Enabling Resilient Cyber-Physical-Human Infrastructures

Speaker: Prof. Nalini Venkatasubramanian

🗓 10:00 A.M.-10:45 A.M. (Taiwan), 6:00 P.M.-6:45 P.M. Nov. 26 (California)

Advances in mobile computing, cyber-physical systems, Internet-of-Things (IoT), cloud computing and big data technologies are making available new modalities of information and new channels of communication. They have enabled the interconnection of objects and data to provide novel services that are changing the landscape of cities and communities worldwide. The impact of these technologies are being felt as they provide new capabilities for personal sensing and in our vehicles and homes. They also have the potential to enable new levels of resilience and efficiencies in community-wide infrastructures such as water, power and transportation that serve as lifelines to human society. We discuss opportunities to create the future cyber-physical-human infrastructures (CPHI) and discuss challenges in operating them reliably and effectively. During large-scale disasters and unexpected events (e.g., fires, floods and earthquakes), infrastructures and the society they support are severely impacted – we discuss how operational CPHI systems can be morphed to gain improved situational awareness and better decision support for response personnel, agencies and citizens. The ability to ensure resilient operations under small events and large disasters requires intelligent data collection and data exchange from diverse devices and data sources and interpretation of this information for higher-level semantic observations. Drawing on our recent efforts in smart buildings, smart spaces and smart water infrastructures, I will discuss the role of IoT, adaptive middleware, and big data technologies to enable resilient CPHIs. The ability to combine novel technologies at multiple layers will open up new possibilities for resilient communities of the future.

Enhancing IoT Services for Situational Awareness using Mobility

Speaker: Ms. Fangqi Liu

🗓 11:00 A.M.-11:15 A.M. (Taiwan), 7:00 P.M.-7:15 P.M. Nov. 26 (California)

IoT deployments with media-rich sensors are revolutionizing smart city applications; the resulting data volumes place a heavy burden on long-range data transmission in IoT. Besides, some IoT applications like environmental monitoring or emergency surveillance are implemented in remote areas, where network infrastructures or sensors are unavailable. To complement the insufficiency of network resources and the unavailability of network infrastructures, we explore the ways to integrate mobility into IoT platforms to enhancing sensing and data transmission. As a use case of this, we explore using public transportation fleets to enhance data transmission for smart city applications in urban scenarios. We develop techniques to create a cost-effective data collection network using existing transportation fleets with predefined schedules to collect sensor data from IoT zones and upload them at locations with better network connectivity. In another use-case, we explore the usage of drone-based sensing to improve situational awareness in the high-rise fire scene. More specifically, we design a control platform for using multiple drones with visual recording devices to monitor the exterior of a high-rise structure in a coordinated fashion to collect image data for detecting the presence of fire/smoke, the open/close status of windows, and tracking the presence of humans.

Learning from Decentralized Datasets – Optimization and Communication Aspects

Speaker: Prof. Y.-W. Peter Hong

🗓 11:15 A.M.-12:00 P.M. (Taiwan), 7:15 P.M.-8:00 P.M. 26 Nov. (California)

Most modern applications in artificial intelligence (AI) rely on centralized data centers to perform the training and inference over large datasets. These data centers are often owned by industrial super-powers, such as Google, Facebook, and Baidu. Individual users or businesses must sacrifice their privacy in order to benefit from the AI services. Distributed (or federated learning) techniques allow learning to be moved from large data centers to distributed edge entities, where users have more control of both their own data and computational resources. In this case, local entities can collaborate to perform learning and inference tasks without explicitly exchanging local raw information with the data center.

In this talk, I will first introduce some basic optimization algorithms, such as gradient descent and alternating direction method of multipliers (ADMM), used for learning from decentralized datasets. Then, through the introduction of our recent works on distributed matrix factorization, distributed clustering, and personal authentication etc, we identify several key design challenges and possible ways to address them. The concept of learning from decentralized datasets is consistent with the recent trend in edge computing, where conventional cloud resources are moved towards edge servers or devices. Hence, we will also discuss some novel cross-layered communication or networking algorithms that can be used to reduce the communication overhead under the edge computing framework.