User-Based Predictive Caching of Streaming Media

Abstract: Streaming media is a growing market all over the world which sets astrict requirement on mobile connectivity. The foundation for a gooduser experience when supplying a streaming media service on a mobiledevice is to ensure that the user can access the requested content.Due to the varying availability of mobile connectivity measures has tobe taken to remove as much dependency as possible on the quality ofthe connection. This thesis investigates the use of a Long Short-TermMemory machine learning model for predicting a future geographicallocation for a mobile device. The predicted location in combinationwith information about cellular connectivity in the geographical areais used to schedule prefetching of media content in order to improveuser experience and to reduce mobile data usage. The Long Short-Term Memory model suggested in this thesis achieves an accuracy of85.15% averaged over 20000 routes and the predictive caching managedto retain user experience while decreasing the amount of dataconsumed.