12-05-2011, 02:39 PM
Abstract
Self-adaptive management and quality adaptationof multimedia services are open challenges in the heterogeneouswireless Internet, where different wireless access points potentiallyenable anywhere anytime Internet connectivity. One ofthe most challenging issues is to guarantee streaming continuitywith maximum quality, despite possible handoffs at multimediaprovisioning time. To enable handoff management to self-adaptto specific application requirements with minimum resourceconsumption, this paper offers three main contributions. First, itproposes a simple way to specify handoff-related service-levelobjectives that are focused on quality metrics and tolerabledelay. Second, it presents how to automatically derive from theseobjectives a set of parameters to guide system-level configurationabout handoff strategies and dynamic buffer tuning. Third, itdescribes the design and implementation of a novel handoffmanagement infrastructure for maximizing streaming qualitywhile minimizing resource consumption. Our infrastructure exploitsi) experimentally evaluated tuning diagrams for resourcemanagement and ii) handoff prediction/awareness. The reportedresults show the effectiveness of our approach, which permits toachieve the desired quality-delay tradeoff in common Internetdeployment environments, even in presence of vertical handoffs.Index Terms—Resource management, handoff management,quality of service, wireless networks, multimedia streaming.
I. INTRODUCTION
THE rapid and widespread diffusion of wireless networksis leading to the realization of the so-called wirelessInternet, which aims to offer Internet-based services anytimeanywhere via heterogeneous wireless access points [1], [2] usingWi-Fi, Bluetooth, General Packet Radio Service (GPRS),Universal Mobile Telecommunications System (UMTS), etc.Service providers are taking this opportunity to put on themarket new and advanced mobile multimedia services, rangingfrom digital television (TV) and Voice over Internet Protocol(VoIP) to video calls and video surveillance.In this wireless Internet scenario, classic strategies and architecturesfor service delivery and management are seriously challenged. For example, services need to be continuouslyavailable to mobile devices, which are commonly equippedwith multiple wireless communication interfaces. The availabilityof such interfaces potentially increases the coverageof content delivery services for mobile users. Mobile clientscould, in principle, choose at anytime the access point andnetwork type that best fit their service and user requirements.But the heterogeneity of these wireless interfaces, thebandwidth fluctuations that are inherent to wireless networks,and the temporary loss of connectivity due to handoffs [3],[4], [5] pose tough technical problems. Another example ofchallenging opportunity is mobile streaming continuity, i.e.the property of maintaining audio/video content delivery byminimizing Quality of Service (QoS) degradations as clientsroam through different access points.Current commercial solutions of wireless multimedia offer alimited support for streaming continuity: they usually providea predetermined set of service profiles (usually defined bynetwork operators), each one relying on predefined QoSobjectives. These objectives are statically mapped into systemlevelconfiguration parameters, typically via over-provisioningof resources to face bandwidth fluctuations and handoff issues.However, it is widely recognized that static resourcemanagement is unsuitable for dynamic wireless environments[6], [7]. We claim that resource allocation should depend onpersonalized QoS requirements and be fully aware of handoffoccurrences. In other words, resources should be effectivelytuned at runtime depending on i) wireless characteristics(supported handoff strategies, origin/target wireless technology,available bandwidth, network delay, etc.) and ii) servicerequirements (tolerable delay, desired QoS level, etc.).In this context, this paper proposes three contributions tothe research area of self-adaptive management for mobilestreaming continuity in the wireless Internet.First, it defines a simple and intuitive way to let serviceproviders specify service-level objectives to be enforced onthe last wireless hop during handoffs, namely Handoff ServiceLevel Objectives (HSLO). HSLOs include a small set of easyto-use parameters, based on
Download full report
http://ieeexplore.ieeeiel5/4275028/53748...er=5374829