Analysis of user demand patterns and locality for YouTube traffic

Abstract

Video content, of which YouTube is a major part, constitutes a large share of residential Internet traffic. In this paper, we analyse the user demand patterns for YouTube in two metropolitan access networks with more than 1 million requests over three consecutive weeks in the first network and more than 600,000 requests over four consecutive weeks in the second network.In particular we examine the existence of "local interest communities", i.e. the extent to which users living closer to each other tend to request the same content to a higher degree, and it is found that this applies to (i) the two networks themselves; (ii) regions within these networks (iii) households with regions and (iv) terminals within households. We also find that different types of access devices (PCs and handhelds) tend to form similar interest communities.It is also found that repeats are (i) "self-generating" in the sense that the more times a clip has been played, the higher the probability of playing it again, (ii) "long-lasting" in the sense that repeats can occur even after several days and (iii) "semiregular" in the sense that replays have a noticeable tendency to occur with relatively constant intervals.The implications of these findings are that the benefits from large groups of users in terms of caching gain may be exaggerated, since users are different depending on where they live and what equipment they use, and that high gains can be achieved in relatively small groups or even for individual users thanks to their relatively predictable behaviour.

Power and cost modeling for 5G transport networks

“…In such a case, large size caches can be placed at each of the MN unlike small caches placed at the APs in Case II. Referring again to the YouTube and Netflix cache models in [8] and [9], we get a higher offloading factor by employing large size caches, i.e., 100 TB. When the caches are placed at the MN, the offloading factor for the YouTube traffic is equal to 51%, while a Netflix cache pre-fetching the top 20% of the movies can offload up to 98% of the traffic.…”

“…After scaling the analysis of YouTube traffic in [8] to the Megacity values, we conclude that we can offload the YouTube traffic by 24% when the cache is placed at the AP. According to the cache model for Netflix traffic presented in [9], we find that by placing the top 100 watched movies in a cache at the AP, we can achieve an offloading factor of 77.7% for the Netflix traffic.…”

“…According to the cache model for YouTube traffic presented in [8], the highest cache hit-rate is achieved when the cache size is equal to the total size of all the unique video clips requested in a 10 days period. After scaling the analysis of YouTube traffic in [8] to the Megacity values, we conclude that we can offload the YouTube traffic by 24% when the cache is placed at the AP.…”

Cacheability of YouTube videos in cellular networks

“…A more recent study by Arvidsson et al analyses the demand patterns for YouTube of 35, 000 mobile devices over several weeks [3]. The authors study the cacheability of YouTube content on a traditional network, with cache proxies at the access points, whereas we study the cacheability in cellular networks.…”

Performance analysis of local caching replacement policies for internet video streaming services

“…the so-called user generated content (UGC) video services such as YouTube or TV-on-demand services, much research has been carried out to study video streaming traffic patterns and user behavior in order to investigate the potential of local network caching gains [10][11][12][13][14]. The results show that online streaming video services usually have significant traffic locality and hence good potential local network caching gains if local network cache servers with an unlimited cache size were employed [11][12][13]. Nevertheless, as traffic patterns of different VoD services are dependent on the types of video content, potential local network caching gains can vary significantly.…”