[ Pobierz całość w formacie PDF ]
.Conversational;2.Streaming;3.Interactive;4.Background.The characteristics of each of these classes are described in the followingsections.9.1.1 Conversational ClassAs its name implies, this is the class to which conversations belong.It applies to not only speech itself, but to any application that involves person-to-person communication in real-time, such as videoconferencing and interac-tive video games.221222QoS in Integrated 3G NetworksThe basic qualities required for speech itself are low delay, low jitter(delay-variation), reasonable clarity (codec quality), and absence of echo.In the case of multimedia applications, such as videoconferencing, it is also necessary to maintain correct relative timing of the different media streams, for example, lip synchronization.This class is tolerant of some errors, as dropping or corruption of avoice packet lasting for a typical 20 ms is unlikely to be detected by a user.The degree of error protection required is variable with some applications.Some compression algorithms divide the subject material into different setswhose importance differs.For example, with H.324 videoconferencing,information associated with lip movement is more critical to quality thanbackground scenery, and hence may be offered a higher level of error protec-tion; while for speech, the AMR codec for UMTS has three distinct catego-ries of information bits.One of the key preliminaries to QoS negotiation for this class is thechoice of codec to use.If the codecs used by the two peer end users were different, then there would be a need for transcoding by a media gateway lead-ing to major delays and probable loss of content quality.For CS and ATMcore networks, this involves H.245 capability exchanges; while for IM corenetworks and remote IP networks, the initial SIP INVITE contains an SDPlist of available codecs for the recipient to respond to, with the final decision made by the session initiator.The choice of codec must also be compatiblewith the network resources available, and so is influenced by RNC, SGSN,and GGSN QoS profile negotiations.The role of gateways and control func-tions for this was described in Section 8.1.3.9.1.2 Streaming ClassThe streaming class consists of real-time applications that send informationto a viewer or listener, but without having any human response.Examples ofthis include video-on-demand, live MP3 listening, Web-radio, news streams,and multicasts.Because of the absence of interaction, there is no longer a need for lowdelay, but the requirements for low jitter and media synchronization remain.The error tolerance remains, but higher quality is required when listening to hi-fi music than for speech.The removal of the low delay criterion makes it possible to use buffer-ing techniques in the end-user equipment to even out the delay variation, sothe acceptable level of network jitter is higher than for the conversationalUMTS Classes of Service223class.As in the conversational case, the choice of codec needs to benegotiated.9.1.3 Interactive ClassThis class covers both humans and machines that request data from anotherdevice.Examples of this include some games, network management sys-tems polling for statistics, and people actively Web-browsing or searchingdatabases.The first requirement is for a delay that is within an application time-out, or reasonably prompt, for the human activities, but not as low as for the conversational class.The second need is for data integrity.9.1.4 Background ClassBackground class covers all applications that either receive data passivelyor actively request it, but without any immediate need to handle the data.Examples of this include e-mails, short message service, and file transfers.The only requirement is for data integrity, although large file transferswill also require an adequate throughput.9.2 UMTS QoS Implementation9.2.1 Architecture and QoS ProfileThe basis of provision of the required QoS in UMTS is the selection of bear-ers with the appropriate characteristics.Several different bearers will be used in support of any call, and the different types are shown in the bearer architecture diagram of Figure 9.1.Each UMTS bearer service is characterized by a number of quality andperformance factors, and these are listed below.Some applications entail several distinct subflows over the radio access bearers that require separate QoSrequirements, especially error protection, but otherwise they use the samecharacteristics as the UMTS bearer.Where a radio access bearer has morethan one subflow, it also has a corresponding number of radio bearers.• Traffic class: This is conversational, streaming, interactive, orbackground.• Maximum bit rate (Kbps): This is the maximum number of bits thata UMTS bearer can deliver to a service access point (SAP) in a224QoS in Integrated 3G NetworksUMTSTEMTUTRANCN IuCNTEEDGEgatewayNODEEnd-to-end serviceTE/MT localUMTS Bearer ServiceExternalbearer serviceUMTS bearer servicebearer serviceRadio access bearer serviceCN bearerserviceRadio bearerIu bearerBackboneserviceservicebearer serviceUTRAPhysicalFDD/TDDbearerserviceserviceFigure 9.1 UMTS QoS architecture.(Source: [1] © ETSI 2002.)specified interval [ Pobierz całość w formacie PDF ]