and sensors on board to map the environment, find shadows, and recognize people in real time. When the system sees that the room is vacant, it turns on the UV-C lamps to kill germs on surface. If it sees motion or a person, it turns off the lamps immediately. The robot also has a graphical user interface (GUI) that lets hospital workers keep an eye on procedures and make reports. The design follows technical rules that focuses on safety, dependability, and efficiency while also taking into account cost and simplicity of manufacturing. During the system’s development, the project looked at and prepared for five main design constraints: compliance with radiation safety standards, accuracy of navigation, energy and thermal efficiency, cost-effectiveness, and ethical use. By combining robotics, embedded systems, computer vision and healthcare safety standards from different fields, the project makes sure that the robot can work on its own without any help from humans and meets all the technical and legal requirements. Market research shows that there is a rising need for automated disinfection systems, notably in hospitals, labs, and congested public places after the COVID-19 pandemic. The suggested UV-C robot is a cheap and easy-to-make alternative to imported devices that nevertheless works well for cleaning and is safe to use. The design helps the environment by using fewer chemical disinfectants and automating healthcare operations, which lowers expense. This project shows the whole engineering process, from figuring out a healthcare problem to making, testing and building a prototype solution. The technology is new because it combines targeted linear actuator with attached gimbal UV-C disinfection, real-time human detection, and fully autonomous navigation into a small, energy-efficient platform. As a result, there is a dependable, safe, and long- lasting robotic solution that will improve infection control and help Qatar move forward in a smart facility and healthcare automation technology.
Project ID = F25SDP 33 CE M
Supervisor: Prof. Uvais Qidwai
Abdulla Alansari, Majd Alhakim, Amar Badran, Amr Mohamed
Indoor localization is a critical challenge in environments where Global Positioning System (GPS) is unavailable or unreliable, such as shopping malls, hospitals, and schools. In such spaces, ensuring child safety and enabling parents to locate their children quickly is a growing need. Traditional indoor localization solutions often rely on costly infrastructure, complex calibration, or pre-built maps, making them unsuitable for rapid deployment in dynamic public spaces. This project addresses this challenge by designing and implementing TrakN, a low-cost real-time indoor child localization system, based on a wearable two-board IoT tag, a cloud backend, a venue mapping workflow, and Android applications for deployment and live tracking. The system uses a Beetle ESP32-C6 main tag board to sample an Inertial Measurement Unit (IMU) the MPU6050 sensor at 100 Hz and post data securely to a cloud server, while a second Beetle ESP32-C6 board performs passive Wi-Fi scanning every approximately 10s and forwards Received Signal Strength Indicator (RSSI) observations. On the backend, the system tracks each device independently using its own state, performs Pedestrian Dead Reckoning (PDR), smooths RSSI values using Kalman filtering, estimates position using log-distance ranging and weighted multi-literation (Weight Centroid Localization), and applies soft RSSI anchoring to suppress inertial drift. A WebSocket pipeline then streams live position updates to the parent application.
Project ID = F25SDP 34 CE M
Supervisor: Prof. Qutaibah Malluhi
Abdulaziz Alkathiri, Mohamad Elhaddad, Moaz Jemmieh, Zaid Obaid
This project presents the design and implementation of a Smart Roadway System for Traffic Monitoring and Adaptive Street Lighting. The system mainly addresses two major challenges present in Qatar’s modern urban environments: the need for effective speed enforcement approach to enhance road safety, and the reduction of energy waste caused by continuous streetlighting at night. The proposed solution utilizes a microprocessor with a camera positioned carefully at multiple camera checkpoints to monitor road activity and capture license plate images. Each camera node acts as an edge device that performs real-time vehicle detection using a lightweight computer vision based vehicle detection module, while Optical Character Recognition (OCR) for license plate extraction is handled at the central server using Plate Recognizer’s external API. The server correlates detection from consecutive camera checkpoints using timestamps to obtain the vehicle speed, storing the data in a scalable database, which is optimized for efficiency and accuracy. Vehicles exceeding the allowed speed segment for the defined segment are logged with the captured images as evidence of their violation, while images corresponding to non-violating records are automatically deleted to conserve storage. The system also identifies vehicles that fail to reach subsequent checkpoints in the same segment, generating alerts for potential incidents such as car accidents or abrupt stops. Simultaneously, an adaptive lighting module dynamically adjusts the segment’s streetlight brightness, maintaining dim luminosity during nighttime and increasing to full brightness upon vehicle detection which is triggered through the same camera-based vehicle detection system described above. This integrated approach demonstrates a reliable, low-cost and sustainable solution for smart road infrastructure, aligning with the objectives of Qatar National Vision 2030 in promoting sustainable and intelligent infrastructure solutions. The system design was further refined through consultation with a professional engineer working at Ashgal in developing intelligent transportation systems.
Project ID = F25SDP 35 CS M
Supervisor: Prof. Khaled Shaban
Sidi Chaikh, Wardan Fakhouri, Amar Massoud, Mohanad Massoud
Conversational Artificial Intelligence (AI) is rapidly becoming a key part of the modern digital experience, from e-commerce to customer service. However, despite the power of today's Large Language Models (LLMs), their practical application is often superficial. Most businesses and users still contend with simple chatbots that are disconnected from live, proprietary data, creating a critical "integration gap" that prevents AI from performing meaningful, context-aware tasks. This project addresses this gap by designing and implementing Dabber, a hybrid AI-interaction platform for automated provisioning and integration of AI assistants into websites. The system's core innovation is the engineering of a standardized middleware, the Model Context Protocol (MCP), which functions as a robust, domain-agnostic bridge . This architecture enables the platform's AI to interact with both static, indexed knowledge (via Retrieval-Augmented Generation) and live, dynamic data (via client APIs), transforming the LLM from a general conversationalist into a task-oriented digital assistant . The project's key achievement is a scalable, multi-tenant platform featuring a centralized orchestration dashboard. Through this dashboard, clients can register, upload their OpenAPI specifications, and trigger the automated, containerized deployment of an isolated, per-client stack. This unique stack, which includes a dedicated MCP Client, MCP Server, and a local vector database, ensures complete data privacy and performance isolation for each client . This work successfully demonstrates a viable and scalable solution to the AI integration challenge. By providing a platform that is both modular and easy to integrate, Dabber democratizes access to advanced AI, offering a practical alternative to the complex and "closed-ecosystem" solutions that currently dominate the market.
Project ID = F25SDP 36 CS M
Supervisor: Dr. Noora Fetais
Essa Al-Mannai, Ahmed Alamoudi, Yousef Saeed, Faisal Taleb
This project presents the development of a system designed to facilitate organizational policy assessment and support the creation of new or improved policies aligned with established frameworks such as ISO 27001, CREST, NIA, NCSA, as well as the QDB guidelines. The increasing complexity of compliance standards often forces organizations to rely on external consultants, creating a disconnect between employees and their internal policies while increasing the risk of audit failures, especially for small and medium-sized enterprises (SMEs). The proposed system leverages Artificial Intelligence (AI), specifically Natural Language Processing (NLP) techniques, to analyse organizational policies and compare them against multiple frameworks simultaneously. Using semantic analysis, the system identifies compliance gaps, generates context- aware recommendations, and assists in drafting missing policies. An integrated chatbot further enhances accessibility by providing interactive guidance to users, while also generating dependable, implementation-ready policies. The system outputs are presented through a visual dashboard that highlights compliance strengths and weaknesses using clear, non-technical language, making it suitable for users without specialized expertise while maintaining quality for users with a higher level of competence. The system was developed using an Agile methodology, enabling iterative improvements in policy analysis accuracy, framework mapping, and user interface design and maintaining modularity. Evaluation of the prototype demonstrates a measurable reduction in the time and effort required for policy assessment, while maintaining consistent alignment with multiple standards. However, the AI- driven approach introduces some variability in output consistency and remains dependent on the quality and structure of input policies. Overall, the system provides an efficient and practical baseline for organizations to monitor, evaluate, and maintain compliance with evolving regulatory requirements.
Project ID = F25SDP 37 CS M
Supervisor: Dr. Moutaz Saleh
Mohammed Al-Romaihi, Ahmed Al-Surmi, Omar Amdadullah, Mostafa Youssef
Modern educational institutions encounter multiple problems because students must navigate through disorganized academic resources. Students who depend on unverified study materials and expensive private instruction develop irregular learning patterns, which decrease their productivity and create unfair academic opportunities. Academic platforms currently fail to deliver integrated features, certified content, and accessible tutoring services to their users. The current academic resource management system requires a single platform that unites all educational materials while ensuring students can access them easily and maintain consistent learning performance. The implementation of this system will lead to better academic results and student financial relief while creating a learning environment that promotes fairness for all students. This project describes the complete development process of YAD (Your Academic Desk), which functions as an academic support platform to enhance student learning material access and course preparation. The platform unites fundamental academic capabilities into one environment, allowing students to obtain authorized course materials, schedule tutoring sessions, participate in organized discussions, and receive customized academic assistance. The platform functions as a unified system that connects students to their instructors and academic departments through enhanced communication while providing streamlined access to vital educational resources. The system includes resource management capabilities, booking systems for online and offline classes, discussion rooms, and performance-based learning recommendation tools. The solution operates through a multi-layered structure that combines academic workflow-specific database design with software design best practices. The system implements optimal design patterns to enhance system reliability and improve both modularity and maintainability. The project includes a cost evaluation to determine platform viability and a complete workflow description that shows how tasks will be distributed and how the project will progress. This document presents the architectural choices made during development, along with descriptions of functional components and development approaches, and outlines the main obstacles that arose during project execution. YAD demonstrates multiple benefits to users through its ability to enhance academic resource accessibility, decrease private tutoring needs, improve resource organization, and strengthen student-instructor collaboration. It addresses academic support system deficiencies to establish an efficient learning environment that supports equal opportunities for students.
Project ID = F25SDP 38 CS M
Supervisor: Prof. Rehab Duwairi
Ayoub Abdoun, Hamed Abumatar, Mahmoud Al-Qudeimat, Yahya Taha
This project addresses a clear gap faced by students transitioning from block-based programming to text-based coding: the jump is abrupt, discouraging, and difficult without structured guidance. Learners who are comfortable with visual programming often struggle once syntax, debugging, and unfamiliar development environments are introduced. This slows their progress, impacts confidence, and limits early exposure to real programming skills. Creating a smoother, more supportive transition pathway is essential for improving learning outcomes and lowering the entry barrier into computer science for younger students. Our proposed solution, Brick2Script, is an educational platform that helps learners understand how the logic they build with blocks maps to real code, while gradually shifting their reliance from visual construction to written programming. Instead of simply “converting blocks to code,” the system focuses on revealing the underlying computational concepts, synchronizing block actions with code representations, and guiding the learner through increasingly text-centered challenges. The platform integrates a block-based workspace, an in-browser code editor, AI-assisted hints on demand, gamified progression, and a scaffolded challenge system aligned with introductory programming curricula. The team began by thoroughly examining the problem through research, competitor benchmarking, and evaluation of existing learning platforms. We defined the platform’s architecture, core features, and user flows, and outlined the major subsystems: authentication and user management, curriculum and challenge engine, block–code dual-view logic, execution sandbox, and the evaluation and feedback module. These foundations guided the full development of the platform throughout the project lifecycle. The platform was built on a robust technology stack using a modular monolith architecture suitable for secure, scalable development. All major subsystems were fully implemented and integrated: the block–code dual-view workspace, the in-browser code execution sandbox, the AI-assisted hint system, and the gamified challenge progression engine. The system was validated through functional and non-functional testing, confirming that it meets the defined requirements. In summary, Brick2Script was successfully designed, implemented, and evaluated as a complete educational platform. The project addresses a genuine gap in programming education and delivers a functional, tested solution that helps learners transition from block-based to text-based coding in a structured and engaging way. The team concludes that the platform is impactful, technically sound, and ready to support real educational use.
Project ID = F25SDP 39 CS M
Supervisor: Prof. Rehab Duwairi
Ahmed Abdumahfouth, Elbaraa Delmi, Mohamed Elansary, Abdulla Jamali
This project focuses on fixing common problems in traditional dine-in restaurants. Today, customers often wait for a waiter just to place an order, face delays or mistakes when they have special requests, and sometimes have to stand in line to order or pay. All of this wastes time, creates room for miscommunication, and lowers customer satisfaction. At the same time, especially after the COVID- 19 pandemic, more people now prefer technology-assisted and contactless dining. Together, these factors highlight the need for a more streamlined, digital way to handle table service. To address this, we propose and develop a QR-based ordering and payment system that digitizes the entire in-restaurant ordering process. Each table has its own unique QR code that customers can scan to open a digital menu, customize their items, and send orders directly from their smartphones. The system then sends these orders in real time to the kitchen interface, updates table status and restaurant capacity, and supports secure online payment. It offers several user-facing interfaces: a customer view for browsing the menu and ordering, a staff view for monitoring and managing active orders, and an admin/manager view for managing menus, tables, and analytics. In addition, the system lays the foundation for future AI-driven, personalized menu recommendations based on customer behavior and order history. The completed system achieves several important results. Functionally, it supports end-to-end order handling, including QR scanning, menu browsing, item customization, shared basket management for group dining, budget filtering, bill splitting, and integrated payment processing. Operationally, it keeps customer devices and restaurant dashboards synchronized in real time, allowing staff to track orders and table occupancy efficiently. From a user-experience perspective, preliminary tests in a simulated restaurant environment show shorter ordering times, fewer order errors caused by miscommunication, and positive feedback from both customers and staff about how easy the system is to use. In conclusion, the project demonstrates that a QR-based, multi-interface restaurant platform can significantly improve ordering efficiency, transparency, and customer satisfaction, while also giving restaurant owners better visibility into their operations. It also provides a solid technical base for future features such as full POS integration, more advanced analytics, and AI-based personalization.
Project ID = F25SDP 40 CS M
Supervisor: Prof. Rehab Duwairi
Omar Aboelrous, Ahmad Almashhadani, Josh Erun Bryce Calma, Hateim Elagha
Ransomware has been a leading threat in the cyber threat landscape for years, affecting almost all industries across both the public and private sectors. Organizations of all types remain vulnerable to ransomware due to the high prospect of monetary gain for attackers and the widespread dependence on digital infrastructure. This is reflected in recent statistics: 65% of organizations suffered attacks in 2024, with annual global losses exceeding $265 billion. While enterprises are equipped with security teams and tools, the traditionally used signature-based detection systems are not designed to deal with highly sophisticated and novel attacks. On the other hand, manual analysis and triage are too slow to prevent or mitigate the catastrophic impacts of ransomware attacks, where every second of downtime translates to significant financial losses, operational disruption, and reputational damage. This project introduces ResiWare: an AI-powered autonomous ransomware defense solution for enterprise environments. ResiWare solves the problem by integrating traditional rule-based security monitoring with modern AI-powered reasoning to create a hybrid detection architecture capable of identifying both known threats and zero-day ransomware attacks. The system implements a four- layer architecture, where it collects logs and telemetry from different sources (e.g.: servers, endpoints, workstations, etc.), feeds the logs to the detection layer for analysis to identify threats, then the reasoning layer takes the data and the detection decisions to correlate it with threat intelligence and produce well-informed insights about the threat; the response layer can then develop a response to the detected incident based on the analysis and correlation of the reasoning layer, and the system ends with the learning layer that takes feedback from humans in the loop (i.e.: security analysts) to improve the system’s accuracy. A key innovation in the system is what we call the event correlation engine, which analyses patterns across multiple security events to detect multi-stage ransomware campaigns that would otherwise appear to be benign when examined in isolation. At the time of writing this abstract, ResiWare successfully analyzed 900+ security events and detected 86 ransomware attacks with a confidence score ranging from 0.7 to 0.9. Thorough testing with different ransomware indicators, including file integrity violation, suspicious process execution, privilege escalation attempts, backup deletions, and mass file encryption, was used to validate the system’s robustness across multiple threat vectors. The system was tested by using pre-prepared scripts that simulate ransomware behavior. We used malicious scripts, benign scripts, and suspicious scripts to see how ResiWare reacted to each of them in terms of detecting the malicious indicators and correlating them (true positives) while not flagging the benign indicators (false positives). However, no specific enterprise production environments were analyzed yet; this is something we choose to address in the second semester. ResiWare represents a huge advancement in automated ransomware detection for enterprise environments. The system uses a pre-trained large language model, Meta’s Llama3.1 (8B) in particular, rather than requiring extensive labeled datasets to detect sophisticated attacks without the burden of training custom machine learning models. The modular architecture ensures ease of future improvements and addition of capabilities planned for the second semester. The system's success validates the approach of using AI as an intelligent assistant that enhances human decision- making by filtering noise, highlighting genuine threats, and providing actionable intelligence at machine speed.
Project ID = F25SDP 41 CS M
Supervisor: Dr. Mohammad Saleh
Mahmod Abdelmawgood, Ibrahim Al-Ajmi, Mohd Hoque
The growing demand for scalable, personalized academic support has exposed significant limitations in traditional education. Fixed question banks and static course materials cannot adapt to individual student needs, leaving students without timely, targeted feedback. Instructors simultaneously face high workloads—creating diverse assessments, grading open-ended responses, and tracking class- wide performance—while identifying learning gaps in real time. These challenges are amplified in online and hybrid environments where direct instructor–student interaction is inherently limited. Eureka was designed and implemented as a web-based AI-assisted tutoring platform to address these challenges. Built with Next.js for the frontend, Supabase for cloud-hosted data storage and authentication, and the GLM-4.5 Flash large language model (LLM) for intelligent academic functions, the platform serves three distinct user roles—students, instructors, and administrators. Instructors generate questions automatically via the LLM or enter them manually, specifying question type (theory, multiple-choice, or true/false), difficulty level, learning objective, and class code. The class- code mechanism ensures each student’s practice queue contains only questions relevant to their enrolled class. Open-ended theory responses submitted by students are evaluated by the AI, providing formative feedback while substantially reducing the manual grading burden on instructors. Students access a dedicated dashboard to browse the question bank, answer questions, and request on-demand explanations from the AI tutor for specific questions or any course topic. A performance analytics view reports total attempts, overall accuracy, accuracy broken down by question type, and per-learning-objective mastery, helping students identify knowledge gaps and self-regulate their learning. An administrative panel provides full CRUD management of users, courses, classes, majors, semesters, and colleges, with CSV bulk-import support to streamline institutional onboarding. The project delivered a fully functional prototype validated through systematic unit testing across all major use cases, with every test case passing as expected. Secure role-based authentication and routing were achieved using Supabase Auth and Next.js middleware. Reliable relational data models were implemented for questions, student answers, courses, and classes. LLM integration proved effective across all three target functions—question generation, open-ended answer evaluation, and contextual student tutoring—while structured prompt engineering and defensive response-parsing logic addressed the output consistency challenges inherent in language model APIs. Eureka’s defining contribution is the unified deployment of a single LLM API across three distinct academic workflows within a coherent, multi-role platform that includes an institutional-grade administrative backend. Unlike most academic AI tutoring prototypes, the system addresses practical deployment requirements: semester-aware class management, bulk data imports, password recovery, and a privacy policy. The platform demonstrates that modern cloud-native web technologies and publicly available LLM APIs can be combined to deliver an affordable, scalable AI tutoring solution. It also surfaces critical lessons in prompt reliability, AI evaluation consistency, and data privacy—considerations that must guide future deployment of AI-assisted tutoring in real educational environments.
Project ID = F25SDP 42 CS M
Supervisor: Dr. Osama Halabi
Mohamed Hadi, Hisham Bassam Ahmed Hammad, Abdallah Omar
The Cybersecurity Fundamentals course provides students with a strong theoretical foundation but lacks a practical laboratory component where learners can apply concepts in a hands-on manner. This gap reduces opportunities for students to directly experience how cyberattacks unfold, how defenses are implemented, and how theoretical principles translate into practice. Addressing this issue is significant because modern cybersecurity education must balance conceptual understanding with applied skills to prepare students for real-world challenges. This project, called V-LabSec (Virtual Reality Laboratory for Cybersecurity Fundamentals), introduces a Virtual Reality (VR) based Lab Course designed to extend the Cybersecurity Fundamentals syllabus with immersive, puzzle-based training. Inspired by prior work such as CySecEscape 2.0, the system advances beyond awareness level training by aligning every lab module with topics from the course, including the CIA triad, malware, authentication, access control, cryptography, penetration testing, and privacy. Each VR “room” presents students with an interactive challenge for example, decrypting a key to unlock a resource, identifying malicious processes, or configuring secure access policies. By solving these puzzles, students engage with authentic scenarios that promote deeper understanding, problem-solving skills, and retention of knowledge. The main contributions of this project are threefold. First, it develops a structured framework for mapping lecture content into VR-based labs. Second, it demonstrates the feasibility of combining immersive VR interaction with automated feedback and performance logging. Third, it enhances the value of the existing course by providing a reusable and scalable laboratory that benefits both students and instructors. What makes V-LabSec unique is its direct alignment with the Cybersecurity Fundamentals syllabus, combined with integrated telemetry for tracking student performance and automated reporting that provides instructors with actionable insights. Unlike general-purpose security VR games, V-LabSec is purpose-built as a structured academic lab environment, ensuring that every puzzle directly reinforces course learning outcomes while offering measurable progress indicators. In summary, this project delivers an innovative and practical VR laboratory that transforms a theory- only course into an experiential learning environment, equipping students with applied cybersecurity skills needed in academic and professional contexts.
Project ID = F25SDP 43 CS M
Supervisor: Dr. Noora Fetais
Ali Abouelkhir, Mohammed Alloh, Amr Mohamed, Yazan Mohammad Alsaleh
Traditional marathon and race timing systems rely heavily on hardware-based solutions such as Radio Frequency Identification (RFID) tags and timing mats distributed along the track. In some professional events, photo-finish cameras are also used to determine precise finishing times. While these systems provide accurate results, they come up with several drawbacks. RFID systems require specialized equipment, logistics for distributing and collecting tags, and often depend on external suppliers. Similarly, photo-finish camera systems are expensive and require trained personnel for setup and maintenance. These limitations make such solutions less accessible, especially for smaller or locally organized events. To address these challenges, the project at first introduces a cost-effective and scalable timing system based on computer vision. The system leverages modern technologies such as OpenCV for real-time video processing and QR code detection, eliminating the need for expensive hardware components. In earlier versions of the project, the system relied on a dual-camera setup. One camera was positioned at the finish line to capture timing events, while another camera was placed in front of the runners to detect and read their bib numbers using optical character recognition (OCR). Although this approach was functional, it introduced several practical limitations. In particular, when multiple runners crossed the finish line close to each other—especially when one runner was directly behind another—the front-facing camera struggled to correctly identify each individual. This occlusion problem significantly affected the reliability of the system in real-world race conditions. To overcome these limitations, the system’s first redesigned adopted a single-camera approach. In the proposed solution, a high-resolution, high-frame-rate camera is mounted directly above the finish line at a height of approximately 1–2 meters, providing a bird’s-eye view of the crossing area. This camera simultaneously handles both detection and identification tasks, simplifying the system architecture and improving robustness. In addition, the identification method was initially changed from traditional bib numbers to QR codes. Each runner is assigned a unique QR code, which is placed on their shoulder to ensure visibility from the top-mounted camera. While this approach improved detection compared to text-based bib numbers, it still presented some challenges in real-world conditions. QR codes can become difficult to detect under motion blur, partial occlusion, or non-optimal viewing angles, which are pretty common during high-speed race scenarios. To further improve reliability and robustness, the system was enhanced by replacing QR codes with AprilTags. AprilTags are specific design for computer vision project and applications, as well as they provide more stable detection under varying conditions such as lighting and motion or when only some of the AprilTag is visible. Unlike QR codes AprilTags can be detected with high accuracy and faster compared to QR codes. In addition, AprilTags algorithms are optimized for real-time performance. During operation, the system continuously monitors a predefined virtual finish-line zone. When a runner enters this zone, the system detects the AprilTag and records the corresponding timestamp. In the intended final deployment, each runner is recorded only once when crossing the finish line for the first time, and any repeated detections within a short time window (e.g., five minutes) are ignored to prevent duplicate entries. For testing and development purposes, the current implementation allows two detections per runner to simulate start and finish events. The recorded data, including runner identification and timestamps, is stored in a structured CSV file for further analysis. The system’s performance can be evaluated based on detection accuracy, timing precision, and robustness under different race conditions, such as varying runner speeds and densities. Overall, this approach significantly reduces cost and complexity compared to traditional marathon timing systems while maintaining reliable performance. By replacing specialized hardware with a camera-based intelligent system, the solution provides a practical and scalable alternative suitable for a wide range of sporting events.
Project ID = F25SDP 44 CS M
Supervisor: Prof. Khaled Khan
Mohammed Al Shebani, Islam Al-Absi, Nasser Al-Kuwari, Mohammed Talib
The increasing prevalence of credential fraud and the inefficiency of traditional verification methods pose significant challenges to educational institutions, governments agencies, employers, and individuals worldwide. Relying on physical documents or centralized digital systems often results in slow verification, a lack of transparency, and susceptibility to forgery. These issues are particularly critical in contexts that demand trust and authenticity whether in academic qualifications, professional certifications, national identity, or licensing credendials. To address this, our project proposes a blockchain-based credential system that enables secure, transparent, and tamper-proof digital documentation. By leveraging the immutability and decentralized nature of blockchain, our solution enables authorized issuers (including educational institutions, government bodies, and licensing authorities) to issue verifiable credentials whose authenticity can be independently verified without relying on a centralized authority. The system is designed to support a broad spectrum of document types, including academic certificates, Qatar National Identity Cards (QID), driving licenses, and any additional credential categories that may be introduced in the future. Our implementation includes a web-based platform for issuers to create and publish credentials, a secure verification mechanism, and a recipient-facing interface for credential management and sharing. The architecture follows a document-agnostic model, meaning new credential types can be onboarded with minimal configuration by defining their required fields, issuing authority, and validation rules ensuring long-term extensibility without structural redesign. The system supports open standards to ensure compatibility with future SSI (Self-Sovereign Identity) frameworks and integrates a lightweight, cost-efficient blockchain network to reduce operational overhead. Key accomplishments include a fully functioning end-to-end prototype, successful demonstration of real-time credential issuance and verification across multiple document types, and integration of cryptographic signature and hash validation using blockchain anchors. Testing showed that our system can issue credentials in batches while maintaining low cost and high scalability. These results highlight the feasibility and effectiveness of combining blockchain technology with a flexible credential framework to modernize digital documentation, providing a future-ready solution that enhances trust, privacy, and efficiency across education, employment, and public administration sectors.
Project ID = F25SDP 45 CS M
Supervisor: Dr. Abdelkarim Erradi
Ahmed Adam, Abdalrehman Daud, Yousif Elsherif, Abdul Mahmoud
Academic advising is critical to student success, yet traditional appointment-centric advising often fail to meet the expectations of today’s digitally connected learners. Students often struggle with complex degree requirements, unclear university policies, and inconsistent guidance. Advisors face heavy workloads particularly during registration periods and lack quick access to a unified dashboard showing student profiles, issue summaries, notes from prior interactions and risk indicators. These are essential for informed and personalized support and improve decision-making. This gap leads to poor advising quality, uninformed decisions, missed deadlines, and ultimately lower retention, delayed graduation, and higher dropout rates. Universities need intelligent, scalable solutions to modernize advising and improve retention. To bridge this gap, we present SUHAIL, an AI-powered academic advising assistant to transform student-advisor interactions. SUHAIL leverages artificial intelligence and natural language understanding to deliver timely, accurate, personalized guidance through a conversational interface. Students can access real-time answers about courses, programs, and policies, supported by citations. Key features include personalized academic planning, meeting summaries from chat or voice sessions with the academic advisers, deadline reminders, and insights from course reviews to guide study planning. These features empower students with accurate information while reducing advisor workload. SUHAIL architecture ensures privacy-by-design with role-based access control, secure handling of academic records, audit logs, confidence indicators, and citation transparency. Its modular design enables easy adaptation to any institution by updating datasets and policies, ensuring scalability and customization. Early prototype demonstrates SUHAIL’s ability to reduce repetitive inquiries and support informed decision-making. Students gain immediate and reliable advising, while advisors reclaim for mentorship and academic counseling. Expected outcomes from SUHAIL include a 30-50% reduction in advisor workload by automating routine tasks and inquiries, average response times under 5 seconds for AI-driven queries to enhance real-time support, and response accuracy exceeding 95% through RAG implementation, citations, and continuous feedback loops to minimize errors and build trust.
Project ID = F25SDP 46 CS M
Supervisor: Prof. Saeed Salem
Ahmad Abuarrah, Mohammed Bakhit, Ghanim Masoud, Yahia Zagad
Spots Mobile is a Qatar-focused local discovery application designed to help residents, students, visitors, and organizations find nearby places, community updates, and promoted events through one location-aware mobile platform. The project addresses the problem of fragmented local information, where users often depend on separate sources such as social media, private messaging groups, static map reviews, and informal recommendations. This fragmentation makes it difficult to find timely, trustworthy, and relevant information about what is happening nearby. Therefore, the project aims to provide a more organized, community-driven, and user-friendly way to discover local activities and updates. The project was developed in two main stages. In the first stage, the team designed the initial requirements, system architecture, and proof-of-concept to validate the core idea. This early version focused on essential features such as user authentication, local post creation, map-based exploration, and an initial AI summarization concept. In the second stage, the system was expanded into a fuller MVP with a more complete implementation and stronger backend support. The final solution uses React Native, Firebase Authentication, Cloud Firestore, Firebase Cloud Functions, Firestore security rules, maps/geolocation services, and OpenAI-backed area summaries. The final system implements authentication, Home discovery, Explore map and heatmap behavior, local post creation, saved spots, comments, reactions, reports, moderation, profile settings, English/Arabic localization, in-app notifications, XP and leaderboard features, admin analytics, subscription state management, and organization-only promoted events. AI summaries are generated through backend Cloud Functions rather than direct client-side OpenAI calls, which protects API credentials and keeps summarization logic under server control. The project achieved a modular layered architecture consisting of a presentation layer, application/service layer, repository/data access layer, backend layer, and external service integrations. This structure improves maintainability, supports future expansion, and separates user interface logic from backend and database operations. The system was evaluated through functional testing, non-functional testing, TypeScript type checking, ESLint validation, manual workflow verification, and backend trust tests. The backend trust test suite passed 10 out of 10 test cases using Firebase emulators, confirming that key security-sensitive operations were properly protected. Overall, Spots Mobile demonstrates a feasible and extensible community discovery platform for the Qatari context. Its main contribution is the integration of local posts, map-based discovery, AI area summaries, organization promotion, moderation, notifications, XP, and leaderboard features into one unified system. The project also identifies future improvements such as full media upload, production push notifications, real payment integration, broader usability testing, stronger privacy workflows, and production deployment preparation.
Project ID = F25SDP 47 CS M
Supervisor: Prof. Cagatay Catal
Saoud ALMehsen, Abdulla Al-Hajri, Hamza Aljaji, Abubaker Elfagih
Many universities receive a large number of applications whenever they announce new faculty positions across their colleges and departments. Recruitment committees must manually read and compare dozens of CVs, research profiles, and teaching records for each position. This process is slow, repetitive, and can become inconsistent when different members interpret the same information differently. As the volume of applications grows, it becomes harder to ensure fairness, transparency, and timely hiring decisions for all academic units. Hire AI was developed to address this challenge by providing an AI-assisted recruitment tool for faculty hiring across the university. The system allows committees to upload job descriptions and candidate CVs, then automatically parses and organizes relevant information such as academic degrees, publications, teaching experience, service activities, and other scholarly achievements. Using an agentic AI workflow based on large language models, Hire AI helps map candidate profiles to job-specific criteria and generates an initial ranking according to configurable weights for teaching, research, and service. The platform offers several useful features for committees. It produces structured summaries of each candidate, highlights strengths and potential concerns, and generates an evidence-based score breakdown rather than a single opaque score. Committees can adjust the weighting of different criteria to reflect the needs of each department and can export a shortlist for further review and interviews. In our prototype evaluation with sample faculty calls from different disciplines, Hire AI was able to significantly reduce the time spent on initial screening while maintaining consistent criteria across candidates. Overall, Hire AI does not replace academic judgment but supports it. The project demonstrates how AI-driven analysis, when properly constrained and explained, can help universities handle faculty recruitment more efficiently, make shortlisting more transparent, and better support committees in selecting qualified candidates.
Project ID = F25SDP 48 CS M
Supervisor: Dr. Moutaz Saleh
Yahya Abdulselam, Aiman Alhetari, Abdelrahman Kotb, Ahmed Mohamed
Dysarthria is a motor speech disorder caused by neurological conditions such as stroke, cerebral palsy, or Parkinson's disease that severely limits communication, reduces independence, and lowers quality of life. Traditional therapy requires intensive in-person sessions with speech-language pathologists, which are costly, time-consuming, and often inaccessible due to limited specialist availability. This project introduces an AI-powered Speech Therapy System for Dysarthria that integrates three core components: Severity Detection classification of speech impairment levels using fine-tuned pre-trained encoders such as WavLM and HuBERT, Error Spotting analysis of patient recordings to identify phoneme-level mispronunciations and speech errors, and Dynamic Therapy Planning generation of personalized therapy plans and daily practice exercises using large language models (LLMs), tailored to severity, age, and language background. The platform is delivered through a web application, enabling patients to record speech, receive AI-driven feedback, and follow adaptive therapy exercises. This design reduces reliance on constant therapist supervision while still allowing professional oversight for validation. By combining machine learning-based assessment with personalized therapy planning, the project provides a scalable and accessible approach to speech rehabilitation. It establishes a foundation for future clinical evaluation and deployment in healthcare and educational contexts in Qatar, the GCC, and beyond.
Project ID = F25SDP 49 CE-CS M
Supervisor: Dr. Osama Halabi
Mohd Bashar, Fahrel Hidayat, Syed Mohammed Hamza, Marcus Monteiro
Search and rescue operations often require robots to navigate dark, cluttered, and hazardous environments where human responders face significant risk. Effective teleoperation in these settings depends on reliable situational awareness, low-latency streaming, and clear identification of victims or hazards. However, conventional visual-only interfaces are limited by occlusion, low-light conditions, and cognitive overload, which reduce operator performance and increase the likelihood of missed detections. Recent advances in 360° video, diminished reality, haptics, and multimodal sensing offer promising pathways for improving remote situational awareness and guiding operators toward task-relevant cues. This study aims to design and evaluate an integrated teleoperation framework that combines 360° VR streaming, diminished reality for visual clutter reduction, multimodal sensing such as thermal, stereo audio, and directional haptic belt for spatial guidance. The objective is to enhance operator performance in simulated SAR scenarios by improving victim detection, reducing false positives and negatives, and providing better awareness of the environment. The scope of the project includes developing a stable 360° streaming environment, implementing a DR pipeline to highlight detected victims, integrating thermal and audio-based cues, and testing endurance, usability, workload, and cybersickness within VR-based robot control. To achieve these goals, we constructed a full teleoperation prototype composed of a mobile robot platform equipped with RGB and thermal cameras, onboard audio sensing, and a VR interface that streams panoramic video to the operator. The system incorporates DR-based background suppression and a haptic belt that conveys the direction of detected cues. We developed evaluation tasks in mock SAR environments featuring clutter, low-light conditions, and designed user trials to measure detection performance, cognitive workload, comfort, usability, and system reliability. The experimental protocol compares visual-only (baseline) teleoperation against, DR assistance, haptic assistance, and the combination of DR and haptic-assisted configuration to assess its impact on time- to-victim, detection rate, and teleoperator experience.
More in Past Senior Projects
Stress is a significant public health concern in Qatar and the Gulf region. Autism Spectrum Disorder (ASD) affects 1.14% [1] of school-aged children, which is estimated to be 1 in 87, which is higher than the global...
Getting proper guidance at university can sometimes be frustrating as the traditional advising systems are not effective or ineffective. Many students face the challenge of not being able to get answers to question...
The desert climate in Qatar makes agriculture difficult due to water scarcity and extreme temperatures that affect the health of crops. The Smart Candy Farm is an innovative solution that addresses the biggest...