Bentaleb, A.Beğen, Ali CengizZimmermann, R.Harous, S.2017-10-252017-10-252017-101520-9210http://hdl.handle.net/10679/5692https://doi.org/10.1109/TMM.2017.2733344Due to copyright restrictions, the access to the full text of this article is only available via subscription.HTTP adaptive streaming (HAS) is receiving much attention from both industry and academia as it has become the de facto approach to stream media content over the Internet. Recently, we proposed a streaming architecture called SDNDASH [1] to address HAS scalability issues including video instability, quality of experience (QoE) unfairness, and network resource underutilization, while maximizing per player QoE. While SDNDASH was a significant step forward, there were three unresolved limitations: 1) it did not scale well when the number of HAS players increased; 2) it generated communication overhead; and 3) it did not address client heterogeneity. These limitations could result in suboptimal decisions that led to viewer dissatisfaction. To that effect, we propose an enhanced intelligent streaming architecture, called SDNHAS, which leverages software defined networking (SDN) capabilities of assisting HAS players in making better adaptation decisions. This architecture accommodates large-scale deployments through a cluster-based mechanism, reduces communication overhead between the HAS players and SDN core, and allocates the network resources effectively in the presence of short- and long-term changes in the network.engrestrictedAccessarticle19102136215100041124760000210.1109/TMM.2017.2733344Bitrate adaptation logicConvex optimizationDASHfastMPCHTTP adaptive streaming (HAS)InstabilityOpenFlowQuality of experience (QoE)Reinforcement learningScalabilitySoftware defined networking (SDN)Streaming architectureUnfairnessUnderutilization2-s2.0-85029151389