Distributed scheduling of recording tasks with interconnected servers

We consider a system with multiple interconnected video servers storing TV programs that are received through satellite antennas. Users, equipped with set-top boxes, submit requests for TV programs, to each of which they assign a utility value according to their preferences. We develop a distributed scheduling algorithm that selects the programs to be recorded and the servers to store them, so that a high total utility is generated to the users' population. Our scheduling algorithm is based on the programs' broadcasting information, the users' preferences, the constraints regarding the capabilities of simultaneous recordings and storage, and the system's topology. In fact, servers belonging to the same cluster co-operate in order to attain increased efficiency by exchanging content through streaming or replication. The efficient performance of our scheduling algorithm is shown by means of experiments. The algorithm constitutes a practically applicable solution, already implemented and integrated in the testbed of the IST project UP-TV.