Periodic Reporting for period 1 - VSDN (Video streaming services with Software-Defined Networking)
Reporting period: 2018-01-01 to 2019-12-31
Recent advances in mobile devices and video streaming services have motivated large-scale media consumption both in fixed and mobile environments. In this context, the main objective of network operators, media creators and service providers is to improve the QoE of end-users by providing high-quality services and interactive mechanisms for seamless adaptation to the specific network conditions of each user. However, the initial design of the Internet as a best-effort network makes it inappropriate for high-volume and bandwidth-intensive applications like video streaming, and recent studies show that just a couple of seconds of video buffering have a significant impact on users’ abandonment rate. Ideally, a service provider should place servers close to the end-users to overcome the limitations of the Internet architecture: Content Distribution Networks (CDNs), overlay networks built on top of the Internet infrastructure , offer the use of their distributed content delivery infrastructure service which can handle rapid changes in traffic demand. However, this is neither scalable nor desirable. Additionally, network operators currently employ centralized architectures (esp. in mobile networks) that lead to long communication paths between end-users and servers, waste of network resources and increased delays. This issue becomes increasingly important with the wider deployment and advances in wireless networks, which have more stringent resource limitations and fast-changing conditions.
VSDN developed a cross-layer QoE-aware media delivery platform of video streaming services and propose a network architecture, which combines service negotiation at the application level, with the concept of SDN, leading to optimal (in a QoE sense) path assignment. Specifically, the research objective of VSDN was to develop a novel SDN-based network architecture for network control and optimal resource allocation for video streaming: the aim was to investigate SDN principles to perform optimized path assignment while achieving the desired level of QoE. This includes the specification of the SDN controller functions which allocates resources to the network elements and configures the routing mechanisms. Moreover, the objective was to develop an end-to-end QoE-aware video streaming platform that adapts content delivery, considering their individual network conditions of each user and the overall topology and resources of the network in order to maximize the end-user QoE.
VSDN developed a cross-layer QoE-aware media delivery platform of video streaming services and propose a network architecture, which combines service negotiation at the application level, with the concept of SDN, leading to optimal (in a QoE sense) path assignment. Specifically, the research objective of VSDN was to develop a novel SDN-based network architecture for network control and optimal resource allocation for video streaming: the aim was to investigate SDN principles to perform optimized path assignment while achieving the desired level of QoE. This includes the specification of the SDN controller functions which allocates resources to the network elements and configures the routing mechanisms. Moreover, the objective was to develop an end-to-end QoE-aware video streaming platform that adapts content delivery, considering their individual network conditions of each user and the overall topology and resources of the network in order to maximize the end-user QoE.
During the project, we have developed a novel QoE model for long-term video streaming applications, which takes into consideration the typical artefacts that are introduced in a streaming application, such as startup delay, stalling during playback, and video loss. The QoE model was developed based on subjective tests that were designed and conducted by the fellow. The developed QoE model was employed in a video client to enable video streaming based on QoE metrics.
Moreover, we developed a QoE Monitoring Framework using SDN, in order to detect when the network conditions, such as packet loss and delay affect the video streaming and we developed a methodology to preserve the QoE in video applications. This is achieved by using an SDN Controller and implementing extra functionality on top of it in order to change the traffic’s transmission path to an alternative one, when QoE falls below a specified threshold. Also, we evaluated the proposed approach in a variety of network conditions to demonstrate the validity of the proposed methods.
The subjective tests have been contributed to ITU-T Study Group 12 Question 14, which provides recommendations on “Performance, Quality of Services and Quality of Experience” for multimedia services over networks. The fellow participated in several physical meetings of Question 14 of SG12 during the fellowship and contributed towards the standardization of the newly developed standard “Development of parametric models and tools for multimedia quality assessment”, and the outcome of the project will be directly applied to the new work item in Q.14 on adaptive streaming, named P.NATS which was standardized in December 2019 as ITU-T Rec. P.1204.
Moreover, we developed a QoE Monitoring Framework using SDN, in order to detect when the network conditions, such as packet loss and delay affect the video streaming and we developed a methodology to preserve the QoE in video applications. This is achieved by using an SDN Controller and implementing extra functionality on top of it in order to change the traffic’s transmission path to an alternative one, when QoE falls below a specified threshold. Also, we evaluated the proposed approach in a variety of network conditions to demonstrate the validity of the proposed methods.
The subjective tests have been contributed to ITU-T Study Group 12 Question 14, which provides recommendations on “Performance, Quality of Services and Quality of Experience” for multimedia services over networks. The fellow participated in several physical meetings of Question 14 of SG12 during the fellowship and contributed towards the standardization of the newly developed standard “Development of parametric models and tools for multimedia quality assessment”, and the outcome of the project will be directly applied to the new work item in Q.14 on adaptive streaming, named P.NATS which was standardized in December 2019 as ITU-T Rec. P.1204.
In the context of end-user quality improvement, previous work has proposed inclusion of different application-level network functions for negotiation and Quality-of-Service (QoS) optimization decision making . However, SDN has emerged as a networking paradigm in which the data and control planes are decoupled, and the control is logically centralized and programmable via standardized interfaces. The OpenFlow protocol provides an interface between the SDN controllers and the infrastructure elements. The interfaces between the control and the application layer provide the means for the applications to use network services and capabilities as needed, without knowing the network specifics, such as network topology. Thus, applications can issue requests (at the application layer), which are translated by the control layer to device-specific configurations. As the control layer in SDNs has a global view on the network topology graph, it is possible to implement control applications that use traditional graph optimization algorithms for traffic engineering. Such examples include the SDN-based architecture to optimize flows and user associations in wireless mesh networks , and the OpenFlow architecture for balancing the load between servers while considering the network capacity. While these works optimize flows with the goal to maximize the network throughput or minimize communication delays, it is not yet clearly quantified how the network parameters affect the perceived quality in video streaming services. From these studies it becomes evident that an understanding of the correlation between QoS, i.e. the network-related factors, and QoE is necessary in order to maximize quality.
The main focus of research in this area so far is on the rate adaptation algorithms and an evaluation in mobile heterogeneous environments . However, the evaluation in these works primarily focuses on objective quality indicators (e.g. stalling/re-buffering events, number of quality switches) without assessing their synergistic impact on the actual QoE . Indeed, MPEG-DASH and its relationship to QoE are not yet well researched; first ideas on this topic are based on objective metrics not exactly reflecting QoE. According to its definition, QoE is influenced by user expectations, context, and personal preferences. Therefore, capturing and understanding end-users’ QoE goes beyond purely measuring video quality . To the best our knowledge, there are no studies to study the cross-layer synergistic cooperation and optimisation between the network architecture, the network virtualisation controls and functions, and the optimal streaming client adaptation to optimize the overall QoE of the end-users.
The main focus of research in this area so far is on the rate adaptation algorithms and an evaluation in mobile heterogeneous environments . However, the evaluation in these works primarily focuses on objective quality indicators (e.g. stalling/re-buffering events, number of quality switches) without assessing their synergistic impact on the actual QoE . Indeed, MPEG-DASH and its relationship to QoE are not yet well researched; first ideas on this topic are based on objective metrics not exactly reflecting QoE. According to its definition, QoE is influenced by user expectations, context, and personal preferences. Therefore, capturing and understanding end-users’ QoE goes beyond purely measuring video quality . To the best our knowledge, there are no studies to study the cross-layer synergistic cooperation and optimisation between the network architecture, the network virtualisation controls and functions, and the optimal streaming client adaptation to optimize the overall QoE of the end-users.