We are currently witnessing an evolution from building and home automation to smart homes, driven by progressing maturity of the Internet of Things and the use of artificial intelligence. However, signifi-cant technological challenges such as immature home intelligence, huge network and central server processing load; and embedded resource usage, still need to be addressed.
Until now, most of the research in this area has focused on centralized architectures for smart homes. This work contributes by developing a new distributed framework, which comprises a real-time dis-tributed system with autonomous behavior, parallel processing, context awareness, and node com-munication. In particular, it introduces a novel approach to adapt and distribute the artificial intelligence to match the distributed system architecture in the smart home. The proposed solution addresses important issues such as real-time learning, temporal detection with a high probability, battery lifetime, network communication, integration with smart objects, and embedded processing power.
A multi-agent smart object model is provided to support the artificial intelligence framework with a new distributed architecture. This model focuses on the embedded resources, the sensor frameworks, and the employed algorithms and leads to considerable savings in battery power consumption, processing resources and network load. Significant parts of the framework are simulated to validate their performance, and it is shown that the performance of the distributed system is comparable to state-of-the-art centralized smart home architectures.
In a larger perspective, the proposed framework supports and facilitates the coming era of Internet of Things. The distributed approach and elements of the framework can be applied in many related are-as, such as ambient-assisted living, intelligent transportation systems, and many other sustainable solutions based on ICT.
Date of defense: 16.12.2013