Abstract :
Server-based adaptive video streaming is gaining popularity in recent years. This is because clients (client-based) and in-network devices (network or proxy-based) are not powerful enough to run state of the art adaptation algorithms, for example, traffic shaping and machine learning. When decision making is placed at the server new and exciting possibilities are obtained for next best segment selection. This work highlights server-based solutions to adaptive video streaming. It provides a taxonomy of current state of the art solutions. It then illustrates various approaches used for server-based adaptive video streaming. Advantages and disadvantages are discussed. Network-assisted or in-network DASH solutions have certain advantages over traditional client-based approaches. It is proposed that the sharing of information would result in better network and client bandwidth estimations.
This measure would ensure better next segment selections. In this paper a novel network-assisted DASH taxonomy is proposed. It consists of cache-based, optimization, rate-quality model, and co-operative elements. Recent approaches using the elements of the taxonomy are illustrated. These approaches show the advantages of using network-assisted entities in DASH-based systems.
Keyword :
Server-based; adaptive video streaming; traffic shaping; machine learning; taxonomy; networkassisted; in-network; bandwidth; segment; cache; optimization; rate-quality; co-operative; DASH