A Mobile Edge Computing–Based Architecture for Improved Adaptive HTTP Video Delivery Yue Li *‡ , Pantelis A. Frangoudis ‡ , Yassine Hadjadj-Aoul ‡ , Philippe Bertin * * Orange Labs, Cesson S´ evign´ e, France ‡ IRISA/University of Rennes 1, France Email: *‡ yue2.li@orange.com, ‡ {pantelis.frangoudis,yhadjadj}@irisa.fr, * philippe.bertin@orange.com Abstract—Dynamic Adaptive Streaming over HTTP (DASH) is currently a widely adopted technology for video delivery over the Internet. DASH offers significant advantages, enabling users to switch dynamically between different available video qualities responding to variations in the current network conditions during video playback. This is particularly interesting in wireless and mobile access networks, which present unexpected and frequent such variations. Moreover, mobile users in these networks share a common radio access link and, thus, a common bottleneck in case of congestion, which may cause user experience to degrade. In this context, the Mobile Edge Computing (MEC) emerging standard gives new opportunities to improve DASH performance, by moving IT and cloud computing capabilities down to the edge of the mobile network. In this paper, we propose a novel architecture for adaptive HTTP video streaming tailored to a MEC environment. The proposed architecture includes an adaptation algorithm running as a MEC service, aiming to relax network congestion while improving user experience. Our mechanism is standards-compliant and compatible with receiver-driven adaptive video delivery algorithms, with which it cooperates in a transparent manner. Keywords—DASH, QoE, Mobile Edge Computing. I. I NTRODUCTION Video streaming is becoming one of the most popular services for mobile consumers. In fact, mobile video traffic, which accounted for 55% of the total mobile traffic in 2015, will represent more than 70% in 2020 [1]. This significant growth is accompanied by the wide adoption of Dynamic Adaptive Streaming over HTTP (DASH) standards [2], [3], particularly suitable for such network settings. DASH allows clients to switch between different video qualities (and, thus, bitrates) to improve viewing experience, matching current network conditions and device capabilities. However, when mobile clients share the same bottleneck link, as is often the case for mobile networks, the DASH protocol can show instabilities, unfairness and bandwidth underutilization [5]. Therefore, there is a need to be aware of the end-user device context and network conditions to optimize video delivery. The emerging ETSI standard on Mobile Edge Computing (MEC) [6] may play an important role in this direction. MEC allows leveraging the cloud computing power by deploying application services at the edge of the mobile network. This can facilitate content dissemination within the access network and can offer new business opportunities by integrating Mobile Network Operators (MNO) into the video delivery value chain. A key component for enabling MEC are servers integrated within the operator’s Radio Access Network (RAN) (e.g., 3GPP, Wi-Fi or small cells). MEC opens the door for autho- rized third parties, such as content providers (CP), to develop their own applications hosted in the MEC servers. These applications can add the flexibility to handle the traffic from/to mobile users. Moreover, operators can expose their RAN edge API to authorized third parties to provide them with radio network information in a real-time manner. We propose a novel architecture for adaptive HTTP video streaming in a MEC context (Section III), introducing an adaptation algorithm running on a MEC server, which aims to reduce network congestion while offering Quality of Exe- prience (QoE) improvements for mobile users. The main idea is to dynamically control the video representations available to clients based on the current network status, thus driving the client-side video adaptation mechanism. To achieve this, we exploit (i) the awareness of network-level information, which could be retrieved via a MEC API exposed by the MNO, and (ii) specific features of the DASH standard. Our approach illustrates a use case for a mutually ben- eficial CP-MNO collaboration over the MEC infrastructure: We show (Section IV) that fairness among mobile users is guaranteed, the overall perceived QoE is improved, and, by responding to congestion, better network resource utilization is achieved. Also, by actively involving the MNO in the video delivery chain, our solution allows for compensating the loss of MNO average revenue per user after the introduction of flat rates. It is worth noting that the proposed mechanism is standards-compliant and transparently coexists with client-side bitrate adaptation algorithms. Indeed, the final quality level se- lection decision is still made by the clients, which corresponds to the current DASH client standard specifications. II. BACKGROUND AND RELATED WORK Dynamic Adaptive Streaming over HTTP, also known as 3GP-DASH (3GPP) [3] or MPEG-DASH (MPEG ISO/IEC) [2], is a popular standard for video streaming over the Internet, allowing for improved user experience in the presence of variable network conditions. DASH aims to avoid playout interruptions, which are known to be harmful for the perceived video quality [4]. DASH servers offer a variety of quality levels (i.e., bitrates) for the same content. When a net- work degradation (e.g., reduced available bandwidth, increased latency) is identified by the video client, the latter can switch