The Research Institutes represent the core of Khalifa University’s research framework, serving as interdisciplinary hubs that bridge the gap between academic discovery and real-world application. By centralizing expertise and infrastructure, these institutes transcend traditional departmental boundaries to tackle complex global challenges and ensure direct alignment with the UAE’s national vision.
The Digital Future Institute (DFI) is Khalifa University’s applied AI and ICT institute, established to design, engineer, and deploy intelligent digital systems across communications, networked infrastructure, sensing environments, energy, climate, cybersecurity, and secure digital platforms. DFI operates as a project-driven institute that bridges advanced AI research with next-generation ICT architectures, integrating foundation models, cyber-physical systems, and hardware–software platforms into operational environments.
The Institute’s mandate is not limited to advancing algorithms, but to architect intelligent systems that are scalable, resilient, and deployable in real-world contexts. DFI’s strategic objective is to accelerate the maturation of AI- and ICT-enabled technologies from laboratory validation toward operational readiness. This includes the development of domain-specific and multi-modal foundation models, AI-native communications systems, distributed and edge computing architectures, and integrated digital infrastructures capable of supporting real-time intelligence.
Aligned with Khalifa University’s research strategy, DFI emphasizes measurable impact, structured TRL progression, and commercialization pathways. The Institute prioritizes applied research that advances technology readiness, supports industry co-investment, and enables validation in relevant operational environments. Rather than focusing solely on publication-driven outputs, DFI supports technology development, benchmarking, validation pilots, and intellectual property generation to ensure that research outcomes translate into deployable systems, scalable platforms, and long-term economic and industrial impact.
DFI is structured around measurable impact and deployment readiness. Its model integrates:
Senior Director of the KU Digital Future Institute
Research Scientist
Postdoctoral Fellow
Postdoctoral Fellow
Postdoctoral Fellow
Postdoctoral Fellow
Postdoctoral Fellow
Research Associate
Research Engineer
PhD Student
DFI research is structured around the design and deployment of intelligent ICT systems. We build next-generation architectures for communications, sensing, and edge computing, and we ensure AI capabilities are embedded throughout the lifecycle, from data and modeling to orchestration, monitoring, and continuous improvement in real environments.
A central focus of DFI is the development of vertical foundation model tracks. Each track is executed as an end-to-end program: problem definition with stakeholders, dataset strategy and governance, model adaptation and evaluation, safety and robustness testing, and engineering integration through APIs and tools. This approach enables foundation models that are purpose-built for specific sectors and workflows, with clear performance targets and deployment readiness.
DFI research is inherently cross-disciplinary. We bring together expertise in wireless and networking, machine learning, systems engineering, security, and applied domain knowledge. Our methodology emphasises reproducibility, benchmarking, and operational validation. The output is not only publications, but also deployable prototypes, reference implementations, and scalable platforms that can transition into partner systems and products.
The output of DFI research is not only publications, but also deployable prototypes, reference implementations, and scalable platforms, including datasets, benchmarks, tooling, APIs, and integration templates, that can transition into partner systems and products, support long-term adoption and technology transfer, and serve as reusable foundations for new tracks and new deployments across telecom, digital infrastructure, energy, climate, and security.
DFI measures impact through deployment outcomes and measurable performance gains, focusing on how our systems reduce operational complexity while improving the reliability, efficiency, and intelligence of digital infrastructure across networks and connected systems, including the practical outcomes that matter in real environments such as higher service quality, stronger security posture, faster incident handling, and energy-aware operation that helps operators and infrastructure owners run smarter without increasing operational burden.
Our deployment impact is defined by what changes after integration into real workflows, because DFI solutions are designed to improve operational execution rather than only produce offline accuracy metrics, which means we target outcomes such as more stable performance under load, fewer failure points through automated reasoning and orchestration, faster troubleshooting through decision support grounded in operational data, and better end-to-end consistency through repeatable pipelines that connect sensing, analytics, and action inside production systems.
DFI delivers sector-level impact through vertical foundation models and applied platforms that are tailored to domain constraints and operational data, so that models do not behave as generic assistants but as domain-aligned systems that can support faster decision-making, higher automation, and improved customer and system outcomes, while remaining deployable at scale through APIs and integration patterns that allow teams to embed the capability directly into tools, dashboards, and operational procedures rather than adding a separate “AI layer” that sits outside the workflow.
DFI also creates ecosystem impact by strengthening Khalifa University’s role as a national hub for applied ICT and AI innovation, because partner programs, deployable platforms, talent development, and commercialization pathways are treated as first-class outcomes of the Institute, enabling a sustainable innovation pipeline that supports new products, new ventures, and long-term competitiveness in strategic digital domains where sovereign capability, trusted deployment, and real-world performance are critical.
Professor Mérouane Debbah’s distinguished honours and recognitions: Professor Mérouane was awarded the Khalifa University Leadership Impact Award (2025), reflecting exceptional leadership, mentorship, and sustained institutional impact. He was also named a Clarivate Highly Cited Researcher 2025, a distinction reserved for researchers with exceptional global scientific influence. In addition, Khalifa University has highlighted major international recognition of his work, including the IEEE Communications Society Industrial Innovation Award (EMEA, 2024), a strong signal of impact beyond academia into real-world technology and industry.
TelecomGPT-Arabic: award-winning applied AI for telecom operations: DFI team received the Telecom Review Excellence Award for “Best AI Application for Vendors in the Middle East” for TelecomGPT-Arabic, developed with du, Nokia, Microsoft, and the ITU. This award validates DFI’s ability to deliver vertical AI that is not only technically advanced but operationally relevant, designed to support real telecom workflows through domain-aware reasoning and scalable integration.
Khalifa University’s DFI won 1st place worldwide in the ITU Large Wireless Model (LWM) Challenge 2025 (with 65 teams), demonstrating strong capability in model design, experimentation, and pipeline optimisation for AI-native wireless intelligence. These challenge outcomes show repeatable execution; building high-performing systems under strict evaluation protocols, an important indicator of readiness for real deployments and large-scale adoption.
Includes deployable systems and IP
DFI builds and maintains shared, deployment-grade technology platforms that power our vertical tracks and partner pilots. These platforms are engineered to integrate with real operational environments, telecom infrastructure, enterprise IT, edge/cloud systems, and secure deployments, while keeping governance, observability, and performance targets built-in from day one.
Our approach is platform-first: every model is packaged with the tooling needed for production use, including retrieval and grounding pipelines, evaluation harnesses, APIs/SDKs, and lifecycle mechanisms (monitoring, regression testing, and controlled rollout). This ensures that partners can move from prototype to pilot to scaled adoption without rebuilding the foundations each time.
DFI’s core platforms below provide telco-native intelligence, Arabic-first interaction, real-time operations support, RF-aware reasoning, standards grounding, and measurable benchmarking for deployment-ready adoption.
TelecomGPT: TelecomGPT is a telco-first LLM platform built for telecom standards, engineering workflows, and operational reasoning. It targets the gap where general LLMs fail to capture telecom-specific depth and decision logic. It is developed through a structured adaptation pipeline using telecom datasets across key tuning stages, and it is evaluated using telco-specific tasks that reflect real engineering needs (not generic chat). TelecomGPT is designed for practical adoption through tooling and ecosystem assets (knowledge grounding, benchmarks, and accessible interfaces) so it can be trusted in real deployments.
TelecomGPT-Arabic: TelecomGPT-Arabic brings telco-first AI into Arabic-first operational environments. It enables Arabic interaction for telecom workflows without depending on English-only tools. This project is a collaboration between the DFI, du, NOKIA, MICROSOFT and ITU. It is designed to support real operational functions such as issue handling, device troubleshooting, and operational insights, with Arabic fluency that matches local terminology and usage. TelecomGPT-Arabic expands accessibility across the region, reduces friction and configuration mistakes, and supports Arabic-native automation and assistant experiences.
TelecomGPT-RT: TelecomGPT-RT is the first release in the TelecomGPT-Reasoning (TelecomGPT-R) Series, a domain-specialized AI model designed to understand and explain telecom standards and technical documentation.
Unlike traditional LLMs, TelecomGPT-RT does not only generate answers, but it also provides structured reasoning about telecom technical content, including ownership domains, architectural implications, and cross-functional impacts.
This is the foundation for AI-native telecom operations and telecom intelligence.
RFGPT is a radio-frequency language model that brings RF perception into AI workflows. It enables natural-language understanding and structured outputs directly from RF representations. It converts IQ waveforms into spectrograms and encodes them into RF tokens injected into a decoder-only LLM, enabling RF-grounded reasoning without requiring heavy manual labeling. RFGPT supports RF-centric use cases such as modulation and technology recognition, RF overlap analysis, WLAN user counting, and 5G NR information extraction, enabling AI-native radio operations.
Telecom Knowledge graphs: Telecom Knowledge Graphs provide a structured grounding layer for telecom AI. They link entities across standards and engineering knowledge to improve correctness and traceability. They support reliable retrieval and reduce ambiguity in model answers, especially for standards-heavy questions, by grounding outputs in structured relationships and verified sources.
Knowledge graphs strengthen governance and trust, enabling provenance-aware reasoning and more robust integration into telecom operational workflows.
Open Telco Benchmarks: Open Telco Benchmarks provide standardized evaluation to measure whether a model is truly telco-ready. They focus on telecom-specific capability, not generic language performance. They include domain-centered evaluations that probe standards understanding and telecom reasoning (e.g., telecom Q&A, standards structuring, and quantitative telecom math). In DFI, these benchmarks drive the engineering loop: model selection, fine-tuning targets, regression testing, and partner pilot validation with measurable KPIs.
DFI develops and showcases deployment-ready demos and products that translate applied research into operational systems. Our deliverables are engineered for real environments: telecom infrastructure, enterprise IT, edge/cloud platforms, and secure government deployments, supported by clear governance, integration pathways (APIs/SDKs), and measurable performance targets.
Each demo or product is built within DFI’s project-driven model: seeded internally for rapid prototyping, then matured through partner-led vertical tracks with validation, scalability testing, and commercialisation options. Outcomes can include open-source releases, partner deployments, shared IP, and licensing pathways, depending on the track structure and partner needs.
NeuroWave: NeuroWave is DFI’s Private 5G AI Platform designed to deliver secure, high-performance, and energy-efficient access to LLMs over telecom infrastructure. It enables organisations to deploy and use LLM capabilities with strong control over data locality, latency, and compute usage, making LLM access practical for sensitive enterprise and government workflows. NeuroWave is positioned as a game-changer for scalable AI deployment on telecom infrastructure. Instead of moving data to public clouds, NeuroWave supports a model where AI services can be brought closer to users through operator-grade infrastructure, unlocking new business opportunities for telecom operators to offer AI services as a managed, secure product.
Arabic Voice-to-Slice: Intent-Based Networking for 6G enables network operators and customers to create, manage, and adjust network slices using natural language instructions in Arabic, through either voice or text. Instead of manual slice configuration, users can express service needs in everyday Arabic, making advanced network capabilities faster to activate and easier to operate across diverse stakeholders. When a user says something like “أريد في شبكة فائقة السرعة”, the system combines Arabic speech recognition with a telecom-aware language model to interpret the request, extract the technical intent, and convert it into a structured policy. This policy is then sent to the BubbleRAN orchestration platform to automatically create or modify the network slice end-to-end, enabling zero-touch provisioning and consistent operational execution.
LLM for Energy Management (LLMOPT): LLMOPT is an energy-management demo track aimed at supporting operational teams with decision support and optimization assistance grounded in real telemetry and constraints. It targets use cases where operators need fast, explainable insights and recommendations while keeping humans in the loop.
DFI builds partnerships and collaborations to ensure our research translates into deployable systems and trusted platforms, which means we engage with global industry ecosystems, open-source communities, and international standards and research forums so our outputs remain interoperable, adoption-ready, and aligned with real operational requirements.
Through GSMA engagement, DFI connects directly with the global mobile industry ecosystem and operator priorities, enabling our AI-native network work and telecom-facing platforms to reflect real deployment constraints, real adoption pathways, and practical requirements for large-scale operator environments.
DFI strengthens open innovation and interoperability by aligning with mature open-source governance and community engineering practices, supporting reference implementations and platform components that are reusable, auditable, and easier to integrate across heterogeneous infrastructures.
DFI participates in the AI-RAN Alliance to contribute to and benefit from global efforts shaping AI-native radio access networks, keeping our research, prototypes, and evaluation practices aligned with emerging directions for performance, robustness, interoperability, and operational readiness.
DFI aligns with ETSI to connect research outputs to international standardization pathways that enable multi-vendor interoperability and large-scale deployment, ensuring our system designs and interfaces remain compatible with evolving standards and real-world compliance expectations.
DFI engages with WWRF to strengthen long-horizon research connectivity and international collaboration, supporting strategic dialogue, research exchange, and joint initiatives that help position KU and DFI within the ecosystem shaping the next decade of wireless and intelligent digital infrastructure.