Cell BE and Bluetooth applied to Digital TV

This paper presents the results of a research in progress about a system for multiple users' behavior analysis and program recommendation for digital TV called SmarTV. The research explores a multi-user's behavior analysis and program recommendation using Bluetooth cell phones and a Playstation 3 working as a set-top box connected to a digital TV network with return channel or an IPTV network. Supposing that every viewer has a Bluetooth enabled cell phone, each mobile will identify one spectator to the set-top box wirelessly. Since the set-top box used is a Playstation 3, it is a full computer with IBM Cell BE processor capable of executing computer-intensive algorithms. This device was used to store the history of programs watched by every person near the TV connected to the set-top box, and to identify their profile and preferences using data mining algorithms. The program recommendation system is a hybrid of a content-based approach with a collaborative approach using a rate system, and uses the k-Means and k-NN data mining algorithms modified to explore the Cell BE features.