MDP-based Network Friendly Recommendations

Controlling the network cost by delivering popular content to users, as well improving streaming quality and overall user experience, have been key goals for providers (CP) in recent years. While proposals improve performance, through caching or other mechanisms (DASH, multicasting, etc.) abound, works proposed turn problem on its head complement such efforts. Instead of trying reduce deliver every possible a user, potentially very expensive endeavour, one could leverage omnipresent recommendations systems nudge users towards low(er) cost, regardless where this is coming from. In paper, we focus latter problem, namely optimal policies “Network Friendly Recommendations” (NFR). A contribution use Markov Decision Process (MDP) framework that offers significant advantages, compared existing works, terms both modeling flexibility computational efficiency. Specifically show subsumes some state-of-the-art approaches, can also optimally tackle additional, more sophisticated setups. We validate our findings with real traces suggest up almost 2X 10X speed-up works.