Multisensor Data Fusion in Pervasive Artificial Intelligence Systems

Intelligent systems designed to manage smart environments exploit numerous sensing and actuating devices, pervasively deployed so as to remain invisible to users and subtly learn their preferences and satisfy their needs. Nowadays, such systems are constantly evolving and becoming ever more complex, so it is increasingly difficult to develop them successfully. A possible solution to this problem might lie in delegating certain decisions to the machines themselves, making them more autonomous and able to self-configure and self-manage. This work presents a multi-tier architecture for a complete pervasive system capable of understanding the state of the surrounding environment, as well as using this knowledge to decide what actions should be performed to provide the best possible environmental conditions for end-users, in line with the Ambient Intelligence (AmI) paradigm. To achieve such high-level goals, the system has to effectively merge and analyze heterogeneous data collected by multiple sensors, pervasively deployed in a smart environment. To this end, the proposed system includes a context-aware, self-optimizing, adaptive module for sensor data fusion. Contextual information is leveraged in the fusion process, so as to increase the accuracy of inference and hence decision making in a dynamically changing environment. Additionally, two self-optimization modules are responsible for dynamically determining the subset of sensors to use, finding an optimal trade-off to minimize energy consumption and maximize sensing accuracy. The effectiveness of the proposed approach is demonstrated with the application scenario of user activity recognition in an AmI system managing a smart home environment. In order to increase the resilience of the system to highly uncertain and unreliable information, the architecture is enriched by a filtering module to pre-process raw data coming from lower levels, before feeding them to the data fusion and reasoning modules in the higher levels.