Yazılım tanımlı ağlar üzerinden yönetimli video servisleri

The best effort nature of the current Internet leads to inefficiencies for all parties involved in a video service; namely, network service providers (NSP), over-the-top (OTT) video service providers (VSP), and users (clients). From the perspective of the NSP, increasing volume of video services does not contribute to their revenues since NSPs commonly charge a flat monthly subscription fee for the best-effort Internet access. From the perspective of OTT VSP, unlike IPTV services that are offered over a managed private IP network, they cannot promise specific quality of experience (QoE) to their users as they rely on the best-effort service offered by some NSP. From the perspective of end-users, they do not have a choice to receive video services with some level of QoE for an extra-cost. In this thesis, recognizing these shortcomings, we investigate the vision of open multi-service inter-networking, including different service-levels and service-level awareness of users, and propose managed video service architectures using software defined networking (SDN) that is a programmable networking paradigm and a key technology for 5G networks. First, we propose centralized and distributed architectures for collaboration between NSP, VSP and users to provide NSP-managed or VSP-managed streaming services over SDN with quality-of-service (QoS) reserved network slices. We show that QoS reservation alone is not sufficient to overcome QoE fluctuations per user and unfairness between heterogeneous video clients, and clients also need to employ TCP receive-window adaptation knowing their fair-share bitrate through collaborative streaming service models that are NSP-managed (centralized) and VSP-managed (centralized or distributed). Second, we propose a video service architecture and a novel resource allocation optimization framework to enable NSPs to offer value-added video services (VAVS) over single operator SDN including multiple service-levels and associated business models. To this effect, we introduce a new batch-optimization framework, where resource (path, bitrate and admission control) allocations for a small group of flows (consisting of new service requests and some existing ones) are performed simultaneously. In order to compute dynamic resource allocations online, we propose a heuristic group-constrained-shortest-path procedure that aims for a fair allocation of resources among a group of requests with the same service level, while maximizing the total NSP revenue. Third, we propose an SDN-enabled distributed open exchange framework for dynamic optimization of end-to-end (e2e) QoS paths over multi-operator networks, which enables individual NSPs to offer inter-operator services with e2e QoS guarantees while allowing each NSP to manage their own network resources. The SDN controllers of all NSPs communicate with each other to advertise a set of QoS-enabled paths across their own network with ask prices, and each NSP dynamically selects the best e2e price-performance path for its customers by bidding on a subset of these advertised paths, and form e2e paths by concatenating them.