QoE-enabled big video streaming for large-scale heterogeneous clients and networks in smart cities

SPECIAL SECTION ON SMART CITIES

Received November 21, 2015, accepted November 28, 2015, date of publication December 8, 2015,date of current version February 29, 2016.

Digital Object Identifier 10.1109/ACCESS.2015.2506648

QoE-Enabled Big Video Streaming for Large-ScaleHeterogeneous Clients and Networksin Smart CitiesBO-WEI CHEN1, (Member, IEEE), WEN JI2, (Member, IEEE),FENG JIANG3, AND SEUNGMIN RHO41Department of Electrical Engineering, National Cheng Kung University, Tainan 701, Taiwan2Beijing Key Laboratory of Mobile Computing and Pervasive Device, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100864, China3School of Computer Science, Harbin Institute of Technology, Harbin 150001, China4Department of Multimedia, Sungkyul University, Anyang 430-742, Korea

Corresponding author: S. Rho ([email protected])

This work was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by theMinistry of Education (2013R1A1A2061978).

ABSTRACT The rapid growth of the next-generation communication and networks is bringing videoservices into more pervasive environments. More and more users access and interact with video contentusing different devices, such as smart televisions, personal computers, tablets, smartphones, and wearableequipments. Providing heterogeneous Quality of Experience (QoE) that supports a wide variety of multime-dia devices is critical to video broadcasting over the next-generation wireless network. This paper reviewspractical video broadcasting technologies and examines current requirements ranging from heterogeneousdevices to transmission technologies. Meanwhile, various coding methodologies, including QoE modeling,scalable compression efficiency, and flexible transmission, are also discussed.Moreover, this paper presents atypical paradigm as an example for video broadcasting with large-scale heterogeneity support, which enablesQoE mapping, joint coding, flexible forward error coding, and cross-layer transmission, as well as optimaland dynamic adaptation to improve the overall receiving quality of heterogeneous devices. Finally, a briefsummary of the key ideas and a discussion of interesting open areas are summarized at the end of this paperalong with a future recommendation.

INDEX TERMS Quality of service, video coding, broadcast technology, communications technology.

I. INTRODUCTIONWith the integration of telecommunications, television net-works, and internet, future networks gradually become anintegrated medium with high broadbands and large-scaletraffic. Providing users with satisfactory Quality of Expe-rience (QoE) that supports the mass heterogeneous multi-media devices is essential for video broadcasting over thenext-generation networks. High-quality video services at anydevices, anytime and anywhere are becoming an inevitabletendency. This significantly challenges the innovation anddevelopment in both networks and video technologies.

Recent years have seen a flourishing change in video traf-fic. Internet video traffic has already exceeded more than halfof any other traffic of consumer networks in 2013. Mobilevideo streaming will grow at a compound annual growth rateof 69.0% between 2013 and 2018, the highest growth rate ofany mobile applications, as reported in [1]. According to the

statistical report from the China Internet Network Informa-tion Center (CNNIC) [2], the number of mobile-phone usershas reached 500 million with an annual growth rate of 19.1%.The most noticeable growth also occurs in mobile-phoneusers. The proportion of consumers that access the internetthrough mobile phones increases from 74.5% in 2012 to81.0% in 2013, much higher than that through the othermobile devices.

Currently, with more global 4G deployments, higher band-width, and more intelligent services, mobile applications pro-viding various videos attract more users. For example, thenumber of mobile video users in China exceeds 247 millionand increases to 83.8% in 2013 compared with that in 2012.Mobile video streaming has become the major and primarycontributor to the growth of global mobile applications. Sucha trend significantly challenges wireless and internet serviceproviders as well as content providers. Therefore, how to

VOLUME 4, 20162169-3536 2015 IEEE. Translations and content mining are permitted for academic research only.

Personal use is also permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

97

www.redpel.com +917620593389


B.-W. Chen et al.: QoE-Enabled Big Video Streaming for Large-Scale Heterogeneous Clients and Networks

FIGURE 1. Architecture of classical video broadcasting to heterogeneous devices.

ensure the high quality of user experience is of prior concerns.Despite decades of research on video coding, communica-tions, and networking, there still remain challenging tasksto provide the mass interactive applications and satisfactoryQoE for mobile users with diverse devices.

Providing good QoE for the users with heterogeneousmultimedia devices is essential for video broadcasting inthe next-generation networks. Much effort has been madeto enhance transmission capacity to satisfy the exponentialincrease of traffic-driven and highly-diverse devices. Onecritical requirement for future ubiquitous environments isthe ability to handle the heterogeneity, such as various userpreferences, display characteristics, device capabilities, andemerging interactive modes.

Therefore, researchers are intensively studying what thecore heterogeneous factor in video broadcasting is and hownew techniques should be designed for better QoE. Thisarticle makes an investigation into the following questions.

• What heterogeneous factors need to be considered invideo broadcasting systems?

• How does a new intelligent interaction affect broadcast-ing architectures?

• How do the researchers model QoE metrics to meet therequirements of mass users?

• How do the service providers offer heterogeneous QoEvideos from the source side?

• How do the service providers efficiently broadcast videostreams in consideration for heterogeneous QoE thatsupports heterogeneous circumstances?

• What are the future challenges?

To this end, this article surveys the aforementionedpossible challenges for heterogeneous video broadcastingunder heterogeneous circumstances. Feasible solutions tothe above questions are offered in this study by proposing

heterogeneous QoE coding and transmission schemesfrom the perspectives of architectures, strategies, andmethodologies.

The remainder of this article is organized as follows.Section II firstly gives the overview of the whole systemarchitecture. Subsequently, Sections III–VI present the abovechallenges, followed by the foundation and a series of techni-cally impressive solutions. Then, numerical results and con-clusions are provided in Section VII. Conclusions are finallydrawn in Sections VIII and IX, along with recommendationsfor future research.

II. OVERVIEWIn current video broadcasting systems, heterogeneity existsalmost everywhere, for example, user terminals, user experi-ence, network access, network types, video contents, and datasupport. Fig. 1 illustrates the key components for classicalvideo broadcasting. Different aspects, including devices,networks, servers, video content, and clients, are discussed asfollows. On the device side, each video stream is broadcastto various devices based on different characteristics rang-ing from terminal display sizes, bandwidth requirements,reception capabilities, channel conditions, battery capacities,to energy consumptions. Moreover, devices on the networkside might access the same video service from different net-works, such as DVB, WIFI, GSM (i.e., 2G systems) [3], [4],WCDMA/CDMA2000/TD-SCDMA (i.e., 3G systems),LTE/WIMAX (i.e., 4G systems) [5], or future 5G net-works [6]. Regarding the server side, videos might be deliv-ered via heterogeneous clouds and storages. Even when itcomes to video data, heterogeneity still exists in video con-tent. For example, video data in different scalable domainsusually have different rate distortions. The last is the end-user side, or the client side, where users might have various

98 VOLUME 4, 2016




preferences and interactions, so that they desire differentQoE even for the same video service.

Conventionally, transmissions, coding, and QoE measure-ments are seldom designed in consideration of heteroge-neous characteristics, and most of the above challenges arepractically ignored during video broadcasting. To provideflexible and reliable QoE for broadcasting, it is necessaryto deeply exploit the heterogeneities in joint coding, trans-missions, QoE measurements, and adaptivity. The feasiblesolution for heterogeneous QoE provision in broadcastingnetworks requires an integrated design effort. The nextsubsection firstly gives a survey on the above key componentsof video broadcasting systems, including QoE metrics, videocoding, and transmissions. Subsequently, a system with het-erogeneous support is presented with the evaluation of systemperformance in terms of broadcasting efficiency.

III. HETEROGENEITY IN QoE METRICSThe International Telecommunication Union (ITU) leads thestudy group on QoS/QoE for IPTV services. According tothe definition by the ITU [7], the QoE is ‘‘The overall accept-ability of an application or service, as perceived subjectivelyby the end-user.’’ The QoE includes two major aspects –Quality of Service (QoS) and human components. The for-mer involves services, transmissions, and applications, whichare measured by objective methods. The latter depend onemotions, preferences, experience, and so on. Unlike QoS,human components are measured in subjective ways [8].More details were discussed in [9], where the authors summa-rized the QoE assessment and the corresponding standardiza-tion activities of the ITU in detail. At present, QoE has beengradually developed and has a wide range of applications inacademia and industry. QoE has outformed IPTV initiallydesinged by the ITU.

So far, current multimedia services in communicationand network systems have achieved a certain level becauseuser-centric concepts have deeply affected the designs forthe whole process of network deployments, service acti-vation, consumptions, management, and evolution severalyears before. As the main metric of user-centric analysis,QoE however has a complex layer-based archi-tecture [11], [85]. In the Universal Mobile Telecommunica-tion System (UMTS) [12], many techniques in the physicallayer already supported the concept of QoE, such as down-link/uplink control and relevant radio/power resource man-agements. However, in the next-generation network, the focuslies in how to guarantee QoE across layers and to ensure QoEbetween applications and transmission layers [13]. Since boththe UMTS and the next-generation systems were originallybuilt on the basis of QoS, the monitoring framework betweenQoS and QoE becomes a feasible route.

The application layer concentrates on context-awareend-to-end transmissions through quality control parame-terizations [14]. Since the uniqueness of QoE lies in thesubjective perception of users and the ability of a systemto fulfill users’ expectations, QoE in the application layer

introduces more human components than current architec-tures [14]. Unfortunately, it is still difficult to simultaneouslyobtain the complete information of QoE parameters from allusers. Accordingly, Zhou et al. [15] proposed using dynamicresource allocation to optimize the total subjective qualityof all the users with or without prior QoE information andtests. As current applications, communications, and networksystems are gradually migrating to cloud-based architectures,an universal model is therefore required to manage cloud-computing ecosystems. Fortunately, QoE-based managementprovides an interactive, rapid, and convenient solution tocloud computing [10]. Through understanding, monitoring,and estimatingQoE, cloud-service providers can easily adapt,control, and manage all the interactions between users andservers. For instance, Laghari et al. [16] examined recentQoE models and developed a practical taxonomy of therelevant variables and interactions based on a communica-tion ecosystem. Fig. 2 summarizes the major factors in aQoE model. As illustrated in this figure, human-relatedcomponents are the key difference between the future andconventional QoS architectures. As a user can access anytype of services through any device, like computers, laptops,smartphones, tablets, wearable gadgets, and televisions,personal experience is actually based on the features of termi-nal equipments, including types, displays, batteries, and oper-ational modes. How to formalize the actual preferences of auser involves further research. This is because the personalperceptual quality significantly depends on – Personal infor-mation (e.g., ages, genders, professionals, and educations),visual quality, video formats (e.g., High Definitions (HDs)and Standard Definitions (SDs)), interactive modes(e.g., videos on demand), urgent degree (e.g., on-displayproportions in an online video), content (e.g., news ormovies), costs of services, the presence of advertisements,and environments (e.g., in a bus or at an office). In addition,factors involving user subjective perceptions, like emotionsand personal habits, dominate satisfaction of a service.Although, through experience parameter estimation andextraction under a specific scenario, QoE control and assur-ance can be rapidly deployed. Nevertheless, there is still noconsensus about an effective QoE model.

For multiuser video communication systems (e.g., broad-casting), research on the relation between heterogeneous userdevices and video streams, which supports user-centric QoE,has recently become a research hotspot [17].

Video sources can be roughly classified into the followingfour categories – 1) Spatial domain that dominates displaysizes. 2) Temporal domain, which determines the fluencyof playback. 3) Quality domain that influences visual quality.4) Error domain, which decides the reliability during broad-casts. Since classical theories on image fidelity could notsimultaneously measure the above-mentioned four domains,hence, current research [19]–[23] developed a hybrid-domainmethod across controls by considering multidimensionalfeatures of videos. For instance, Ou et al. [19] mod-eled the impact of temporal and quality domains based

VOLUME 4, 2016 99




FIGURE 2. QoE architecture in video communication systems with heterogeneous support.

on frame rates and quantization step-sizes by fitting framerates, fidelity of Signal-to-Noise Ratios (SNRs), and Mean-Opinion Scores (MOSs) into a function. A subjective qualitytest was conducted for evaluating the relation between the rateand the perceptual quality of a scalable video with temporaland quality scalability. Chan et al. [18] built multidistortionalmeasures between the fidelity of Peak SNRs (PSNRs) andthe temporal downsampling, as well as between the PSNRfidelity and the spatial downsampling. They subsequentlyapplied the rate-distortion optimized scheduling to analyzea diverse range of target devices. These methods [18], [19]basically followed the idea of PSNR fidelity and were furthermodified to support the subjective cross-measure with frame-rate and resolution scalability.

Recently, more and more approaches addressed thespatio-temporal quality problem by using user experiencemaximization. Such studies include [19]–[22]. In [20],Wang et al. created a generalized and classifier-based pre-diction framework to provide multidimensional adaptiveoperations and different SNR-temporal resolutions by usingthe human vision system. Similarly, the authors [21], [22]modeled the spatio-temporal utility through homogeneousand heterogeneous QoE decomposition. Rather than focus-ing on individual domains, the hybrid multiple distortion

measure [19] has become a tendency due to the effectivemultidimensional feature of QoE.

IV. SCALABLE SUPPORT FOR VIDEO CODINGBroadcasting videos to multiple heterogeneous devices usu-ally involves two major techniques – Coding and transmis-sions. Thus, scalable control and its performance are criticalfor broadcasting. From the view of video coding, today’svideo coding paradigm typically uses spatial and temporalfeatures as well as quality redundancies when serving adiverse range of display resolutions and transmission chan-nels. Cumulative video coding and non-accumulative codingare typical examples.

The former, cumulative coding, classifies video sourcesinto one base layer and multiple enhancement layers. Thebase layer can be independently decoded, whereas theenhancement layer can be successfully decoded only whenthe base layer and the anterior enhancement layers are recov-ered. One of the most famous cumulative coding methods isscalable video coding (SVC) [23], which has pioneered theresearch trend in academia and industry for years. Since SVCcan dependently encode video sources based on video subsetsfrom generation sides to receiver sides, scalability can bedirectly achieved based on different requirements, like quality

100 VOLUME 4, 2016




(fidelity), display sizes (spatial domain), and frame rates(temporal domain). Multiresolutional source coding [24] pre-sented another cumulative coding method, where embeddeddata descriptions were given during encoding in the man-ner of progressive and successive refinement. By doing so,SVC has brought the video processing area into a mile-stone and subsequently enabled encodes/decoders to processa variety of rates and resolutions. Likewise, wavelet videocoding [25] was a particularly useful technique for spatio-temporal scalability with low complexity. It has becomea popular algorithm in modern multiresolutional videocompression.

Unlike multiresolutional or layered source coding, asmentioned in accumulative coding, there is no hierarchicaldescription in non-accumulative coding. Multiple descriptioncoding (MDC) [26] was a typical instance and can be used forthe heterogeneity issue. This is because MDC can decom-pose a video source into multiple descriptions and subse-quently convert this video into several subsets based on thedescriptions.

Since MDC is performed in the encoded streams, userdevices benefit from path diversity over different net-works when multiple paths are available [27]. For exam-ple, if descriptions are not successfully received dueto unknown errors, packet loss, transmission delay, orjitter, the decoder can still reconstruct the original videofrom the received descriptions. Such a property providesflexible robustness for multiple heterogeneous communica-tion systems against noise [28]. Similar studies like [29]devised a system that could assign a multiple descriptionvideo by constructing multiple multicast trees. Despite therobustness of heterogeneous communications, MDC is stillsusceptible to the problem of coding efficiency, for MDCrecovers a certain video quality from every description. How-ever, since the probability of losing every description atthe same time is quite low, MDC still demonstrates satis-factory reliability and robustness in practical transmissionsystems.

In summary, based on the aforementioned reasons, videocoding techniques with scalable support have become widelyused in modern video broadcasting systems, such as mobilebroadcasting/multicasting [30], multiantennal broadcastingsystems [31], opportunistic broadcasting/multicasting [32],and multimedia broadcasting networks [33].

V. VIDEO TRANSMISSION IN HETEROGENEOUSCIRCUMSTANCESWith the time-varying and error-prone characteristics ofchannels, the variety of devices, and the complex of QoE,conventional video broadcasting usually faces unreliableproblems. To overcome such an issue, researchers havedeveloped a new field called reliable video broadcasting,where reliability in wireless networks was realized via trans-mission techniques. Most of related works focused on:1) using opportunistic transmissions to improve the diversitygain in multiuser scenarios; 2) developing cross-layer-based

forward error correction (FEC) to simultaneously provideheterogeneous QoE support; 3) introducing fair streamingschemes to satisfy variable requirements for multiple hetero-geneous users. The following content respectively elaboratesthese three categories.

1) Among the aforementioned three approaches, oppor-tunistic transmissions exploited the variations in channels toachieve high utilization of scarce wireless resources. Suchtransmissions have revealed potentials in cross-layer and real-time applications for wireless broadcasting networks. Relatedworks can be found in [34]–[37]. The authors [34] proposedopportunistic spectrum selection that could allocate avail-able channel resources orderly to users based on their QoEexpectations, with joint support of channel characteristics,QoE measures, and current channel resources. In contrast,another approach ‘‘opportunistic user selection’’ chose userswith maximum channel gains or states [35], [36] to improvebroadcasting efficiency. Opportunistic listening and condi-tional demodulation among video layers [37] could enhancethe system performance. The work by Huang et al. [38]showed that opportunistic-based layered multicasting couldobtain improvement in efficiency through suitable schedul-ing and resource allocation. Consequently, the utilizationof limited resources was accordingly improved by oppor-tunistically transmitting video substreams in considera-tion of those heterogeneous characteristics and multiuserrequirements.

2) To guarantee the acceptable visual experience, QoE cancowork with FEC and error protection strategy in a cross-layer designed framework. Based on such an idea, the sec-ond transmission category emphasizes joint channel coding,resource allocation, and scheduling design under the cross-layer control. FEC concentrates on reliable transmission pro-vision in error-prone wireless circumstances. With adaptivechannel coding, a video stream is capable of adapting itselfto channel dynamics. A common method of adaptive channelcoding, like [39], used joint source and channel coding tominimize the end-to-end distortion. In wireless video broad-casting/multicasting, layered transmissions are viewed as aneffective approach to support heterogeneous receivers withvarying requirements. The work in [40] used a utility func-tion for modeling the reception features in terms of physicalcapacity, actual received bandwidth, and numbers of receivedlayers. Furthermore, this approach also offered layered videotransmissions through multiple video sessions. The workin [41] employed adaptive channel coding and extended thescalable multilayered transmissions to time-varying wirelesschannels. Ji et al. [22] [43] proposed layer-adaptive videosbased on suitable rateless coding protection. In [35] and [44],the authors devised resource allocation and scheduling strate-gies to improve the resource utilization, including wirelessnetwork-flow resources [43] and wireless radio resour-ces [34]. In general, through cross-layer optimization, such assystematic [44], application-centric [45], network-oriented,and wireless-oriented approaches [46], the quality of videostream can be improved. [47] was an instance of cross-layer

VOLUME 4, 2016 101




FIGURE 3. Framework of classical video broadcasting to heterogeneous devices.

optimization by using the utility maximization or distortionminimization approaches.

3) To simultaneously satisfy variable requirements andfairly utilize limited available resources, the third cat-egory focuses on balancing the quality of experienceamong all heterogeneous users. From the view of multi-ple users, a polling-based strategy can directly guaranteethe fairness of all the users. However, the strategy usuallypresents low utilization of available resources because itcannot adapt itself to the variety of user channels well.As the available resource is usually constrained in thecase of multiusers, the server can improve the mini-mal QoE of all the users and then maximizes all theQoE by applying max-min fairness. This is one approach.There is also another method that proportionally allocatesthe resource to the users based on proportional fairness.As video content in different scalable domains has differentrate distortions, and end-users care about the video qual-ity rather than the bandwidth, resource allocation by usingcontent-based fairness is an efficient way [48]. Nonetheless,the bottleneck in heterogeneous video broadcasting still liesin variety, unreliability, and limited resources. This subse-quently makes video broadcasting difficult to provide reliablereal-time video streaming for multiusers.

VI. ARCHITECTUREIn this article, we present a framework of video broad-casting with heterogeneity support as shown in Fig. 3.This solution considers the scenario of multicontent videobroadcasting, where videos are distributed to multiple het-erogeneous devices. The techniques on both of the serverside and the client side are listed in the figure. The serverside includes utility-driven joint source coding/optimization,

QoE mapping, content-aware fair resource allocation,flexible FEC, joint source/channel coding, cross-layeroptimization, layer channel coding, adaptive modula-tion, and diversity modules. Moveover, resource-aware,cooperative-transmission, adaptive-computing, interaction,and QoE-capturing modules are presented on the client side.To support heterogeneous QoE, several dynamic monitor-ing operations are required to simultaneously serve diversedevices with resource constraints under a heterogeneouscircumstance. Such operations correspondingly need device-aware mechanisms from receivers, QoE-aware propertiesfrom users, and circumstance-aware services from broadcast-ing systems.

A. APPLICATION-LAYER CODING AND ADAPTIONQoE provision from the application layer has become themost active and effective method in recent years. From videosource coding, a video stream is encoded into progressivelayers that have unequal importance for serving different usergroups. QoE mapping directly introduces the parameters tovideo source coding such that video streams are generatedaccording to the requirements from users.

Scalable video sources providemore adaptability by apply-ing a variety of schemes, such as scalable stream extraction(e.g., [39], [49]–[52]), layer generation with different prior-ities (e.g., [40], [53]–[55]), and summarization (e.g., [56]),before they are dispatched to the next layers.

In broadcasting systems, it is critical to efficiently utilizeavailable bandwidth resources so as to provide guaranteedquality of service for multiple users. Generally, utility isdefined as the satisfaction level of a user with respect to het-erogeneous characteristics or defined as the utility summationfrom all the users that are serviced. Since the satisfaction

102 VOLUME 4, 2016




TABLE 1. Model comparison.

is parameterized through the QoE mapping module,developing a corresponding metric is conducive to content-aware allocation in fair broadcasting systems.

Since layered video data are sensitive to transmissionfailures, it is acceptable for servers to eliminate the retrans-mission and lower the overhead of the unnecessary recep-tions using FEC. To develop more flexible FEC, most ofrelated works focused on: 1) finding an optimal bit allo-cation between video coding and channel coding, suchas [57] and [58]; 2) designing a new encoder for target sourcerates under a given channel condition, such as [59]; 3) propos-ing novel channel coding to achieve the required robustness,such as low-density parity check (LDPC) [60], Turbo [61],Reed-Solomon (RS) [62], and Fountain [63] codes; 4) creat-ing a joint optimization framework that covers all availableerror control components along with error concealment andtransmission control to improve entire system performance,like [64].

B. PHYSICAL-LAYER CONTROL INCROSS-LAYER OPTIMIZATIONBesides the performance of unequal error protection, effi-ciency improvement of transmissions in multilayers isthe major purpose of the physical layer. The remarkablehigh rates with high reliability innovation include Diver-sity Embedded Space-Time Codes (DE-STCs) [65], [66],which allow servers to provide multiple levels of reliabil-ity to satisfy different QoS requirements. DE-STCs real-ized a form of communications, where the high-rate codeopportunistically took advantage of good channels and madedecisions [67]. Through cross-layer designs, joint controlwith DE-STCs could benefit the diverse rates and reli-able transmissions in a wide range of channel conditions,especially in broadcasting/multicasting [68]–[70]. WhenDE-STCs were combined with opportunistic transmissions,the utilization of the scarce wireless resource was furtherimproved, particularly in variable channel conditions [71].Current video broadcasting services are expected to pro-vide more experience-enriched videos for consumers thanbefore. With the diversity of multiple devices and the vari-able demands from mobile users, video streams are normallyinitiated and delivered through multiple layered substreams.Under the framework of cross-layer control, broadcastingmultiple video streams with multiuser QoE support can berealized through adaptive modulation and joint diversity-embedded high-rate reliability coding from physical layers.

C. INTELLIGENT PROCESSING ON THE DEVICE SIDEWith the increase of pervasive computing, current devices arebecoming more ubiquitous [72]. Generally, video serviceson mobile devices are usually computationally intensiveand power-consuming. Consequently, emerging wirelessapplications have to face a challenge of resource-constrainedvideo networking, such as wireless low-power surveil-lance networks, mobile video phones, etc., because com-putational power, memory, and batteries are limited.However, the high resolutions and complex functionali-ties of encoding require high resources. Thus, the videoencoder should have the capability and the scalability ofprocessing videos based on remaining battery capacity,and power-scalable video encoding is a smart solution forenergy-constrained devices. This scheme performs gametheoretical analysis and models the power consumptionas a game problem. It uses game theory to solve thetradeoff between encoding and power consumptions, andit allows video services to work under variable energeticconstraints while keeping stable performance. Since theuser device is the direct terminal to collect the QoE, thehuman perceptual method offers another approach for resolv-ing power consumption. For example, fine-grained models,such as perceptual macroblock-level power control based onJust-Noticeable Distortion (JND), can adapt to availableenergy resources at macroblock levels in consideration ofhuman perceptions. For those devices with large displays,cooperative communications have been proven to be robustagainst variable data rates [37].

VII. PRACTICAL CHALLENGES ANDCOMPARATIVE RESULTSThis section provides numerical results of the performancewith a focus on the aforementioned concepts in the article.

As described above, parametric QoE models have beenproposed in the past years. Different models with vari-ous parameters were designed for different conditions. Thissection firstly gives a summary of representative modelsand then compares the performance of different models.Table 1 presents five QoE estimation models in typical videobroadcasting scenarios. Different QoSs are highlighted in thefigure.

1) Mean Perceived QoS (MPQoS) model: A qualitativemetric that was designed for CIF- or QCIF-sized videos. Thismodel did not consider factors in transmissions. The modelparameters were derived from video content based on [73].

VOLUME 4, 2016 103




FIGURE 4. Frame-level video quality based on PSNRs, PSPNRs, and SSIMs for sequence ‘‘Crew.’’

2) Model by Ma et al.: In [74], Ma et al. have presented arate-and-quality model based on frame rates and quantizationstep-sizes. Model coefficients were predicted by using videofeatures, but the transmission impairment was not modeledherein.

3) Video Quality Metric (VQM) model: A qualitativeevaluation proposed by National Telecommunications andInformation Administration (NTIA), which consisted of thelinear combinations of features derived from several HumanVisual Systems (HVSs) [76]. The degradation introduced inthe transmission process was not evaluated herein.

4) Method by Liu et al.: Liu et al. [75] have proposedthe video quality model by considering packet losses. Losspositions and loss severity as well as error lengths were fullyinvestigated in their method. The authors used a VQM basedon PSNRs (e.g., VQMp) proposed in [77] for coding artifacts.5) Motion-based video integrity evaluation (MOVIE):

In [78], a video qualitative evaluation was presented for mod-eling not only spatial and temporal domains but also spatio-temporal domains. The analysis was carried out by evaluatingmotion quality along computed motion trajectories.

6) Approach by You et al. : In this method [79], You et al.developed an attention-driven foveated quality model, whichgenerated the perceived representation of a video by integrat-ing visual attentions into the foveation mechanism.

TABLE 2. Evaluation of different QoE models on LIVE database.

For fairness, the experiments on LIVE database [80] werecarried out based on all the aforementioned models exceptfor MPQoS as the authors did not quantitatively present howto derive model coefficients from video content. Table 2 liststhe performance of different QoE models in terms of PearsonCorrelations (PCs), root-mean-square errors (RMSEs), andEpsilon-insensitive RMSEs (E-RMSEs) based upon the 95%

confidence interval of the video subjective scores. As dis-played in the table, all the leading QoE models perform wellin the LIVE database. Interestingly, there is no dominantmodel, which can comprehensively consider coding artifacts,transmission factors, and HVS-related features at the sametime.

The following section provides an overview of the perfor-mance with support of scalable video coding. Three versionsof the same video were manually selected. They were respec-tively designated as ‘‘high-quality level,’’ ‘‘medium-qualitylevel,’’ and ‘‘low-quality level’’ after processed by using theJSVM SVC reference encoder [81].• High-Quality Level: 704× 576 at 30 fps, QP = 32• Medium-Quality Level: 352× 288 at 15 fps, QP = 38• Low-Quality Level: 176× 144 at 10 fps, QP = 44Fig. 4 presents the performance of SVC with three scala-

bility dimensions. Three representative quality metrics wereused for the evaluation. The horizontal axis represents theframe index, whereas the vertical axis respectively spec-ifies the measurements for PSNRs, structural similarities(SSIMs) [82], and peak signal-to-perceptual-noise ratios(PSPNRs) [83]. These three metrics reveal a similar trendwhen the video quality degrades. Furthermore, all the threemetrics present numerical losses.

FIGURE 5. Packet loss ratio of different SVC layers.

The following test simulates the layered video transmis-sions over wireless networks with flexible FEC techniques.Fig. 5 compares the loss ratio of different SVC layers.In this simulation, the SVC video stream was encoded into

104 VOLUME 4, 2016




a layered structure, which included the base layer and twoenhancement layers. The base layer used QCIF formats at7.5Hz, whereas the first enhancement layer applied the QCIFformat at 15Hz, and the second enhancement layer wasbased on CIF formats at 30Hz. To ensure that users couldbrowse the basic quality version of the video, the enhancedFEC was implemented to protect the base layer. The secondenhancement layer was used for the comparison, so it was notprotected by FEC because of least importance. Judging fromFig. 5, the result reveals that the packet loss ratio does notincrease dramatically until the packet error rate reaches 0.2.When the packet error rate is between 0.3 and 0.5, the packetloss ratios of the base layer and the first enhancement layerare almost the same. However, the second enhancement layerhas a higher packet loss ratio than the other two layers due tothe lack of proper FEC techniques.

VIII. FUTURE RESEARCH DIRECTIONS1) A UNIFIED QoE MEASURE MODELIN HETEROGENEOUS NETWORKSAs mentioned above, the heterogeneity of devices directlyinfluences the design of video broadcasting systems.Although a large number of significant works on QoE under-standing have been conducted, there is still no clear descrip-tion of a unified QoE model for comprehensive broadcastingeven in communication systems. The intrinsic property ofvideo signal itself presents complex scalability, especially inhybrid domains. After video streams are encoded and trans-mitted, the error propagation caused by quality degradation,packet losses, delay, format inconformity, etc., is difficult toevaluate. Nevertheless, to make the devices and inner videoservices more ubiquitous, new interactive techniques shouldbe developed. Subjective quality assessment in laboratoryenvironments is losing its relevance to realistic user termi-nals [84]. How to combine the QoE with user background,emotions, behavior, habits and social influences is still anopen topic.

2) DEVICE- AND USER-AWARE ADAPTIVEJOINT CODING MODELIn typical, an entire video stream is initiated after it is dividedinto multilayered substreams to ensure the diversity of multi-ple devices and to satisfy the various demands from users.Following the initiation, these substreams are distributedand transmitted through multiple subchannels in parallel todiverse end-users. Finding a way to transmit these substreamswith support of multiuser experience has a major impact onperformance. However, in current user-centric broadcastingsystems, video coding, channel coding, and joint codingshould develop adaptability and robustness to cope with massinteractions under ubiquitous environments. Thus, how tointelligently, dynamically, and cooperatively encode and pro-tect video streams so that videos can adapt themselves to vari-able circumstances with limited resources is still challenging.

3) NETWORK COGNITIVE COOPERATIVE TRANSMISSIONIn homogeneous networks, the quality of network varieswith time. Since layered video data are sensitive to failures,the broadcasting system needs joint solutions of coding andtransmissions to adapt to quality fluctuation. Nevertheless,the access, interactive modes, user operations, and terminalsemerge diversely in heterogeneous networks. These resultin high-delay, high-cost, and mismatch-bandwidth problems.Thus, video broadcasting faces a new challenge of how todevelop new revolutionary techniques to support ubiquitouscomputing and communications.

IX. CONCLUSIONVideo broadcasting to heterogeneous devices is a researchsubject that requires comprehensive QoE modeling, codingand transmission strategies with heterogeneity support. Thisarticle firstly investigates the key concept of QoE architec-tures by reviewing recent results from scalable video cod-ing to heterogeneous video transmission. Finally, this studybrings the theoretical models closer to practical implementa-tion by presenting an integrated broadcasting system. How-ever, the community still lacks revolutionary techniques.Developing effective methodologies will need interdisci-plinary efforts from academia and industry in the researchfield of video coding, multiuser communication and broad-casting networks.

REFERENCES[1] Cisco Visual Networking Index: Global Mobile Data Traffic Forecast

Update, 2013–2018. [Online]. Available: http://www.cisco.com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/white_paper_c11-520862.html, accessed Mar. 30, 2014.

[2] 33rd Statistical Report on Internet Development in China. [Online].Available: http://www.cnnic.net.cn/, accessed Mar. 30, 2014.

[3] D. Zhang, Z. Yang, V. Raychoudhury, Z. Chen, and J. Lloret, ‘‘An energy-efficient routing protocol using movement trends in vehicular ad hocnetworks,’’ Comput. J., vol. 56, no. 8, pp. 938–946, 2013.

[4] D. Zhang, H. Huang, J. Zhou, F. Xia, and Z. Chen, ‘‘Detecting hot roadmobility of vehicular ad hoc networks,’’Mobile Netw. Appl., vol. 18, no. 6,pp. 803–813, 2013.

[5] S. Schwarz, J. C. Ikuno, M. Simko, M. Taranetz, Q. Wang, and M. Rupp,‘‘Pushing the limits of LTE: A survey on research enhancing the standard,’’IEEE Access, vol. 1, pp. 51–62, 2013.

[6] T. S. Rappaport et al., ‘‘Millimeter wave mobile communications for 5Gcellular: It will work!’’ IEEE Access, vol. 1, pp. 335–349, 2013.

[7] Definition of Quality of Experience (QoE), document COM 12-LS 62-E,ITU-T, Jan. 2007, pp. 1–2.

[8] Quality of Experience Requirements for IPTV Services,document FG IPTV-DOC-0036, ITU-T, Oct. 2006, pp. 1–6.

[9] A. Takahashi, D. Hands, and V. Barriac, ‘‘Standardization activities in theITU for a QoE assessment of IPTV,’’ IEEE Commun. Mag., vol. 46, no. 2,pp. 78–84, Feb. 2008.

[10] T. Hobfeld, R. Schatz, M. Varela, and C. Timmerer, ‘‘Challenges of QoEmanagement for cloud applications,’’ IEEE Commun. Mag., vol. 50, no. 4,pp. 28–36, Apr. 2012.

[11] P. Rengaraju, C.-H. Lung, F. R. Yu, and A. Srinivasan, ‘‘On QoEmonitoring and E2E service assurance in 4G wireless networks,’’IEEE Trans. Wireless Commun., vol. 19, no. 4, pp. 89–96,Aug. 2012.

[12] D. Soldani, M. Li, and R. Cuny, Eds.,QoS and QoEManagement in UMTSCellular Systems. West Sussex, U.K.: Wiley, 2006.

[13] M. Volk, J. Sterle, U. Sedlar, and A. Kos, ‘‘An approach to modelingand control of QoE in next generation networks,’’ IEEE Commun. Mag.,vol. 48, no. 8, pp. 126–135, Aug. 2010.

VOLUME 4, 2016 105




[14] J. Sterle, M. Volk, U. Sedlar, J. Bester, and A. Kos, ‘‘Application-basedNGN QoE controller,’’ IEEE Commun. Mag., vol. 49, no. 1, pp. 92–101,Jan. 2011.

[15] L. Zhou, Z. Yang, Y. Wen, H. Wang, and M. Guizani, ‘‘Resource alloca-tion with incomplete information for QoE-driven multimedia communi-cations,’’ IEEE Trans. Wireless Commun., vol. 12, no. 8, pp. 3733–3745,Aug. 2013.

[16] K. U. R. Laghari, N. Crespi, and K. Connelly, ‘‘Toward total qualityof experience: A QoE model in a communication ecosystem,’’ IEEECommun. Mag., vol. 50, no. 4, pp. 58–65, Apr. 2012.

[17] J. Kim, T.-W. Um, W. Ryu, and B.-S. Lee, ‘‘IPTV systems, stan-dards and architectures: Part II—Heterogeneous networks and terminal-aware QoS/QoE-guaranteed mobile IPTV service,’’ IEEE Commun. Mag.,vol. 46, no. 5, pp. 110–117, May 2008.

[18] C. W. Chan, S. Wee, and J. Apostolopoulos, ‘‘Multiple distortion measuresfor packetized scalable media,’’ IEEE Trans. Multimedia, vol. 10, no. 8,pp. 1671–1686, Dec. 2008.

[19] Y.-F. Ou, Z. Ma, T. Liu, and Y. Wang, ‘‘Perceptual quality assessment ofvideo considering both frame rate and quantization artifacts,’’ IEEE Trans.Circuits Syst. Video Technol., vol. 21, no. 3, pp. 286–298, Mar. 2011.

[20] Y. Wang, M. van der Schaar, S.-F. Chang, and A. C. Loui, ‘‘Classification-based multidimensional adaptation prediction for scalable video codingusing subjective quality evaluation,’’ IEEE Trans. Circuits Syst. VideoTechnol., vol. 15, no. 10, pp. 1270–1279, Oct. 2005.

[21] W. Ji, Z. Li, and Y. Chen, ‘‘Joint source-channel coding and optimizationfor layered video broadcasting to heterogeneous devices,’’ IEEE Trans.Multimedia, vol. 14, no. 2, pp. 443–455, Apr. 2012.

[22] T. Mei, X.-S. Hua, C.-Z. Zhu, H.-Q. Zhou, and S. Li, ‘‘Home video visualquality assessment with spatiotemporal factors,’’ IEEETrans. Circuits Syst.Video Technol., vol. 17, no. 6, pp. 699–706, Jun. 2007.

[23] H. Schwarz, D. Marpe, and T. Wiegand, ‘‘Overview of the scalable videocoding extension of the H.264/AVC standard,’’ IEEE Trans. Circuits Syst.Video Technol., vol. 17, no. 9, pp. 1103–1120, Sep. 2007.

[24] M. Effros, ‘‘Universal multiresolution source codes,’’ IEEE Trans. Inf.Theory, vol. 47, no. 6, pp. 2113–2129, Sep. 2001.

[25] B.-J. Kim, Z. Xiong, and W. A. Pearlman, ‘‘Low bit-rate scalable videocoding with 3-D set partitioning in hierarchical trees (3-D SPIHT),’’IEEE Trans. Circuits Syst. Video Technol., vol. 10, no. 8, pp. 1374–1387,Dec. 2000.

[26] V. K. Goyal, ‘‘Multiple description coding: Compression meets the net-work,’’ IEEE Signal Process. Mag., vol. 18, no. 5, pp. 74–93, Sep. 2001.

[27] P. Correia, P. A. Assuncao, and V. Silva, ‘‘Multiple description of codedvideo for path diversity streaming adaptation,’’ IEEE Trans. Multimedia,vol. 14, no. 3, pp. 923–935, Jun. 2012.

[28] P. Xia, S.-H. G. Chan, and X. Jin, ‘‘Optimal bandwidth assignment formultiple-description-coded video,’’ IEEE Trans.Multimedia, vol. 13, no. 2,pp. 366–375, Apr. 2011.

[29] O. S. Badarneh, Y. Qian, B. Rong, A. K. Elhakeem, and M. Kadoch,‘‘Multiple description coding based video multicast over heterogeneouswireless ad hoc networks,’’ in Proc. IEEE ICC, Jun. 2009, pp. 1–6.

[30] T. Schierl, T. Stockhammer, and T. Wiegand, ‘‘Mobile video transmissionusing scalable video coding,’’ IEEE Trans. Circuits Syst. Video Technol.,vol. 17, no. 9, pp. 1204–1217, Sep. 2007.

[31] S. Jaeckel and V. Jungnickel, ‘‘Scalable video transmission in multiantennabroadcast systems,’’ in Proc. 14th Eur. Wireless Conf., Jun. 2008, pp. 1–6.

[32] W. Huang and K. L. Yeung, ‘‘Optimal opportunistic multicast for minimiz-ing broadcast latency in wireless networks,’’ inProc. IEEE ICC, May 2010,pp. 1–5.

[33] M. Luby, T. Gasiba, T. Stockhammer, and M. Watson, ‘‘Reliable multi-media download delivery in cellular broadcast networks,’’ IEEE Trans.Broadcast., vol. 53, no. 1, pp. 235–246, Mar. 2007.

[34] T. Jiang, H. Wang, and A. V. Vasilakos, ‘‘QoE-driven channel allocationschemes for multimedia transmission of priority-based secondary usersover cognitive radio networks,’’ IEEE J. Sel. Areas Commun., vol. 30, no. 7,pp. 1215–1224, Aug. 2012.

[35] I. Berenguer, I. J. Wassell, and X. Wang, ‘‘Opportunistic user schedulingand antenna selection in the downlink of multiuser MISO systems,’’ inProc. IEEE 61st VTC, vol. 2. May/Jun. 2005, pp. 1138–1142.

[36] W. Ji, Z. Li, and Y. Chen, ‘‘QoE-based opportunistic transmission forvideo broadcasting in heterogeneous circumstance,’’ in Proc. 20th ACMMultimedia, Oct. 2012, pp. 1029–1032.

[37] S.-C. Wang and W. Liao, ‘‘Cooperative multicasting for wireless scal-able video transmissions,’’ IEEE Trans. Commun., vol. 61, no. 9,pp. 3980–3989, Sep. 2013.

[38] C.-W. Huang, S.-M. Huang, P.-H. Wu, S.-J. Lin, and J.-N. Hwang,‘‘OLM: Opportunistic layered multicasting for scalable IPTV over mobileWiMAX,’’ IEEE Trans. Mobile Comput., vol. 11, no. 3, pp. 453–463,Mar. 2012.

[39] M. Stoufs, A. Munteanu, J. Cornelis, and P. Schelkens, ‘‘Scalable jointsource-channel coding for the scalable extension of H.264/AVC,’’ IEEETrans. Circuits Syst. Video Technol., vol. 18, no. 12, pp. 1657–1670,Dec. 2008.

[40] J. Liu, B. Li, Y. T. Hou, and I. Chlamtac, ‘‘On optimal layering andbandwidth allocation for multisession video broadcasting,’’ IEEE Trans.Wireless Commun., vol. 3, no. 2, pp. 656–667, Mar. 2004.

[41] J. Xu, X. Shen, J. W. Mark, and J. Cai, ‘‘Adaptive transmission ofmulti-layered video over wireless fading channels,’’ IEEE Trans. WirelessCommun., vol. 6, no. 6, pp. 2305–2314, Jun. 2007.

[42] W. Ji and Z. Li, ‘‘Joint layered video and digital fountain coding for multi-channel video broadcasting,’’ in Proc. 18th Int. Conf. ACM Multimedia,Oct. 2010, pp. 1223–1226.

[43] W. Tu, ‘‘Efficient resource utilization for multi-flow wireless multi-casting transmissions,’’ IEEE J. Sel. Areas Commun., vol. 30, no. 7,pp. 1246–1258, Aug. 2012.

[44] F. Fu and M. van der Schaar, ‘‘A systematic framework for dynamicallyoptimizing multi-user wireless video transmission,’’ IEEE J. Sel. AreasCommun., vol. 28, no. 3, pp. 308–320, Apr. 2010.

[45] M. van der Schaar, Y. Andreopoulos, and Z. Hu, ‘‘Optimized scalable videostreaming over IEEE 802.11 a/e HCCA wireless networks under delayconstraints,’’ IEEE Trans. Mobile Comput., vol. 5, no. 6, pp. 755–768,Jun. 2006.

[46] J. Huang, Z. Li, M. Chiang, and A. K. Katsaggelos, ‘‘Joint source adap-tation and resource allocation for multi-user wireless video streaming,’’IEEE Trans. Circuits Syst. Video Technol., vol. 18, no. 5, pp. 582–595,May 2008.

[47] E.Maani, P. V. Pahalawatta, R. Berry, T. N. Pappas, and A. K. Katsaggelos,‘‘Resource allocation for downlink multiuser video transmission overwireless lossy networks,’’ IEEE Trans. Image Process., vol. 17, no. 9,pp. 1663–1671, Sep. 2008.

[48] Y. Li, Z. Li, M. Chiang, and A. R. Calderbank, ‘‘Content-aware distortion-fair video streaming in congested networks,’’ IEEE Trans. Multimedia,vol. 11, no. 6, pp. 1182–1193, Oct. 2009.

[49] M. Wien, H. Schwarz, and T. Oelbaum, ‘‘Performance analysis of SVC,’’IEEE Trans. Circuits Syst. Video Technol., vol. 17, no. 9, pp. 1194–1203,Sep. 2007.

[50] Y. Wang, L.-P. Chau, and K.-H. Yap, ‘‘Spatial resolution decision inscalable bitstream extraction for network and receiver aware adaptation,’’in Proc. IEEE ICME, Jun./Apr. 2008, pp. 577–580.

[51] T. B. Abanoz and A. M. Tekalp, ‘‘Optimization of encoding configurationin scalable multiple description coding for rate-adaptive P2P video multi-casting,’’ in Proc. 16th IEEE ICIP, Nov. 2009, pp. 3741–3744.

[52] E.Maani andA.K.Katsaggelos, ‘‘Optimized bit extraction using distortionmodeling in the scalable extension of H.264/AVC,’’ IEEE Trans. ImageProcess., vol. 18, no. 9, pp. 2022–2029, Sep. 2009.

[53] Q. Zhang, Q. Guo, Q. Ni, W. Zhu, and Y.-Q. Zhang, ‘‘Sender-adaptiveand receiver-driven layered multicast for scalable video over the Internet,’’IEEE Trans. Circuits Syst. Video Technol., vol. 15, no. 4, pp. 482–495,Apr. 2005.

[54] D. Taubman and J. Thie, ‘‘Optimal erasure protection for scalably com-pressed video streams with limited retransmission,’’ IEEE Trans. ImageProcess., vol. 14, no. 8, pp. 1006–1019, Aug. 2005.

[55] W. Ji and Z. Li, ‘‘HeterogeneousQoS video broadcastingwith optimal jointlayered video and digital fountain coding,’’ in Proc. IEEE ICC, Jun. 2011,pp. 1–6.

[56] Z. Li, G. M. Schuster, A. K. Katsaggelos, and B. Gandhi, ‘‘Rate-distortionoptimal video summary generation,’’ IEEE Trans. Image Process., vol. 14,no. 10, pp. 1550–1560, Oct. 2005.

[57] M. Bystrom and J. W. Modestino, ‘‘Combined source-channel codingschemes for video transmission over an additive white Gaussian noisechannel,’’ IEEE J. Sel. Areas Commun., vol. 18, no. 6, pp. 880–890,Jun. 2000.

[58] D. C. Cernea, A.Munteanu, A. Alecu, J. Cornelis, and P. Schelkens, ‘‘Scal-able joint source and channel coding of meshes,’’ IEEE Trans. Multimedia,vol. 10, no. 3, pp. 503–513, Apr. 2008.

[59] Z. He, J. Cai, and C. W. Chen, ‘‘Joint source channel rate-distortionanalysis for adaptive mode selection and rate control in wireless videocoding,’’ IEEE Trans. Circuits Syst. Video Technol., vol. 12, no. 6,pp. 511–523, Jun. 2002.

106 VOLUME 4, 2016




[60] N. Raja, Z. Xiong, andM. Fossorier, ‘‘Combined source-channel coding ofimages under power and bandwidth constraints,’’ EURASIP J. Adv. SignalProcess., vol. 2007, Dec. 2007.

[61] X. Jaspar, C. Guillemot, and L. Vandendorpe, ‘‘Joint source–channel turbotechniques for discrete-valued sources: From theory to practice,’’ Proc.IEEE, vol. 95, no. 6, pp. 1345–1361, Jun. 2007.

[62] L. Qian, D. L. Jones, K. Ramchandran, and S. Appadwedula, ‘‘A generaljoint source-channel matching method for wireless video transmission,’’ inProc. DCC, Jun. 1999, pp. 414–423.

[63] Q. Xu, V. Stanković, and Z. Xiong, ‘‘Distributed joint source-channelcoding of video using raptor codes,’’ IEEE J. Sel. Areas Commun., vol. 25,no. 4, pp. 851–861, May 2007.

[64] F. Zhai, Y. Eisenberg, T. N. Pappas, R. Berry, and A. K. Katsaggelos,‘‘Rate-distortion optimized hybrid error control for real-time packe-tized video transmission,’’ IEEE Trans. Image Process., vol. 15, no. 1,pp. 40–53, Jan. 2006.

[65] S. N. Diggavi, A. R. Calderbank, S. Dusad, and N. Al-Dhahir, ‘‘Diversityembedded space–time codes,’’ IEEE Trans. Inf. Theory, vol. 54, no. 1,pp. 33–50, Jan. 2008.

[66] S. Dusad, S. N. Diggavi, N. Al-Dhahir, and A. R. Calderbank, ‘‘Diver-sity embedded codes: Theory and practice,’’ IEEE J. Sel. Topics SignalProcess., vol. 2, no. 2, pp. 202–219, Apr. 2008.

[67] Y. Li, M. Chiang, A. R. Calderbank, and S. N. Diggavi, ‘‘Optimal rate-reliability-delay tradeoff in networks with composite links,’’ IEEE Trans.Commun., vol. 57, no. 5, pp. 1390–1401, May 2009.

[68] J. Chui and A. R. Calderbank, ‘‘Multilevel diversity-embedded space-timecodes for video broadcasting over WiMAX,’’ in Proc. IEEE ISIT, vol. 2.Jul. 2008, pp. 1068–1072.

[69] C. Bilen, E. Erkip, and Y. Wang, ‘‘Layered video multicast usingdiversity embedded space time codes,’’ in Proc. IEEE Sarnoff Symp.,Mar./Apr. 2009, pp. 1–5.

[70] O. Alay, P. Liu, Y. Wang, E. Erkip, and S. S. Panwar, ‘‘Cooperative layeredvideo multicast using randomized distributed space time codes,’’ IEEETrans. Multimedia, vol. 13, no. 5, pp. 1127–1140, Oct. 2011.

[71] Q. Zhang and Y.-Q. Zhang, ‘‘Cross-layer design for QoS support in multi-hop wireless networks,’’ Proc. IEEE, vol. 96, no. 1, pp. 64–76, Jan. 2008.

[72] R. Ballagas, J. Borchers, M. Rohs, and J. G. Sheridan, ‘‘The smart phone:A ubiquitous input device,’’ IEEE Pervasive Comput., vol. 5, no. 1,pp. 70–77, Jan. 2006.

[73] H. Koumaras, A. Kourtis, D. Martakos, and J. Lauterjung, ‘‘QuantifiedPQoS assessment based on fast estimation of the spatial and temporal activ-ity level,’’Multimedia Tools Appl., vol. 34, no. 3, pp. 355–374, Sep. 2007.

[74] Z. Ma, M. Xu, Y.-F. Ou, and Y. Wang, ‘‘Modeling of rate and perceptualquality of compressed video as functions of frame rate and quantizationstepsize and its applications,’’ IEEE Trans. Circuits Syst. Video Technol.,vol. 22, no. 5, pp. 671–682, May 2012.

[75] T. Liu, Y. Wang, J. M. Boyce, H. Yang, and Z. Wu, ‘‘A novel videoquality metric for low bit-rate video considering both coding and packet-loss artifacts,’’ IEEE J. Sel. Topics Signal Process., vol. 3, no. 2,pp. 280–293, Apr. 2009.

[76] M. H. Pinson and S. Wolf, ‘‘A new standardized method for objec-tively measuring video quality,’’ IEEE Trans. Broadcast., vol. 50, no. 3,pp. 312–322, Sep. 2004.

[77] S. Wolf and M. Pinson, ‘‘Video quality measurement techniques,’’Nat. Telecommun. Inf. Admin., Washington, DC, USA,Tech. Rep. TR-02-392, 2002.

[78] K. Seshadrinathan andA. C. Bovik, ‘‘Motion tuned spatio-temporal qualityassessment of natural videos,’’ IEEE Trans. Image Process., vol. 19, no. 2,pp. 335–350, Feb. 2010.

[79] J. You, T. Ebrahimi, and A. Perkis, ‘‘Attention driven foveated videoquality assessment,’’ IEEE Trans. Image Process., vol. 23, no. 1,pp. 200–213, Jan. 2014.

[80] K. Seshadrinathan, R. Soundararajan, A. C. Bovik, and L. K. Cormack,‘‘Study of subjective and objective quality assessment of video,’’ IEEETrans. Image Process., vol. 19, no. 6, pp. 1427–1441, Jun. 2010.

[81] J. Reichel and M. Wien, ‘‘Joint scalable video model JSVM-8,’’document ISO/IEC JTC1/SC29/WG11 ITU-T SG16 Q.6, 2006.

[82] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, ‘‘Image qualityassessment: From error visibility to structural similarity,’’ IEEE Trans.Image Process., vol. 13, no. 4, pp. 600–612, Apr. 2004.

[83] X. K. Yang,W. S. Ling, Z. K. Lu, E. P. Ong, and S. S. Yao, ‘‘Just noticeabledistortion model and its applications in video coding,’’ Signal Process.,Image Commun., vol. 20, no. 7, pp. 662–680, Aug. 2005.

[84] J.-S. Lee, F. De Simone, T. Ebrahimi, N. Ramzan, and E. Izquierdo,‘‘Quality assessment of multidimensional video scalability,’’ IEEEWireless Commun. Mag., vol. 50, no. 4, pp. 38–46, Apr. 2012.

[85] L. Atzori, C. W. Chen, T. Dagiuklas, and H. R. Wu, ‘‘QoE managementin emerging multimedia services,’’ IEEE Commun. Mag., vol. 50, no. 4,pp. 18–19, Apr. 2012.

BO-WEI CHEN (M’14) received the Ph.D. degreein electrical engineering from NationalCheng Kung University (NCKU), Tainan, Taiwan,in 2009. Until 2010, he was a Post-DoctoralResearch Fellow with the Department ofElectrical Engineering, NCKU. He is currently aPost-Doctoral Research Fellow with the Depart-ment of Electrical Engineering, Princeton Univer-sity, Princeton, NJ, USA. His research interestsinclude big data analysis, machine learning, social

network analysis, audiovisual sensor networks, semantic analysis, and videoencoders. He serves as the Chair of the Signal Processing Chapter of theIEEE Harbin Section.

WEN JI (M’09) received the M.S. andPh.D. degrees in communication and informationsystems from Northwestern Polytechnical Univer-sity, Xi’an, China, in 2003 and 2006, respectively.From 2007 to 2009, she was a Post-DoctoralResearch Fellow with the Institute of Comput-ing Technology, Chinese Academy of Sciences,Beijing, China, where she was an AssistantProfessor from 2009 to 2010 and is currently anAssociate Professor. Her research areas include

video communication and networking, video coding, channel coding, infor-mation theory, optimization, network economics, and pervasive computing.She is the Vice Chair of the Signal Processing Chapter of the IEEE HarbinSection.

FENG JIANG received the B.S., M.S., andPh.D. degrees in computer science from theHarbinInstitute of Technology (HIT), Harbin, China,in 2001, 2003, and 2008, respectively. He is cur-rently an Associate Professor with the Departmentof Computer Science, HIT, China. His researchinterests include computer vision, pattern recog-nition, and image and video processing. He is theSecretary of the Signal Processing Chapter of theIEEE Harbin Section.

SEUNGMIN RHO received the M.S. andPh.D. degrees in computer science from AjouUniversity, Korea, in 2003 and 2008, respectively.In 2008 and 2009, he was a Post-DoctoralResearch Fellow with the Computer Music Lab-oratory, School of Computer Science, CarnegieMellon University. He was a Research Professorwith the School of Electrical Engineering, KoreaUniversity, from 2009 to 2011. In 2012, he was anAssistant Professor with the Division of Informa-

tion and Communication, Baekseok University. He was a Faculty Memberwith the Department of Multimedia, Sungkyul University, Korea, in 2013.He is currently an Associate Professor with the Department of ComputerEngineering, Mevlana University, Konya, Turkey. His current research inter-ests include database, big data analysis, music retrieval, multimedia systems,machine learning, knowledge management, and computational intelligence.

VOLUME 4, 2016 107



QoE-enabled big video streaming for large-scale heterogeneous clients and networks in smart cities

Education

Transcript of QoE-enabled big video streaming for large-scale heterogeneous clients and networks in smart cities