Abstract:

Online clustering algorithms might be adopted to search the degradation events that occur on the set-top-box (STB) devices that are used by television operators. In adaptive streaming systems, the bitrate changes might be considered as quality changes. The series of bitrate changes can be detected by online clustering algorithms that analyze logs in real-time. This short paper introduces the clustering of events collected from the STB devices connected to INEA television system.  The Mini Batch k-means and evoStream algorithms are used to cluster the events related to bitstream changes.