Constructing an Ambient Intelligence Development Framework: Constructing an Ambient Intelligence D evelopment F ramework Student: Bui Vu Hoang Co-supervisor: Assoc. Prof. Bui The Duy Co-supervisor: Dr. Vu Thi Hong Nhan
Content: Content Introduction Our Ambient Intelligence Development Framework Implementation & Results Conclusion Demo
Content: Content Introduction Our Ambient Intelligence Development Framework Implementation & Results Conclusion Demo
Introduction: Introduction Ambient Intelligence was Introduced by the Information Society Technologies Advisory Group in 2001. There are many Ambient Intelligence development frameworks Exp : WOSH [ http :// wosh.sourceforge.net/docs , 2007] CAMPUS [ Karin et al, 2008]
Contribution: C ontribution Problem: Ambient Intelligence requirements satisfaction Our solution: A framework that intuitively satisfies the requirements
Ambient Intelligence Requirements: Ambient I ntelligence Requirements Ubiquitous computing Distributed, collaborative Plug-n-play Ubiquitous communication No coupling Real-time data streams Intelligent user interfaces
WOSH: WOSH Advantages Feature extensive Suitable for large areas Disadvantages Coupling between components Possible duplications
CAMPUS framework: CAMPUS framework Advantages No coupling Global communication Disadvantages No data storage
MICA framework: MICA framework A framework for agent-based systems Developed by New South Wales University.
MICA: MICA Blackboard: A central place for Communication Data storage Agent and Blackboard transports Developers can focus on the component
MICA: MICA Transports’ mechanisms are slow No buffer to ensure data rate No mechanism to provide seamless data streaming components
Content: Content Introduction Our Ambient Intelligence Development Framework Implementation & Results Conclusion Demo
Our AmI Development Framework: Our AmI Development Framework Agent-based Based on MICA framework Supports real-time data streaming
Design goals: Design goals Plug-n-play Agents can join and start operating Collaboration Multiple agents can access a single data stream No direct coupling Agents operate without knowing the existence of other agents
Overall view: Overall view
Broadcaster & Receiver Registration: Broadcaster & Receiver Registration
Streaming Registration Monitor: Streaming Registration Monitor
Broadcaster Registration: Broadcaster Registration
Receiver Registration: Receiver Registration
Overall view: Overall view
Broadcaster stream: Broadcaster stream
Receiver stream: Receiver stream
Component design: Component design
Overall view: Overall view
Content: Content Introduction Our Ambient Intelligence Development Framework Implementation & Results Conclusion Demo
Implementation: Implementation Video encoder/decoder: Xuggle RTP stack: Efflux Java, Eclipse IDE Video format: MPEG4, 320x240p, 25fps 3 Intel Core i3 2.5GHz machines with 2GB of RAM running on Windows 7 Ultimate 32bit Wired network
Results: Results Avg. speeds Broadcaster Receiver Packets/s 2600 90 bits/s 30M 1000k Standard videoconferencing quality: 384kbps Standard audio quality: 128kbps High quality Skype video call: 500kbps
Content: Content Introduction Our Ambient Intelligence Development Framework Implementation & Results Conclusion Demo
Conclusion: Conclusion A framework design that supports Dynamic components Collaborative components No direct coupling between components A working prototype with high performance
Future work: Future work Create an actual small scaled AmI system in our lab with the framework Make improvements to the framework Fix bugs Add more features Long term goal: promote our framework to be a standard and high quality AmI development framework
Content: Content Introduction Our Ambient Intelligence Development Framework Implementation & Results Conclusion Demo
Demo: Demo
Thank you for listening: Thank you for listening