An Intelligent IoT World
- 講者I-Ling Yen 教授 (Department of Computer Science, University of Texas at Dallas)
邀請人：陳文村 - 時間2016-05-26 (Thu.) 10:30 ~ 12:30
- 地點資訊所新館106演講廳
摘要
With the rapid advances in CPS and IoT systems, a vast number and variety of sensors and actuators are likely to be deployed in the near future. However, most of existing CPS/IoT systems are static. We consider dynamically discover and compose IoT services to cope with dynamically arising situations. A three stage workflow is constructed for an intelligent IoT world: (1) Situation analysis and reactive decision making, (2) IoT service composition to realize the reactive actions, and (3) Service execution and control. Situation recognition can be achieved by video and image processing techniques. Reactive decisions can be made based on techniques such as expert systems, case based reasoning, and historical scenario mining. Service computing technologies can be leveraged to achieve dynamic IoT service discovery and composition. Service execution in IoT may involve a lot of control tasks. We plan to use crowd sourcing and gaming techniques to achieve better control.
In this talk, we specifically focus on techniques for IoT service discovery and composition. SOA technologies can be applied to discover and compose IoT services. However, existing service research focuses on software services while for an IoT service, the provider PT has a significant role. Moreover, the actions of one PT may impact the other PTs and the effects and interferences also need to be considered during composition. We extend the OWL service ontology to PT-SOA ontology to facilitate better modeling of IoT services and enable automated composition reasoning. We also develop new planning techniques to improve the reasoning speed in large scale IoT systems in order to support agile reactions. Autonomous IoT service discovery without a central registry is another topic that require deeper investigation. Existing works on semantic based routing can be used to discover services in peer to peer networks. However, IoT devices have limited computing power and energy constraints. We will discuss a semantic routing algorithm that requires low storage and low computation.