IEEE ISCC 2025

The Internet of Things (IoT) has rapidly expanded, with applications spanning industrial automation, smart cities, healthcare, logistics, etc. By 2030, the IoT landscape is expected to host over 1 trillion devices, ranging from simple wearables to advanced UAVs. These devices will be connected through different network technologies. Also, they will demand diverse Quality of Service (QoS) requirements, in terms of latency, reliability, and security. The integration of Artificial Intelligence (AI) services along the Cloud-to-Things Continuum will necessitate new paradigms for computing, communication, and resource management, while addressing sustainability challenges such as energy efficiency, device disposal, and environmental impact. To address these challenges, this tutorial explores key advancements and strategies shaping the Pervasive IoT. Topics include: battery-less IoT and intermittent computing; the emergence of new communication technologies (such as 6G, WiFi7, and LiFi) to support diverse application scenarios; and the Cloud-to-Things Continuum (C2TC) for scalable and efficient resource management. Strategies for optimal joint computing-network resource allocation and dynamic service reconfiguration, will be required to ensure QoS compliance for critical applications, even in dynamic environments. By surveying the latest research and discussing open issues, this tutorial aims to provide insights into the innovative approaches shaping the future Pervasive IoT ecosystem.

The tutorial will cover an in-depth study of the convergence of post-quantum cryptography (PQC) and quantum key distribution (QKD) in the 6G domain. The first part will provide background information on PQC, QKD, and B5G, and the second part will discuss the integration of these technologies into the 6G network architecture. In the second part, there will be a step-by-step demonstration of PQC integration within the user equipment communications and registration processes. There will also be a QKD demonstration in which the secure key exchange will be implemented and run within the quantum simulation environment. The participants do not need to have any prerequisite components. Application of PQC and QKD concepts in the security of telecommunication networks is an emerging topic and will be very important in the coming years as well for the development of secure 6G networks. The attendees of this tutorial will have the chance to learn more about security threats and potential techniques to provide defense against them from the perspective of 6G network infrastructures. Some special features of this tutorial are a clear link between the PQC and QKD ideas (including proper algorithm selection and establishing quantum channels) and recent developments in PQC, an overview of standardization efforts in 6G, and how these can improve the security of telecommunications networks.

As we accelerate into the era of 6G and hyper-connectivity, intelligence is no longer centralized. It is ambient, distributed, and increasingly agentic. This keynote explores how Edge Intelligence, Large Language Models (LLMs), and next-generation networks are converging to create a powerful new paradigm: a world where machines reason, adapt, and collaborate autonomously at the network's edge. From the battlefield to the city street, intelligent agents embedded in edge environments are already transforming critical sectors. I will delve into recent advances, including LLM-powered autonomous defense vehicles, real-time multimodal traffic cognition systems, and language-model-guided network optimization. The audience will gain insight into how techniques such as prompt engineering, federated learning, and lightweight language models are enabling secure, low-latency decision-making without compromising trust or privacy. Drawing from real-world implementations and collaborative research, this keynote outlines a bold vision for the future: a seamless edge ecosystem that brings together distributed intelligence, multimodal data, and next-generation connectivity to support responsive and scalable AI solutions. The edge is no longer the edge. It is the origin of infinite intelligence.

In this speech we will describe the transformative impact of multi-modality approaches in healthcare. Traditional single-modality methods have long limited the scope of medical decision-making, but we are now at a turning point. The convergence of advanced technologies and diverse data sources has opened unprecedented opportunities to revolutionize patient care. This keynote will examine where we are today in the integration of multi-modality approaches, focusing on their role in enhancing disease diagnosis and prognosis. By fusing medical images, bio-signals, clinical records, and other critical data, these methods provide a holistic view of a patient's condition, paving the way for truly personalized medicine. We will explore how multi-modality approaches enable a comprehensive patient profile, integrating genetic, imaging, and clinical data to inform precise, tailored treatments. Looking ahead, we will address the technical challenges in synthesizing heterogeneous data and highlight the emergence of deep learning as a groundbreaking tool for multi-modal integration. Together, we'll uncover the possibilities and potential of multi-modality in shaping the future of medical support systems.

This talk will go through a journey of critical developments in systems security over the last 25 years. The journey starts from a summary of high-level objectives, protecting user privacy and sensitive information and ensuring the continuous operation of critical infrastructures. It then reviews key underpinnings - cryptography, interoperability via standardization, wide-spread adoption of open-source and commoditized hardware components. Recently, several challenges to these underpinnings have become evident: traditional asymmetric cryptography is at risk due to the emergence of quantum computing. Supply chain security, for both hardware and software components are recognized as a critical area of concern. Finally, the rise and accelerated adoption of AI is imposing additional challenges while at the same time offering huge opportunities. We discuss how the systems security community is trying to address these challenges, including the transition to Post Quantum Cryptography, securing AI models and systems and leveraging AI for Security.

Each generation of telecommunication systems brings additional levels of sophistication to the services offered to end-users. The Ultra Reliable Low Latency Communications (URLLC) services (e.g. remote robotic surgery) promised by the fifth-generation (5G) are compelling examples. They are a far cry from the Short Message Service (SMS) offered by the second-generation (2G), and the simple multimedia services offered the third and fourth generations (3G/4G). The deployment of 6G systems is expected for the 2030s, and much more sophisticated services (e.g. immersive holographic type - communications services) are expected. Clouds are the pillars of 5G and Beyond (5GB) due to the fact that features such as elasticity, scalability, and provisioning on-demand can successfully tackle the everlasting challenges such as lack of flexibility and over provisioning faced by telecommunication systems. 6G requirements are now known and are far more stringent than their 5G counterparts. Expected end-to-end latency for instance is now 0.1 milli-second instead of the 1 milli-second that is hardly met nowadays. Clouds for 5G will certainly fail when it comes to meeting 6G challenges. Thus, the need of a new generation of clouds for 6G. However, the road ahead from clouds for 5G to clouds for 6G will certainly be bumpy due to the numerous challenges. In the first part of this keynote speech, we will introduce the expectations of 6G systems on clouds, and discuss why clouds for 5G cannot meet them. In the second part, we will sketch the research directions that may bring us to clouds for 6G. The third part will show that clouds alone will not be sufficient for 6G. It will be necessary to complement them by other paradigms. In-Network Computing (INC) is a good candidate. This is likely to bring us to a paradigm of "cloud-edge continuum enriched by INC" for 6G.

The importance of software in communication environments has significantly increased over the past decades, thanks to the widespread adoption of virtualization techniques, Cloud-native solutions, and advancements such as service-based architecture, user/control plane decoupling, and network slicing, giving rise to the concept of Network Softwarization. Software-defined Networking (SDN) advocates for a logically centralized view of network infrastructure and resources, utilizing open, standardized interfaces to control packet processing and forwarding. Network Function Virtualization (NFV) imagines software-based network elements and functions, applying a Cloud-like (*aaS) approach to virtualized network infrastructures, offering benefits such as scalability, elasticity, mobility, and replicability. These concepts collectively contribute to Network Programmability, facilitating a programmatic approach to network resource management and service deployment. The goal of this tutorial is twofold: on one side, to review how network management and automation evolved towards a more comprehensive network orchestration approach; on the other side, to show participants that they too can be part of this software (r)evolution in communication networks, demonstrating how easy it can be to "build your own SDN controller" leveraging the available architectures and technologies. This tutorial will address current challenges of network orchestration, such as interoperability and automation across multi-vendor, multi-technology environments characterized by dynamic resource availability, typical of edge/fog computing scenarios. Then, to effectively meet the demands of network orchestration, the idea of disaggregating the network control plane will be illustrated, and a playground framework will be introduced for the design and evaluation of microservice-based SDN controllers, demonstrating how to use it to effectively develop and test customized functionalities for a fully fledged SDN controller running in a sandboxed environment. The ultimate objective is to facilitate participants in understanding the current developments and best practices - and anticipating the upcoming ones - in network softwarization, so that they can drive innovation in their field of work.

Transport Layer Security (TLS) is a widely used protocol for secure channel establishment. However, TLS lacks any inherent mechanism for validating the security state of the endpoint software and its platform. To overcome this limitation, recent works have combined remote attestation (RA) and TLS, named attested TLS. We present the state-of-the-art attested TLS protocols in an emerging application domain, namely confidential computing, and use federated learning as the running example. We present an approach based on ProVerif to discover attacks on such protocols. After introducing network security (via TLS 1.3) and endpoint security (via RA), the tutorial explores the ways to combine the two protocols. Then, we introduce formal methods and the approach to discover protocol-level attacks in attested TLS protocols. Finally, we present open research problems in this domain for the attendees.

Recent advances in computational methods for CSE and AI - such as computational drug discovery and LLMs - combined with new parallel computing hardware like GPUs, demand more than just parallel programming software (e.g., CUDA, MPI, Hadoop) to harness computing power for these applications. To ensure programmer productivity, parallel programming software relies on critical parallel computing system software - compilers, runtime systems, job schedulers - that enable performance, efficiency, and scalability, correctness, and resilience of the application. Moreover, as applications grow in complexity and computational demands evolve, applications require support of such system software across different parallel programming abstraction layers, from intra-node computation to cluster-level parallelism to cross-datacenter work orchestration. This keynote explores advances in system software frameworks that have enabled traditional scientific and emerging AI workloads. We will focus discussing systems software innovations across four layers of programming abstraction: (1) low-level parallel programming models (e.g. CUDA, OpenMP, MPI), (2) libraries for base-language-parallelism (e.g. Charm, Kokkos, C++ and Fortran StdPar), (3) Domain-Specific Libraries (DSLs) offering parallel programming (e.g. Triton, PyTorch, PetaBricks), and (4) AI agent and scientific workflow orchestration tools (e.g. NVIDIA's AgentIQ, Pegasus). In doing so, we will share real-world parallel computing systems software from our research that has supported production workloads in science and AI at the Department of Energy (DOE), and we will discuss how such systems software used in the DOE has impacted systems software for today's rapidly evolving computing needs to enable high-demand genAI in industry. We will close with a perspective of how current parallel computing system software will extend to emerging and future application domains such as robotics, blockchain, and other domains involving real-time computing systems.