Research
Expertise

The emergence of Large Language Models and autonomous multi-agent systems has opened a new chapter in my research. I design hierarchical agentic architectures where swarms of specialised AI agents collaborate, negotiate, and execute complex tasks without human intervention — from securing OT networks to managing oncology patient data and automating enterprise GRC processes.

A core thread is Retrieval-Augmented Generation (RAG) combined with multi-modal models — enabling agents to reason over structured clinical, security, and compliance data in real time. My 2026 work on Agentic SOC introduces a swarm-orchestrated multi-agent LLM for fully autonomous OT cyber defence, while ONCO-LLM targets cancer patient QoL monitoring through agentic LLMs at home.

Security is a cross-cutting theme woven through nearly every domain I work in. My dedicated security research addresses the unique vulnerabilities introduced by 5G/6G networks, IIoT deployments, and AI-driven systems. I proposed a Software-Defined Security Architecture for 6G networks, designed an IT/OT Cyber Range framework for consumer services, and introduced AI-enabled Cyber Threat Intelligence (CTI) resource allocation models for IoE-Edge environments.

In digital forensics, I've tackled adversarial AI threats to medical devices, OSINT automation from social media, and frameworks for drone-based investigations — a uniquely emerging law-enforcement challenge. My 2026 paper on quantum-resistant cybersecurity provides a forward-looking roadmap for post-quantum secure systems.

My healthcare AI research spans the full clinical pipeline — from rapid point-of-care diagnosis to long-term patient monitoring. During the COVID-19 pandemic, I developed a multi-modal deep learning framework integrating chest X-rays and clinical data for rapid diagnosis (published in ACM TOMCCAP), and a secure IoMT edge framework to protect and process patient data at scale (IEEE IoT Journal).

In oncology, I address the critical data gap for cancer patients outside hospital settings through ONCO-LLM and the funded 20M SAR SAARIA project — using AI and digital eye fundus images to detect multiple diseases. My work on Explainable AI for medical IoT ensures clinicians can trust and interpret AI-driven decisions, while adversarial defences protect against manipulation of medical deep learning models.

This is the most prolific domain in my portfolio — 34 publications developed over eight years of intensive research into technology-assisted physical and cognitive rehabilitation. The core innovation is a non-invasive, multi-sensory therapy environment using low-cost depth cameras (Microsoft Kinect), inverse kinematics feedback, and gamification to make therapy engaging and measurable for children with hemiplegia, elderly patients, and individuals with physical disabilities.

I extended this work to cyber-physical therapy systems, cloud-based remote monitoring, and cognitive rehabilitation — culminating in m-Therapy (IEEE IoT Journal) which integrates Social IoT, big data, and mobile computing for in-home therapy management. A dedicated thread also covers dyslexia detection in children using multimedia retrieval and, more recently, multi-modal agentic AI with real-time gaze pattern recognition.

Hajj is one of the world's most complex logistical challenges — over 2 million pilgrims converging on Makkah within days. My research addresses this through a unique combination of spatio-temporal big data analytics, crowdsourcing, IoT sensor networks, and blockchain. The landmark 7.5M SAR KSA RDO project (with Oxford and King's College London) delivered a real-time early warning system for crowd catastrophe prevention.

I won the KAUST Challenge 2020 (1M SAR) with an AI crowd flow prediction system, and built HajjMemo — a mobile crowdsourcing platform for pilgrim feedback. Further work includes the digital extension of Makkah's road network via OpenStreetMap, secure k-anonymous Hajj social networks, and mobile edge computing frameworks for secure Hajj services. These systems also feed into smart city frameworks applicable globally.

The IoT/Edge domain serves as the connectivity backbone for many of my research areas. My contributions include early 5G testbed deployments (with lessons published in IEEE Access 2020), semantic IoT notification architectures, and a cognitive edge framework for smart city services. The 2020 IoEV-Chain paper (IEEE Network) introduced a 5G-based secure inter-connected mobility framework for electric vehicles using blockchain.

More recently, my focus has shifted to 6G-native AI at the edge — designing AI inference pipelines that run on resource-constrained edge nodes while maintaining security and privacy. The 2025 paper in the Journal of Internet of Things examines how AI is reshaping IoE edge networks, while J60 (ACM TOIT 2026) proposes a CTI-based resource allocation model that optimises edge security dynamically.

Blockchain emerged as a critical enabler across my research domains. I was among the early researchers to apply blockchain specifically to healthcare therapy (J31, IEEE Access 2018), securing therapy session data on a mobile edge computing layer. I also applied spatial blockchain to mass dyslexia screening (J32) — protecting sensitive child data while enabling large-scale screening programs.

The British Council-funded "Blockchain for Hajj" project established an international collaboration with UK universities to secure pilgrim identity management and services. My work on blockchain for insurance settlements (CAPCO Journal) demonstrated applicability in fintech, while IoEV-Chain showed how blockchain can secure electric vehicle mobility data in a 5G environment.

This domain underpins much of my smart city and Hajj research. I built frameworks that bridge body sensor networks and social networks (IEEE TIM 2010/2011), enabling contextual social media feeds enriched with physiological and location data. The funded KACST project on querying and visualising real-time social network data led to a suite of 10+ papers at top ACM and IEEE venues.

The crowdsourcing thread introduces spatio-temporal constraint-based social networks for group mobility, multimedia-enhanced crowdsourced routing, and the i-Diary/ST-Diary authoring tools for location-tagged multimedia events. The user profiling work (MTAP 2018) applied the Standing Ovation Model to predict social media influence at scale, while shortest-path algorithm surveys (ACM TSAS 2019) are widely cited in navigation and GIS research.

My research career began in immersive computing. The LORNAV system (IEEE DS-RT 2004) was an early virtual reality tool for navigating and authoring distributed learning object repositories in 3D — a concept well ahead of its time. Subsequent work on haptic interfaces explored adding tactile feedback to YouTube videos (ACM Multimedia 2010) and designing collaborative ambient intelligent virtual environments.

This foundational work directly informed the XR thread in my rehabilitation research (serious games, gesture interfaces) and culminated in the $15M USD EON Reality International Winner Grant in partnership with UPM for XR and immersive learning technology innovation. The thread is now re-emerging as AI-powered XR environments become mainstream in education and healthcare.

The earliest research in my portfolio addresses the fundamental challenge of how devices sense, route, and collaborate in wireless networks. My Q1 paper in IEEE Transactions on Instrumentation and Measurement (2007) applied modified syntactic methods to Bengali handwritten character recognition — an early foray into AI-for-sensing. The SENORA framework (IEEE Sensors Journal 2007, Q1) introduced a P2P service-oriented architecture for multi-robot sensor collaboration.

Ant colony-based reinforcement learning for wireless sensor network routing (IEEE IMTC 2007) demonstrated bio-inspired algorithms in resource-constrained environments — a theme that recurs in my later edge AI work. The autonomous dead-reckoning robot navigation system with intelligent precision calibration established foundational work in mobile robotics that directly informed the cyber-physical therapy systems developed years later.