NEW (Extended) Deadline: August 26th, AOE.
Scope: This workshop solicits original work that advances the science of dynamically composing, operating, adapting, and assessing future intelligent, mission-critical IoT applications that operate in harsh, unfriendly, or adversarial environments. The motivating application examples include disaster response, first-responder support, rescue management, extreme environmental monitoring (e.g., monitoring volcanoes, nuclear plants, bio-chemical incidents, or contagious disease outbreaks), and systems that, by their very nature, are subject to frequent adversarial action such as security/anti-theft systems, intrusion detection systems, anti-jamming systems, and defense systems. A common thread across the above systems is the need for high resilience in the face of a broad array of threats, human or environmental. By soliciting original research on attaining resilient and dependable operation in such a broad spectrum of harsh IoT application contexts, the workshop aims to help the research community collectively distill fundamental insights, key concepts, best practices, and analytical foundations to support a next generation of IoT services for mission-critical applications in adversarial environments. Challenges such as heterogeneity, scale, and fast evolving dynamics are of great interest. Contributions may include, but are not limited to:
- Resilient performance-assurances in the face of threats
- Robust edge AI capabilities
- Robust exploitation of LLMs in IoT settings
- Foundation models for mission-critical IoT systems
- IoT adaptation to meet goals despite perturbations
- Handling model and environment uncertainty
- Accurate learning in adversarial conditions (adversarial machine learning)
- Formal verification of machine learning
- Optimization under uncertainty
- Resilient cyber-physical-human information fusion of contaminated inputs.
