Video Streaming Application Optimization Technologies In Cloud Environment

As an emerging commercial pattern,cloud platform can offer users virtualized computing resources with high reliability and scalability in a real-time on-demand style.Its improved cost-effectiveness attracts more and more individuals and enterprises to migrate their services and applications onto cloud.Here,video streaming applications have become the most representative cloud application because of its distinctive characteristics such as large user scales,unique demand features and wide application scenarios.Actually,video streaming has undergone major development in the past decade.The corresponding network traffic has accounted for the vast majority of all the Internet traffic.The popularity of various portable smart mobile devices propels the development of video streaming and high-bandwidth video streaming applications are emerging in endlessly.The current video streaming has brought more and more pressure to the application providers' servers and networks.Cloud computing is an ideal approach to deal with these challenges:"infinite" cloud resources appeal to the characteristics of being large scale and highly dynamic,while geo-distributed cloud data centers can support good user quality of experience such as low access latency.How to better improve the performance of cloud video streaming applications would be a challenging and valuable issue.Some major challenges in the above problem are listed as follows:1)Due to some factors such as market competition,video streaming cloud services from different cloud vendors are usually assigned some unique features superior to others.One specific video streaming cloud service can adjust its QoS level based on users' real-time demands.This adjustability can upgrade the application's performance.However,it also promotes the difficulty of video streaming cloud service selection greatly.2)The elasticity of cloud resources enable users to scale up and down resource reservation to obtain high resource utilization.As a delay of several minutes is required to update the configuration and to launch new instances for each rescaling operation,the video streaming providers need to proactively proceed with the rescaling operation in advance based on predictive information.However,the existing prediction techniques cannot guarantee the entire accuracy.This mismatch can influence the guarantee of services level agreement.3)As a support of providing reliable and ubiquitous Internet access to mobile devices,cellular-based access technologies play a vital role.However,these wireless network channel conditions vary as users move and fluctuate depending on weather,building shields,congestion,etc.Such random and dynamic characteristics of wireless network condition may damage both stability and fluency of live video streaming.4)Portable mobile devices allow users to share live videos anywhere or watch high-quality on-demand videos anytime.The high visual experience exerts great pressure on the Internet network and access.Some existing D2D communication researches usually build on users' random mobility or position snapshot in cellular networks.The random mobility of mobile users leads to dynamic variation of communication duration and quality,making D2D communication be low-efficient especially for video transmissions.In view of the above challenges,we propose our solutions for cloud video streaming application optimization.Specifically,the main contributions of our work are summarized as follows:1)In order to realize the integrative optimization of cloud video streaming application construction,deployment,mobile access and user cooperation,we design a cloud video streaming application optimization framework.This framework consists of four layers,i.e.,cloud video streaming application construction layer,cloud video streaming application optimization layer,cloud video streaming service access layer and cloud video streaming service user layer.In detail,based on distributed service brokers,cloud video streaming application construction layer focuses on building cloud applications with desired functionalities by using the bottom cloud video streaming services.Cloud video streaming application optimization layer pays attention to optimizing the resource utilization and improving the suitability of application and market by taking some auxiliary information into account such as applications' temporal-spatial usage and user groups.Cloud video streaming service access layer aims at improving the individual access throughput.Cloud video streaming service user layer fully considers the user mobility and analyze the influence of user cooperation pattern to the video quality.2)With the cloud's charismatic storage and computation power,more and more traditional video services such as encoding are being migrated onto cloud platforms.These cloud services on different cloud platforms could be employed to form cross-cloud video streaming applications.However,a cloud service may have various adjustable quality of service(QoS)properties.This characteristic makes it costly and time consuming to mine qualified ones from massive candidate cloud services for developing a desired video streaming application.In view of this challenge,a cloud service selection method,named CSSM,is proposed in this paper.It takes the utility value as the evaluation index and aims at finding optimal or near-optimal trusted service composition solutions from a set of cloud services on users' demands.Technically,the user preference on each QoS metric is formalized as the preference interval for enhancing the fitness of a service composition solution.Furthermore,an extended top-k iteration composition process is performed among cloud services to get an optimal or near-optimal trusted service composition solution.Both theoretical analysis and experimental evaluation are conducted to guarantee the feasibility and efficiency of the CSSM.3)With the prosperity of media streaming applications over the Internet in the past decades,multimedia data has sharply increased(categorized as multimedia big data),which exerts more pressure on the infrastructure,such as networking of the application provider.In order to move this hurdle,an increasing number of traditional media streaming applications have migrated from a private server cluster onto the cloud.With the elastic resource provisioning and centralized management of the cloud,the operational costs of media streaming application providers can decrease dramatically.However,to the best of our knowledge,existing migration solutions do not fully take viewer information such as hardware condition into consideration.In this article,we consider the deployment optimization problem named ODP by leveraging local memories at each viewer.Considering the NP-hardness of calculating the optimal solution,we turn to propose computationally tractable algorithms.Specifically,we unfold the original problem into two interactive subproblems:coarse-grained migration sub-problem and fine-grained scheduling sub-problem.Then,the corresponding offline approximation algorithms with performance guarantee and computational efficiency are given.The results of extensive evaluation show that compared with the baseline algorithm without leveraging local memories at viewers,our proposed algorithms and their online versions can decrease total bandwidth reservation and enhance the utilization of bandwidth reservation dramatically.4)By leveraging powerful mobile devices such as smartphones,people can enjoy a rich real-time sensing cognition of what they are interested in anytime and anywhere.As a key support for the propagation of these richer live media contents,cellular-based access technologies play a vital role to provide reliable and ubiquitous Internet access to mobile devices.However,these limited wireless network channel conditions vary and fluctuate depending on weather,building shields,congestion,etc.,which degrade the quality of live video streaming dramatically.To address this challenge,we propose to use crowdsourcing brokerage in future networks which can improve each mobile user's bandwidth condition and reduce the fluctuation of network condition.Further,to serve mobile users better in this crowdsourcing style,we study the brokerage scheduling problem which aims at maximizing the user's QoE(quality of experience)satisfaction degree cost-effectively.Both offline and online algorithms are proposed to solve this problem.The results of extensive evaluations demonstrate that by leveraging crowdsourcing technique,our solution can cost-effectively guarantee a higher quality view experience.5)The widespread use of mobile devices such as smartphones propels the development of new-fashioned video applications like 3D stereo video and mobile cloud game,exerting more pressure on current mobile access network.To cater for the development of mobile video streaming,D2D(Device-to-Device)communication paradigm is treated as a promising technology to compensate this gap.However,existing D2D communication researches usually build on users'random mobility or position snapshot in cellular networks.This random mobility leads to dynamic variation of communication duration and quality,making D2D communication be low-efficient especially for video transmissions.To address this challenge,we introduce the crowdsourcing paradigm which can facilitate some control on the movement of recruited crowdsourcing users and then study how to optimize movement control decision.Specifically,based on a realistic 4G/LTE network throughput measurement study,we formulate movement control decision to be a cost-constrained user recruitment problem.Considering the hardness property of this problem,we focus first on a special case of the problem and propose a pseudo-polynomial time complexity optimal solution.Then,we apply this solution to solve the general problem and propose a graph-partition-based algorithm.Extensive experiments show that our solution is able to improve the efficiency of real-time D2D communication dramatically for mobile videos.