1、 TECHNICAL REPORT ATIS-0100032 RELATION BETWEEN ITU-T (Y.1541/Y.1221) AND 3GPP UMTS/LTE QOS CLASSES ATIS is the leading technical planning and standards development organization committed to the rapid development of global, market-driven standards for the information, entertainment and communication
2、s industry. More than 200 companies actively formulate standards in ATIS Committees, covering issues including: IPTV, Cloud Services, Energy Efficiency, IP-Based and Wireless Technologies, Quality of Service, Billing and Operational Support, Emergency Services, Architectural Platforms and Emerging N
3、etworks. In addition, numerous Incubators, Focus and Exploratory Groups address evolving industry priorities including Smart Grid, Machine-to-Machine, Networked Car, IP Downloadable Security, Policy Management and Network Optimization. ATIS is the North American Organizational Partner for the 3rd Ge
4、neration Partnership Project (3GPP), a member and major U.S. contributor to the International Telecommunication Union (ITU) Radio and Telecommunications Sectors, and a member of the Inter-American Telecommunication Commission (CITEL). ATIS is accredited by the American National Standards Institute (
5、ANSI). For more information, please visit . Notice of Disclaimer second, some parameters like delay are defined and measured completely differently; and third, the ITU-T QoS classes are end-to-end but the 3GPP classes are not. Recognizing these incompatibilities, Committee T1 previously considered b
6、-3A175 one possible mapping between the Y.1541 QoS classes (and associated traffic descriptors) and a corresponding set of values for 3GPP “bearer service attributes” defined for the four UMTS classes . This Technical Report provides a formal documentation of the earlier Committee T1 effort b-3A175
7、(sections 6 to 9). It then provides descriptions of the newer 3GPP LTE QoS classes (section 10) and demonstrates inconsistencies between the LTE and Y.1541 classes similar to those observed between the UMTS and Y.1541 classes (section 11). Conclusions and recommendations are provided in section 12 a
8、ll in the context of addressing a good interworking solutiona critical precursor to achieving truly harmonized end-to-end QoS as wireless and wireline networks converge. For convenience, a summary of conclusions is presented here as follows. There are two major incompatibilities in deriving a meanin
9、gful and consistent mapping between the Y.1541 classes and the 3GPP UMTS/LTE: Transfer Delay - The 3GPP Transfer Delays are defined as maxima whereas the Y.1541 Transfer Delays are expressed as mean values. Also, the 3GPP Transfer Delays are defined for the 3GPP domain only whereas the Y.1541 Transf
10、er Delays are expressed as end-to-end, but the ITU experts were never able to agree on an allocation to segments like wireless access. Delay Variation Delay Variation in the packet stream is specified for the Y.1541 classes (a distribution statistic) but is not addressed in any way for the 3GPP clas
11、ses. 4As an example of the concerns expressed, b-3P149 states in part: “The QoS Parameters, Parameter Values and QoS Classes defined in 3GPP specifications are different from, and may be incompatible with those in ITU-T specifications. This may result in interoperability problems between 3GPP-based
12、wireless networks and ITU-T based wireline networks. Adversely impacted services may include Voice over IP, Video Streaming, and multimedia services such as Telecommunications for Disaster Relief. Either alignment of specifications or the definition of standardized interworking that will not adverse
13、ly impact service delivery between networks based on 3GPP and ITU-T specifications is therefore required.” ATIS-0100032 4 Going forward, it is recommended to approach the end-to-end QoS problem by developing a detailed specification of interworking between different network segments (e.g., between L
14、TE and optical IP backbones). This would necessitate, for example, developing an industry-wide default markings between things such as LTE QCI-related markings and Diffserv Code Points, or alternatively, development of interworking guidelines that can be used to derive appropriate Service Level Agre
15、ements (SLA) between service providers. 6 COMPARISON OF Y.1541 AND UMTS QOS CLASSES The content of sections 6 through 9 are based on earlier ATIS Committee T1 work by Neal Seitz. The content in these sections is taken from Neals earlier ATIS contributions b-3A175. Table 1 illustrates the Y.1541 QoS
16、classes and associated network performance objectives. These specifications apply between user-network interfaces that delimit end-to-end IP flows. The objectives are designed to be achievable on common IP network implementations. Classes 0 and 1 place upper bounds on packet transfer delay and packe
17、t loss. They also limit packet delay variation. Classes 2 and 3 place upper bounds on packet transfer delay and packet loss, but do not limit packet delay variation. Classes 0 and 2 differ from Classes 1 and 3 in their packet transfer delay objectives. Class 4 limits packet loss and provides a very
18、loose upper bound on delay. A single packet error ratio objective is specified for classes 0-4; this value is chosen to ensure that packet loss is the dominant cause of defects presented to upper layers. Y.1541 also defines a “best effort” QoS class (Class 5) with no specific performance guarantees.
19、 Y.1541 assumes that the user and network provider have agreed on a traffic profile that applies to one or more packet flows in a QoS class. At present, the agreeing parties may use whatever capacity specifications they consider appropriate so long as they allow both enforcement and verification. Fo
20、r example, peak bit rate (including lower layer overhead) may be sufficient. When protocols and systems supporting dynamic requests are available, users may negotiate a traffic contract that specifies one or several traffic parameters. ITU-T Recommendation Y.1221 Y.1221 defines the traffic parameter
21、s in the context of three fundamental types of flows IP-based networks can support (dedicated bandwidth, statistical bandwidth, and best effort). The Y.1221 traffic parameters and corresponding UMTS service attributes are discussed below. Table 1 IP QoS class definitions and network performance obje
22、ctives (footnotes omitted) QoS Classes Network Performance Parameter Nature of Network Performance Objective Class 0 Class 1 Class 2 Class 3 Class 4 Class 5 (Un-specified) IPTD Upper bound on the mean IPTD 100 ms 400 ms 100 ms 400 ms 1 s U IPDV Upper bound on the 1-10-3quantile of IPTD minus the min
23、imum IPTD 50 ms 50 ms U U U U IPLR Upper bound on the packet loss probability 1*10-31*10-31*10-31*10-31*10-3U IPER Upper bound 1*10-4U ATIS-0100032 5 Table 2 identifies some typical applications for each Y.1541 QoS class, and some typical node mechanisms and network techniques that could be used to
24、implement them. For example, the node mechanisms could involve separate queues with preferential servicing and different drop priorities, or traffic grooming; the network techniques could involve constrains on routing and distance.5Y.1541 emphasizes that these guidelines are discretionary; network p
25、roviders using the standard may employ whatever node mechanisms, routing constraints, provisioning strategies, or other QoS control techniques they choose. Table 2 Guidance for IP QoS classes QoS Class Applications (Examples) Node Mechanisms Network Techniques 0 Real-Time, Jitter sensitive, high int
26、eraction(VoIP, VTC) Separate Queue with preferential servicing, Traffic grooming Constrained Routing and Distance 1 Real-Time, Jitter sensitive, interactive (VoIP, VTC). Less constrained Routing and Distances 2 Transaction Data, Highly Interactive, (Signaling) Separate Queue, Drop priority Constrain
27、ed Routing and Distance 3 Transaction Data, Interactive Less constrained Routing and Distances 4 Low Loss Only (Short Transactions, Bulk Data, Video Streaming) Long Queue, Drop priority Any route/path 5 Traditional Applications of Default IP Networks Separate Queue (lowest priority) Any route/path T
28、able 3 illustrates the QoS (also called traffic) classes defined for the 3GPP defined universal mobile telecommunications system (UMTS). Four QoS classes are defined in 3GPP Technical Specification 23-107 TS23-107: conversational, streaming, interactive, and background. The main distinguishing facto
29、r among these classes is the delay sensitivity of the traffic. The conversational and streaming classes are intended to be used primarily in carrying real-time traffic flows. The conversational class supports real-time services like video telephony that are particularly delay sensitive. The streamin
30、g class supports one-way flows, and therefore is somewhat less delay sensitive. The interactive and background classes are primarily meant to be used by traditional Internet applications like WWW, e-mail, telnet, FTP, and news. Because they have looser delay requirements than the conversational and
31、streaming classes, they can provide better error rates using channel coding and retransmission. The main difference between the interactive and background classes is that the interactive class is mainly used by interactive applications (e.g. interactive e-mail or interactive Web browsing), while the
32、 background class is meant for background traffic (e.g. background download of e-mail or other files). Responsiveness of the interactive applications is ensured by separating the interactive and background applications. Interactive traffic is intended to have a higher priority in scheduling than bac
33、kground traffic, so that background applications use transmission resources only when interactive applications 5Recommendation Y.1541 notes that there will be very long paths that cannot support Classes 0 and 2; nevertheless, it was considered important to specify (and offer) low delay services wher
34、e feasible. ATIS-0100032 6 do not need them. TS 23-107 notes that such prioritization is particularly important in a wireless environment, where the bandwidth is low compared to fixed networks. Table 3 3GPP UMTS QoS classes Traffic class Conversational class conversational RT Streaming class streami
35、ng RT Interactive class Interactive best effort Background Background best effort Fundamental characteristics - Preserve time relation (variation) between information entities of the stream - Conversational pattern (stringent and low delay ) - Preserve time relation (variation) between information e
36、ntities of the stream - Request response pattern - Preserve payload content - Destination is not expecting the data within a certain time- Preserve payload content Example of the application - Voice - Streaming video - Web browsing - Background download of emailsComparing Tables 1-3, it appears that
37、 Y.1541 classes 0 and 1 correspond generally with the 3GPP conversational and streaming classes, respectively. In each specification regime, the two classes are intended to support real time services and the first class has a more stringent delay requirement than the second. In both regimes delay va
38、riation is intended to be limited.6Similarly, it appears that Y.1541 classes 2-4 correspond generally with the 3GPP interactive class. In both specification regimes, a key application of interest is interactive data. Y.1541 supports this application category with three classes, distinguished by diff
39、erent quantitative delay limits. TS 23-107 states (para. 6.4.3.2): “There is a definite need to differentiate between quality for bearers within the interactive class. One alternative would be to set absolute guarantees on delay, bitrate etc, which however at present seems complex to implement Inste
40、ad, traffic handling priority is used. SDUs of a UMTS bearer with higher traffic handling priority are given priority over SDUs of other bearers within the interactive class, through UMTS-internal scheduling.” Thus TS 23-107 envisions a QoS distinction for interactive traffic similar to that defined
41、 in Y.1541, but specifies a relative QoS mechanism for implementing it.7Y.1541 class 5 corresponds closely with the 3GPP background class. The most fundamental difference between the Y.1541 and TS 23-107 classes is that the former classes specify quantitative performance limits while the latter clas
42、ses, in themselves, do not. Clearly, such limits would need to be specified in wireless/wireline interworking situations to assure particular QoS levels end-to-end. 6The limit is implicit in the requirement to “preserve time relation (variation) between info entities of the stream” in the 3GPP case.
43、 7Another 3GPP specification (TS 29-207) defines six QoS classes by expanding the interactive class into three classes, distinguished by three traffic handling priorities. This specification would align the UMTS and Y.1541 classes more closely, since a UMTS class would be associated with each of Y.1
44、541 classes 2-4 (see Table 6). ATIS-0100032 7 7 MAPPING BETWEEN Y.1541 CLASSES AND 3GPP UMTS SERVICE ATTRIBUTES Although the 3GPP QoS classes do not in themselves provide a basis for QoS interworking with external IP networks, TS 23-107 specifies a related set of “bearer service attributes” that may
45、. TS 23-107 in fact states that QoS will be defined by specifying such attributes. Any particular set of attributes that can be requested by the user is defined as a “QoS profile.” The QoS profile is communicated among UMTS entities to activate QoS mechanisms ensuring provision of the negotiated UMT
46、S service QoS. TS 23-107 further states: “The end-to-end service is provided by translation/mapping with UMTS external services. A Translation Function converts between the internal service primitives for UMTS bearer service control and the various protocols for service control of interfacing extern
47、al networks. The translation includes converting between UMTS bearer service attributes and QoS attributes of the external networks service control protocol.8Thus, the QoS mapping envisioned in TS 23-107 is a translation of bearer service attributes. The attributes specify values for particular perf
48、ormance (and traffic) parameters. If the set of bearer service attributes translated between 3GPP and external systems were comprehensive enough (including, for example, attributes related to error, delay, etc.), it could be possible to map between the 3GPP bearer service attributes and the Y.1541 Q
49、oS classes. TS 23-107 discusses the attributes that can be specified (and mapped between 3GPP and external systems) for each of the four 3GPP QoS (or traffic) classes.9Table 4 summarizes the defined UMTS bearer attributes and their relevance for each traffic class. The bearer service attributes are listed in the column on the left. Three of these attributes describe bearer service performance (as opposed to traffic or functionality): transfer delay, SDU error ratio, and residual bit error ratio. The latter parameter cannot be easily related to the Y.1541
