Facebook (A)Live?

Abstract

The era of live-broadcast is back but with two major changes. First, unlike traditional TV broadcasts, content is now streamed over the Internet enabling it to reach a wider audience. Second, due to various user-generated content platforms it has become possible for anyone to get involved, streaming their own content to the world. This emerging trend of going live usually happens via social platforms, where users perform live social broadcasts predominantly from their mobile devices, allowing their friends (and the general public) to engage with the stream in real-time. With the growing popularity of such platforms, the burden on the current Internet infrastructure is therefore expected to multiply. With this in mind, we explore one such prominent platform -Facebook Live. We gather 3TB of data, representing one month of global activity and explore the characteristics of live social broadcast. From this, we derive simple yet effective principles which can decrease the network burden. We then dissect global and hyper-local properties of the video while on-air, by capturing the geography of the broadcasters or the users who produce the video and the viewers or the users who interact with it. Finally, we study the social engagement while the video is live and distinguish the key aspects when the same video goes on-demand. A common theme throughout the paper is that, despite its name, many attributes of Facebook Live deviate from both the concepts of live and broadcast.

A Large-Scale Measurement and Optimization of Mobile Live Streaming Services

“…As in other measurement-driven studies [28], [35], [41], our dataset covers only one MLS service provider. Given the huge number of sessions contained in the dataset and that the examined service provides similar functionalities as in other popular services (e.g., Facebook Live), the observations made in this paper as well as the optimizations are likely to be applicable to others.…”

“…View number features Daily median number of broadcast Cat. #2 Broadcaster viewing-specific [12][13][14][15][16][17][18][19] View number features [20][21][22][23][24][25][26][27] View duration features [28][29][30][31][32][33][34][35] View completeness features Cat. #3 Arrival time-specific [36][37][38][39][40][41][42][43] Historical arrival time features [44][45][46][47][48][49][50][51] Broadcasts time, Arv actual .…”

“…The behavioral patterns of MLS and general live streaming systems have been examined in several works. Raman et al [35] explored the video characteristics, and the social engagement of Facebook Live, and highlighted many unwatched broadcasts. Li et al [24] characterized the user behavior in the PPTV mobile live streaming system, and they examined the structure of the abandoned session problem.…”

“…This challenge is exacerbated by the unpredictable nature of user-generated uploads, making capacity management more difficult. For instance, although any user can broadcast, their popularity is highly diverse, with some videos going unwatched, whilst others gain huge traction [35]. Second, both broadcasts and views are much shorter than traditional live streaming [41].…”

“…To address these challenges and streamline MLS systems, we need a deep understanding of the user behavior in practical MLS services. Existing active measurements of MLS services mostly focus on the infrastructure of MLS platforms [35], [41] (like the CDNs), or the popularity of broadcasters [35]. Nevertheless, they do not provide a detailed analysis of both the broadcasters' and the viewers' behavior in terms of the network resource usage, video QoE and viewership characteristics, mostly because of the limitation of datasets.…”