Edge intelligence capability is expected to enable the development of a new paradigm integrated with edge computing and artificial intelligence. However, due to the multisource nature, heterogeneity, and a large scale of the sensory data, it is necessary to improve the data processing and decision-making capacity for the edges. Hence, this paper asserts that information fusion is an important technique to power the capacity of edge intelligence in terms of collection, communication, computing, caching, control and collaboration. Specifically, it provides a comprehensive investigation of four representative scenarios assisted by information fusion at the edge, i.e., multisource information fusion, real-time information fusion, event-driven information fusion, and context-aware information fusion. Moreover, it discusses the future directions and open issues in this field.
- Context-aware,
- Event-driven,
- Information fusion,
- Multisource,
- Real-time
IR deposit conditions: