University of Sharjah, UAE · Assistan Dean for the College of Computing and Informatics · Profile
Dr. Bassel Soudan is an Associate Professor of Computer Engineering at the University of Sharjah, United Arab Emirates. He received his Ph.D. in Electrical Engineering from the Illinois Institute of Technology in 1994 and brings extensive experience in both academia and industry, including his work at Intel Corporation in the United States. His research focuses on artificial intelligence, cybersecurity, the Internet of Things (IoT), embedded systems, and renewable energy, with particular emphasis on intelligent security solutions and the optimization of computing and energy systems. Throughout his academic career, Dr. Soudan has held several leadership positions and contributed to the development and accreditation of engineering programs. He has authored numerous scientific publications in international journals and conference proceedings and actively contributes to the research community through editorial and peer-review activities. He is a Senior Member of IEEE and a recipient of multiple awards recognizing his excellence in teaching and professional contributions.
From Lab to Rubble: Autonomous Air–Ground Robot Teams for Disaster Response
In the first hours after an earthquake or building collapse, the chance of finding survivors falls quickly, yet rescue teams often work blind and at great personal risk. Autonomous robots promise to change this. Aerial drones can survey a disaster area rapidly, and ground vehicles can carry supplies and sensors into places no aircraft can reach. Turning that promise into deployed systems, however, exposes hard problems that laboratory demonstrations rarely face. Robots must navigate without GPS inside damaged structures whose layout no longer matches any map. They must operate in confined, cluttered spaces, process sensor data onboard in real time, and coordinate across heterogeneous air and ground platforms over unreliable communication links. This keynote presents a long-term vision for cooperative UAV–UGV fleets that search for victims, deliver aid, and restore voice communication with trapped survivors. It draws on ongoing research into indoor localization in changed environments and drone inspection of confined spaces. The talk maps the gap between today's lab results and field-ready systems, and outlines the open research problems in autonomy, edge intelligence, and resilient networking that must be solved to close it.
Key Topics
University of Milan, Italy · IEEE Senior Member · ACM Distinguished Scientist · Profile
Dr. Ernesto Damiani is the Senior Director of Robotics and Intelligent Systems Institute at Khalifa University, UAE. He is also a Professor in the Electrical and Computer Engineering department and Director of the Khalifa University Center for Cyber Physical Systems (C2PS). He is on extended leave from the Department of Computer Science, Università degli Studi di Milano, Italy. His research interests include secure service-oriented architectures, privacy-preserving Big Data analytics and Cyber-Physical Systems security. With 494 publications listed on DBLP, he is considered among the most prolific European computer scientists. He is a senior member of the IEEE and ACM Distinguished Scientist.
Securing the Agentic Stream: From TESLA Hash Chains to Model-Independent Guardrails
Multi-agent systems powered by Large Language Models (LLMs) are transitioning from autonomous conversational demonstrators to distributed execution pipelines managing mission-critical operations. In these architectures, dynamic working memory—materialized as streaming token contexts and shared Key-Value (KV) cache structures—is passed across heterogeneous agents, retrieval pipelines, and external tools. However, modern inference engines treat working context as an implicit trust zone. This architectural blind spot exposes collaborative agent workflows to context poisoning, indirect prompt injections, state tampering in transit, and cache desynchronization. Conventional cryptographic defenses, such as asymmetric public-key signatures (PKI) over dynamic token streams or high-dimensional tensor pages, impose latency penalties that cripple real-time inference throughput. In this keynote, we present a unified framework bridging lightweight stream authentication with model-independent security architectures to establish Zero-Trust context provenance for multi-agent systems. We revisit the core mechanics of our Two-Level Initialization μTESLA (TLI-μTESLA) protocol, which decoupled stream bootstrapping from packet-rate message authentication through hierarchical one-way hash chains. We translate this paradigm into Context Hash Chains (CHCs) for agentic memory: high-level master chains bind workflow safety contracts and system invariants, while ephemeral low-level chains recursively anchor streaming token slices and agent handoffs at sub-microsecond overhead. Furthermore, we demonstrate how this lightweight cryptographic substrate interfaces with model-independent guardrails developed under systematic STRIDE threat modeling. By treating guardrails as out-of-band attestation services, the release of hash chain disclosure keys is dynamically gated upon automated security audits of intermediate context (such as raw retrieval payloads). Using a concrete, worked-out retriever–writer pipeline, we illustrate how context hash chains detect tampering in transit, neutralize prompt injection attacks, and enforce strict execution proofs prior to tool invocations, offering a reproducible, low-latency foundation for trustworthy multi-agent systems.
Key Topics