PaperlayerView exact revision on arXiv

Source-linked AI summary

QoE Management of Multimedia Streaming Services in Future Networks: A Tutorial and Survey

Alcardo Alex Barakabitze, Nabajeet Barman, Arslan Ahmad, Saman Zadtootaghaj, Lingfen Sun, Maria G. Martini, Luigi Atzori

arXiv:1912.12467v1cs.NIcs.MMeess.SP

TL;DR

Multimedia streaming growth and heterogeneous delivery conditions make QoE management difficult across users, networks, and services. The paper develops a tutorial and comprehensive survey spanning QoE processes, HAS, SDN/NFV, emerging architectures, and newer application domains, then identifies challenges and research directions. Its review highlights SDN/NFV-supported QoE management as an active area with substantial scope for future research.

  • Problem

    QoE management must address heterogeneous users, changing network conditions, diverse services, and difficult optimization decisions about what, where, when, and how often to control.

  • Method

    The paper provides a tutorial and comprehensive survey of QoE management across QoE processes, HAS, SDN/NFV, emerging architectures, and newer multimedia domains.

  • Results

    The review highlights SDN/NFV and related emerging architectures as important areas for QoE management while identifying challenges and directions for future research.

  • Takeaways & Limitations

    QoE management research spans programmable network control, adaptive streaming, emerging delivery architectures, and applications including immersive media and gaming.

Abstract

from arXiv · show

We provide in this paper a tutorial and a comprehensive survey of QoE management solutions in current and future networks. We start with a high level description of QoE management for multimedia services, which integrates QoE modelling, monitoring, and optimization. This followed by a discussion of HTTP Adaptive Streaming (HAS) solutions as the dominant technique for streaming videos over the best-effort Internet. We then summarize the key elements in SDN/NFV along with an overview of ongoing research projects, standardization activities and use cases related to SDN, NFV, and other emerging applications. We provide a survey of the state-of-the-art of QoE management techniques categorized into three different groups: a) QoE-aware/driven strategies using SDN and/or NFV; b) QoE-aware/driven approaches for adaptive streaming over emerging architectures such as multi-access edge computing, cloud/fog computing, and information-centric networking; and c) extended QoE management approaches in new domains such as immersive augmented and virtual reality, mulsemedia and video gaming applications. Based on the review, we present a list of identified future QoE management challenges regarding emerging multimedia applications, network management and orchestration, network slicing and collaborative service management in softwarized networks. Finally, we provide a discussion on future research directions with a focus on emerging research areas in QoE management, such as QoE-oriented business models, QoE-based big data strategies, and scalability issues in QoE optimization.

I. INTRODUCTION

QoE management is increasingly important yet difficult as multimedia traffic, user expectations, network heterogeneity, and service diversity expand. The paper motivates SDN/NFV-based, programmable and adaptable management approaches while identifying gaps in prior surveys.

  • Multimedia growth and QoE challenges: 82% of all IP traffic was forecast to be video by 2022, increasing the importance of delivering high video quality to end users.The growth of video streaming services creates new revenue potential for ISPs, mobile operators, and OTT providers.
  • Multimedia growth and QoE challenges: QoE is challenging because client devices, request patterns, media contents, transmission conditions, and CDN performance vary substantially.These factors create significant spatial and temporal variation in service delivery conditions.
  • Multimedia growth and QoE challenges: QoE management challenges span variable network resources, emerging services and contexts, heterogeneous networks, and rapidly growing multimedia use across diverse devices.Congested locations such as trains, stadiums, and shopping malls require continuous adaptation of resource allocation to different clients.
  • Emerging network paradigms: SDN, NFV, MEC, and cloud/fog computing are emerging responses to rising QoE demands and the need to upgrade network systems for new use cases.The paper situates these paradigms within transformations toward future networks supporting services such as ultra-high-definition video.
  • Emerging network paradigms: SDN and NFV can make future network control programmable, centrally manageable, adaptable, and cost-effective while supporting QoS, QoE, and ELA requirements.The paper links these technologies to automated network management and video-streaming applications.
  • Gaps in prior work: Prior surveys leave gaps in standardization coverage, research-project and challenge descriptions, and comprehensive treatment of architectures, implementations, and deployment strategies.The paper positions its review as addressing these limitations across current and future networks.

B. Scope and Contributions

The paper combines a tutorial on QoE management and multimedia streaming with a broad survey of QoE-aware techniques across softwarized and emerging architectures. It also covers newer multimedia domains and identifies future research challenges and directions.

  • Scope: The paper’s primary objective is to provide a comprehensive state-of-the-art on QoE management for multimedia streaming services in future networks.Its scope encompasses both tutorial material and a systematic survey of management approaches.
  • Tutorial: The tutorial covers QoE modeling, monitoring and measurement, optimization and control, HAS, server and network assistance, and network softwarization and virtualization.HAS is presented as the dominant technique for streaming video over the Internet.
  • Survey: The survey reviews QoE management using SDN/NFV, including server and network assistance, OTT–ISP collaboration, routing, applications, and transport protocols.It also examines adaptive streaming over MEC, cloud/fog computing, and ICN architectures.
  • New domains: The paper extends QoE management to immersive AR/VR, mulsemedia, light-field, and video-gaming applications.These domains receive limited coverage in most earlier surveys, with exceptions for AR/VR and multisensory applications.
  • Future challenges and directions: The paper identifies future QoE management challenges and research directions concerning emerging applications, network management, orchestration, network slicing, and collaborative service management.It presents these directions in the context of future softwarized networks.

C. QoE Monitoring and Measurement

QoE monitoring and measurement collect information from users, applications, networks, and bitstreams to identify degradation and support continuous optimization. The surveyed approaches use monitoring, analysis, and control mechanisms across delivery-chain components.

  • QoE management requires monitoring terminal capabilities, application information, network information, and bitstream-related data.These measurements help identify the root causes of degraded or unsatisfactory QoE.
  • QoE management continuously optimizes and controls mechanisms from content generation through consumption along the service delivery chain.A stated goal is maximizing end-user QoE through efficient allocation of available resources.
  • A client-side YouTube monitoring application quantifies application status, predicts end-user QoE, and triggers resource-management tools during degradation.YoMoApp passively monitors KPIs on Android smartphones, while a network advisor initiates corrective actions.
  • An autonomic access-network architecture separates monitoring, knowledge-based analysis, and action enforcement for video QoE optimization.Its knowledge plane uses an analytic reasoner and a feedforward neural-network reasoner to determine QoE actions.
  • QoE optimization studies address rate adaptation, admission control, resource allocation, energy consumption, buffering, and cooperative transmission.Reported objectives include fairly maximizing clients’ QoE, reducing energy use and freezes, and balancing startup, rebuffering, and starvation effects.
  • A mobile QoE-management framework combines QoE monitoring, QoE management, and QoE control to estimate per-flow quality and coordinate network decisions.Multiple QoE controllers can improve reliability and availability and help avoid service interruptions.

E. Summary

The paper surveys QoE management for multimedia streaming, emphasizing QoE modelling, monitoring, and optimization, HTTP adaptive streaming, and emerging network architectures. It also reviews limitations of decentralized HAS when multiple clients compete for shared resources.

  • QoE-aware multimedia service management integrates QoE modelling and assessment, monitoring and measurement, and optimization and control.The modelling component creates predictive mathematical models from data related to key quality indicators.
  • The survey discusses HTTP adaptive streaming, including client-side and server-side optimization, Dynamic Adaptive Streaming over HTTP, and Server and Network Assisted DASH.It also considers multimedia delivery-chain and service-management issues involving OTT providers, ISPs, CDNs, transit providers, and IXPs.
  • HAS is widely used because it supports reliable transmission, cache reuse, and firewall traversal.Commercial examples include Microsoft Smooth Streaming, Adobe Dynamic Streaming, Netflix, and Apple HTTP Live Streaming.
  • DASH adapts video quality to network conditions by requesting segments at suitable bitrates from multiple encoded representations.Throughput-, buffer-, or hybrid-based adaptation can reduce playback stalls and use available bandwidth more effectively.
  • HAS can suffer bitrate-switching instability, network-resource under-utilization, and QoE unfairness when multiple clients compete for shared resources.These concerns are aggravated in heterogeneous environments and motivate discussion of MPEG-SAND.

B. Server And Network assisted DASH (SAND)

SAND extends MPEG-DASH with messages exchanged among DASH clients and network elements to expose real-time delivery conditions. The paper connects this architecture with SDN/NFV for centralized, scalable QoE-aware management.

  • SAND extends MPEG-DASH by exchanging messages between DASH clients and network elements or among network elements.The messages expose operational characteristics of networks, servers, proxies, caches, CDNs, and clients.
  • The SAND architecture contains DASH clients, DASH-Aware Network Elements, and regular network elements that remain DASH-unaware.DANEs communicate with clients and possess limited awareness of DASH-formatted objects such as MPDs and segments.
  • SAND defines Parameters Enhancing Reception messages from DANEs to clients and additional messages for network-assisted delivery coordination.DANEs can prioritize, parse, or modify DASH objects when they recognize their format.
  • Client status messages let DANEs monitor QoE, communicate available bandwidth, and report cached segments for device-aware requests.This information supports video data-rate optimization and cache-aware adaptation.
  • SAND requires collaboration between service/application and network providers and modifications to network elements for message exchange.SDN is proposed as a centralized control mechanism for practical and scalable implementation, while SDN/NFV-based QoE management is presented as a vital direction.

1) Client Request Redirection and Optimal Route:

Client redirection and content distribution affect multimedia delivery quality, but DNS-based assignment lacks visibility into ISP congestion. SDN and NFV are presented as mechanisms for more programmable routing, resource management, and movable virtual surrogates.

  • 1) Client Request Redirection and Optimal Route:: Client request redirection assigns users to nearby CDNs or surrogate servers to reduce content retrieval time.The mechanism can use DNS unicast or a combination of DNS and anycast.
  • 1) Client Request Redirection and Optimal Route:: DNS-based redirection relies on link-layer information and cannot react to congestion inside the ISP network.Without information sharing, OTT providers may deliver inadequate quality while ISPs transport data over non-optimal paths.
  • 2) Content Distribution/Replication and Cache Miss Handling:: Content distribution and replication can use hierarchical or flat CDN and surrogate-server organizations.Hierarchical replication uses HTTP redirects and can embed content-location information in the overlay to avoid cache misses.
  • 2) Content Distribution/Replication and Cache Miss Handling:: NFV can deploy and move Virtual Surrogate Servers as VNFs when flash crowds or network congestion threaten quality.This approach requires OTT–ISP collaboration, information exchange, and ISP access to VNF infrastructure.
  • 1) Client Request Redirection and Optimal Route:: SDN separates the control and forwarding planes and enables programmable, fine-grained, network-wide orchestration of applications and services.The paper notes that centralized controllers suit small or single-domain networks but can face scalability and reliability problems.

B. Network Function Virtualization (NFV)

NFV decouples network functions from dedicated hardware and enables their dynamic deployment on commodity infrastructure. The ETSI MANO framework organizes infrastructure, orchestration, management, and virtualized services through connected functional blocks.

  • NFV concept: NFV decouples physical network equipment from network functions and enables virtual network functions to run on commodity hardware.Network functions can be rapidly deployed and dynamically allocated, with resources allocated efficiently for service function chaining.
  • NFV concept: Service Function Chaining is an ordered list of service functions applied to classified packets and flows.
  • ETSI MANO framework: The ETSI MANO framework groups NFV Infrastructure, NFV Orchestrator, Network Management System, and VNFs and Services as connected functional blocks.The blocks communicate through defined reference points.
  • ETSI MANO framework: NFVI combines physical computing, storage, and network resources with virtual abstractions created through a virtualization layer.The Network Management System supports operation of both VNFs and physical resources.
  • ETSI MANO framework: The NFVO orchestrates infrastructure resources and network-service instantiation, while the VNFM manages VNF-instance lifecycles and coordination.
  • NFV use cases: Academic and industry NFV use cases include QoE-based multipath routing and virtualized implementations for broadband access, inspection, RAN, and CPE functions.

3) NFV Standardization Activities:

NFV standardization addresses virtualized architectures, service-function chaining, provider collaboration, and integration with SDN. The surveyed QoE approaches use network and client information to guide bitrate, bandwidth, caching, and adaptive-streaming decisions.

  • NFV standardization activities: NFV standardization spans virtualized architectures, service-function chaining, provider collaboration, and integration of SDN services with NFV resources.Relevant activities involve the IRTF NFVRG, IETF SFC Working Group, ATIS NFV Forum, Broadband Forum, and ETSI-related work.
  • Network-assisted adaptive streaming: SAND standardizes messages between DASH clients and network elements to improve bandwidth utilization, streaming experience, and bandwidth fairness.Bandwidth reservation and bitrate guidance are highlighted as network-assisted strategies.
  • Network-assisted adaptive streaming: The SDN controller can compute fair DASH bitrates using SSIM, assign bandwidth slices among clients with similar optimal bitrates, and use nested control loops.
  • Network-assisted adaptive streaming: SABR uses per-link bandwidth and cache occupancy to guide clients through a REST API and dynamic SDN routing toward suitable caches.The surveyed work reports improved client video bitrate, reduced server load ratio, and higher network utilization.
  • Network-assisted adaptive streaming: QoE-SDN APP combines user movement and network conditions with segment selection, encoding-rate, and caching strategies intended to reduce stalling events.
  • Network-assisted adaptive streaming: The BMS allocates, slices, and monitors bandwidth for heterogeneous HAS clients, with a Viewer QoE Optimizer computing joint presentation and allocation decisions.Its optimization uses network utility maximization together with fastMPC, SOL, and online decomposition methods.

B. QoE-Fairness and Personalized QoE-centric Control in SDN

SDN-based QoE-fairness and personalized-control approaches centrally monitor streaming sessions and allocate or route resources according to client conditions and QoE objectives. The surveyed mechanisms address fairness, path selection, multicast adaptation, and periodic QoS-to-QoE control.

  • QoE-fairness and personalized control: SDN controllers can monitor HAS players, device capabilities, requested content, subscription plans, QoE, and buffer levels to support personalized control.
  • QoE-fairness and personalized control: cDVD exposes a client-level network API and uses bandwidth enforcement to equalize QoE across encrypted SDN-assisted sessions while maintaining low re-buffering ratios.
  • QoE-fairness and personalized control: QFF uses OpenFlow for vendor-agnostic resource management and aims to maximize the number of users meeting a target QoE-fairness level in heterogeneous environments.
  • QoE-fairness and personalized control: SDNDASH addresses quality instability, unfair bandwidth sharing, and resource under-utilization among competing DASH clients sharing a bottleneck link.
  • QoE-fairness and personalized control: 75% users can be supported at the same QoE level by an SDN-based multi-client bandwidth-management architecture.
  • QoE-centric routing: Q-POINT calculates best paths for service flows using SIP negotiation, QoS-QoE mappings, and an optimization function.
  • QoE-centric routing: SDM2Cast adapts multicast layers and customizes multicast paths according to network state, while other approaches select routes or servers using QoE information.
  • QoE-centric routing: SQAPE periodically measures packet loss, delay, and link bandwidth to predict streaming QoE and performance.The monitoring interval described is 60 seconds.

D. QoE-Aware SDN/NFV-based Mechanisms using MPTCP and Segment Routing

MPTCP and Segment Routing are surveyed as complementary mechanisms for QoE-aware traffic engineering in SDN/NFV networks. MPTCP distributes streams across subflows, while Segment Routing expresses paths as ordered segment sequences that reduce switch routing-state requirements.

  • MPTCP: MPTCP creates multiple subflows over disjoint paths, supporting traffic engineering and QoE-aware video delivery in SDN/NFV networks.
  • Segment Routing: TCAM is expensive and limited, with practical switch support reported at approximately 2k–20k rules.
  • Segment Routing: Segment Routing simplifies the control plane by expressing MPTCP subflow paths as ordered sequences of segments between ingress and egress nodes.Segments may represent nodes, links, or packet filters, and their identifiers may be global or local.
  • MPTCP and Segment Routing: A combined MPTCP-SR strategy is presented as a way to reduce TCAM cost while offering flexibility, reliability, scalability, and improved end-user QoE.
  • MPTCP: SDN-based MPTCP implementations are reported to improve throughput, support QoS/QoE-guaranteed DASH services, and address out-of-order packets through path control.
  • Collaborative service QoE management: Collaborative QoE management assigns monitoring to OTTPs and network-wide QoS monitoring and policy implementation to ISPs using SDN/NFV.ISP actions include virtual surrogate-server placement, traffic rerouting, link aggregation, and traffic prioritization.

F. QoE-aware/driven Approaches using Full Adoption of SDN and NFV

Full SDN/NFV adoption supports flexible QoE monitoring, traffic control, and cross-layer coordination in future networks. The survey compares these approaches and identifies continuing research scope for QoE management and orchestration.

  • Integrating SDN and NFV enables flexible placement of QoE measurement points and control of multimedia traffic flows.
  • Software-defined NFV architectures can steer traffic actively and jointly optimize network functions for QoE management.
  • Application-layer QoE requirements can be specified and controlled at the network layer through SDN controllers.
  • Existing SDN/NFV use cases include dynamic bandwidth allocation, video QoE adaptation, and scalable monitoring and discovery for 5G systems.
  • The survey compares approaches by client QoE-fairness, deployment complexity, and the metrics most affecting QoE, while identifying broad scope for future research.

VI. QOE- MANAGEMENT APPROACHES USING EMERGING ARCHITECTURES

Emerging architectures extend QoE-aware streaming toward the network edge, distributed fog/cloud environments, and coordinated service orchestration. The reviewed approaches use contextual information, caching, prefetching, and resource coordination to support delivery quality and fairness.

  • MEC, fog/cloud computing, and ICN are presented as emerging architectures that can benefit from SDN and NFV for interactive and immersive video services.
  • MEC offers high bandwidth and low latency, with QoE-aware architectures and monitoring probes adapting service delivery to RAN and resource conditions.
  • Mobile edge virtualization combines context-aware adaptive prefetching, traffic prediction, and video quality adaptation for QoE-assured 4K video delivery.
  • QoE-aware edge caching and bandwidth provisioning aim to reduce content delivery latency, improve resource utilization, and increase fairness across users.
  • QCSS coordinates client-side contextual measurements with MEC-side throughput prediction and QoE-aware bitrate adaptation.
  • Fog computing distributes resources across networks while SDFog orchestrates flows using service discovery, flow creation, and network-parameter calculation.
  • Fog services can provide low delay and QoE support without significant network overhead, while orchestration evaluations consider throughput, latency, energy efficiency, and fairness.

C. QoE-driven/aware Management Approaches using Information-Centric Networking (ICN)

ICN shifts networking toward content-centric naming and enables caching, controller-assisted prefetching, and adaptive streaming across edge caches. These mechanisms target video quality, delivery performance, and mobile-network load under varying conditions.

  • ICN decouples content from network location by naming information objects directly with routable identifiers.
  • ICN controllers can locate content, manage caches, monitor congestion, and prefetch DASH segments to edge caches using network and session information.
  • DASC overcomes network bottlenecks by progressively converging toward the best available video quality, while ICN complicates client rate adaptation because nodes are hidden.
  • ICN caching exposes video interests and segment relationships to the network, supporting video- and network-aware prefetching during off-peak congestion periods.
  • 20% higher delivered video quality is reported for video- and network-aware prefetching compared with a DASH system without prefetching.
  • ICN approaches can combine fine-grained caching with multipath transmission, potentially providing higher throughput than standard TCP/IP.

A. QoE in Immersive AR/VR and Mulsemedia Applications

QoE management is extended to immersive AR/VR, mulsemedia, and gaming applications whose media formats, sensory effects, and interaction demands differ from traditional streaming. The surveyed work addresses bandwidth, latency, perceptual quality, sensory experience, and flow optimization across these domains.

  • AR/VR: Growing 360° video and VR use increases streaming demands, requiring attention to bandwidth limitations, end-to-end latency, and QoS/QoE.
  • AR/VR: Viewport-dependent 360° video solutions reduce the bandwidth required for streaming by adapting delivery to the user’s dynamically selected view.
  • AR/VR: AR QoE research includes subjective assessment, psychophysics, usability, human factors, ergonomics, ethnography, collective experience, and real-time bitrate adaptation.
  • Mulsemedia: Mulsemedia adds sensory dimensions such as olfactory, haptic, thermal, light, wind, and vibration effects to multimedia experiences.
  • Mulsemedia: 77% higher overall user enjoyment is reported when mulsemedia is used, while nonlinear models address QoE prediction for enriched audiovisual sequences.
  • Gaming: Cloud gaming streams cloud-rendered gameplay to clients, and SDN-based flow distribution can use game type, server load, and path delay to minimize data-center delay.
  • Gaming: Gaming QoE standardization includes recommendations covering factors affecting QoE and subjective evaluation methods for gaming activities.

C. QoE in Light Field Applications

The supplied passages situate immersive multimedia applications within QoE management challenges and future softwarized-network research, while highlighting demanding requirements for light field video.

  • Light Field Requirements: Light field video uses very high resolutions, typically 50–80 megapixels, creating high bandwidth demands for streaming and display.The passage links these requirements to the need for efficient compression and higher network bandwidth.
  • QoE Measurement: QoE measurement for light field applications considers 3D displays and binocular receptive-field properties aligned with human visual perception.
  • Research Challenges: Future QoE management must address emerging multimedia applications alongside network management, orchestration, and collaborative service management.
  • Collaborative Management: Collaborative QoE management may require longer-duration video prediction models and algorithms that jointly consider QoE fairness and business models.

C. QoE-oriented Business Models in Future Softwarized Network

The paper discusses QoE-oriented business models for future softwarized networks, emphasizing Experience Level Agreements and unresolved questions around user churn, pricing, and dynamic management.

  • Business Contracts: Current QoS-based SLAs are described as insufficient for QoE-related contracts between service providers and end-users.
  • Business Contracts: Experience Level Agreements extend traditional SLAs by expressing service performance in terms of users’ quality of experience.
  • Business Contracts: ELAs are proposed as a basis for minimum QoE guarantees and QoE-differentiated service provisioning in future networks.
  • Open Questions: The relationship between QoE and user churn remains under investigation, including subjective validation of churn models and effects of pricing on willingness to pay.
  • Open Questions: Future softwarized networks require research on dynamic QoE models, intelligent slice scheduling, slice isolation, mobility, security, and performance guarantees.

G. Multimedia Communications in Internet of Things (IoTs)

The paper identifies multimedia QoE management challenges in IoT and future softwarized networks, then surveys existing approaches and outlines research directions for emerging networked services.

  • IoT multimedia applications face scalability, mobility, security, privacy, QoE resource-management, and multimedia network-management challenges.
  • The survey covers QoE management solutions using SDN and NFV in current and future 5G networks.
  • It examines QoE-aware adaptive streaming over MEC, fog/cloud computing, and information-centric networking, alongside SDN/NFV design, implementations, projects, standards, and use cases.
  • Future research directions include QoE-oriented network sharing and slicing, business models, big-data strategies, scalability, resilience, optimization, security, privacy, and trust models.
Loading 1912.12467v1…