Anomaly detection from distributed data sources

Home Home
Members Members

Resources: N/A

Anomaly detection is the task of uncovering abnormal or unusual patterns from the data. The problem is well-studied when the data is available at a central location. However, many aviation datasets are distributed across diverse geoghraphic locations and centralizing them is often very difficult due to bandwidth, storage, computation and other constraints. Also many emerging sensing technologies are capable of wireless communication making data collection easier over a wide region. Fast and accurate detection of anomalies in such environments would require algorithms which are in-network, scalable, communication efficient, and above all converge to the exact same result compared to a centralized execution. Local detereministic distributed algorithms may provide possible solutions.