A software system that monitors the frequency bands of interest and performs thorough pattern analysis on the intercepted RF signals by first extracting key metadata features from the raw I/Q samples and then performing pattern analysis on the metadata records using state-of-the-art ML/AI algorithms, establishing a pattern-of-life for “normal behavior/scenario” and detecting RF anomalies caused by a change in underlying scenario/activity.

Some of the highlighted features and functionalities associated with the PoLICE system are:

– Integrated long-term (seasonal, daily) and short-term (seconds, subseconds) spectrum pattern-of-life characterization and anomaly detection via the multiple-stage learning networks consisting of the frontend metadata extractor ML network and backend metadata PoL learning network.

– Easily scalable for distributed learning with multiple input sensor covering a much wider bands of interest.

– State-of-the-art machine learning-based (ML/AI) time-series analysis approach for both the RF signal classification and metadata pattern learning.

– Proven capability to generalize the PoL learning engine to Navy real-world RF environment via rapid re-training / re-programming or transfer learning (enabled by the diverse training and testing environment generated through IFT RF Environment Simulator and IFT Metadata Generator).