INTERNET DRAFT G. Fodor Document: draft-fodor-reqts-cellular-netwks-00.txt F. Persson Expires: Aug 2002 L. Westberg B. Williams J. Wiorek Ericsson Feb 2002 NSIS QoS Signaling Requirements from a Multi-access Wireless Perspective Status of this Memo This document is an Internet-Draft and is subject to all provisions of Section 10 of RFC2026. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/1id-abstracts.html The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html ABSTRACT We consider access technology agnostic applications that use IP level QoS primitives to specify their QoS requirements. In future wireless systems (beyond what is often referred to as 3G), where mobile nodes may access an IP backbone over diverse access networks, the IP level QoS primitives are not only used to specify the requested IP level service, but also (through an access specific translation function) to help the mobile node's wireless link manager to configure the wireless bearer service. In such a scenario, the IP level QoS primitives must meet special requirements in order to support the spectrum efficient management of the wireless resources. In this draft we discuss the main characteristics of wireless access networks and derive the requirements on IP level QoS primitives from a wireless, and especially from a cellular perspective. 1 Background and Motivation Fodor, Persson, Westberg, Wiorek [Page 1] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks One of the key objectives in the evolution and standardization of future wireless access technologies beyond 3G (such as the IMT-2000) systems, is to offer QoS support for a variety of services, including IP services, while considering the critical aspect of optimizing spectrum efficiency. Applications that make use of such services include voice over IP, streaming services over IP and other QoS enabled IP applications that access services of a QoS enabled IP network over wireless accesses. Such future wireless systems are expected to offer services over multiple access technologies in such a manner that applications need not be aware of the specific link layer(s) that is (are) available at any given time. For instance, a device may have a combination of different interfaces such as wired and wireless LAN, and a cellular wireless access. From a user perspective, it is a key requirement that the one application can operate effectively over any of these accesses. For cellular wireless accesses, operating effectively includes consideration of spectrum efficiency (which affects cost/performance). Clearly, such applications need to be able to specify their end-to- end QoS requirements in an access agnostic manner. That is, we assume that each access system can have its own translation function, that takes the IP level QoS parameters as input and generates the access specific parameters from those (see also [1] for further details on the role of such translation functionality and [2] for a suitable set of such IP level parameters). Thus, applications only need to communicate their QoS requirements through IP layer primitives (an API) rather than through an access specific primitives. Using IP level primitives towards diverse accesses necessitates the need to investigate the requirements on IP level QoS primitives from a wireless and specifically from a cellular perspective. Cellular systems are generally characterized by the fact that their wireless resources are scarce and costly. In an end-to-end path containing one or more wireless accesses, it is expected that the wireless links (providing transport over the air interface) will be the most critical ones (_bottlenecks_) for QoS delivery. Thus, the end-to-end service will be mainly influenced by the suitability of the provided wireless link characteristics. One of the challenges in the design of QoS services and the associated enabling protocols in the multi-access environment is that they must be able to be supported effectively by a variety of link layers and QoS mechanisms, including those used by wireless access networks. It is believed that platforms supporting multiple layer 2 (L2) interface types will provide generic (non-interface specific) QoS primitives to applications. Since IP based QoS primitives are expected to be widely available to application developers on some of these platforms (e.g. laptops), it is useful to ensure that spectrum efficient radio services can be provided for applications requesting Fodor, Persson, Westberg, Wiorek [Page 2] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks QoS through such QoS primitives. Spectrum efficiency is a key factor in enabling cellular operators to deploy affordable services. In this draft we focus on applications/hosts requesting QoS through such primitives, and list the requirements on the IP level QoS signaling protocol from the perspective of such applications/hosts. The draft is organized as follows. Section 2 describes the cellular architecture and the main functionalities of the wireless network elements. Note that this model is generic in the sense that it can be applicable to a variety of wireless technologies. Section 3 builds on the discussion of Section 2 and lists these requirements. 2 Wireless Network Architecture and Characteristics We consider an architecture where mobile nodes (MN) access an IP network via a cellular access network. We naturally assume that the IP layer is present in the mobile node such that the user may establish an IP connection to other IP endpoints through the wireless access and the IP network. In such a network, a typical scenario includes the following elements: o The mobile node (MN) is considered to include the physical device connecting to the Wireless Access Network (WAN) and the IP level resource manager and signaling entity that allows applications to request QoS enabled IP bearer services. The IP level resource manager in the MN can provide the wireless link (air interface) manager with QoS related information through a technology specific translator entity. The wireless link provides the transport service over the air interface between the MN and the base station (which is part of the radio network of Figure 1). o The Wireless Access Network (WAN) that consists of base stations (BS), base station controllers BSC, (also referred to as radio network controllers, RNC) and possibly other nodes responsible for mobility management, location management, etc. The WAN connects to the external IP Network (e.g. The Internet) through (a) gateway node(s) (WAN GW). It is important to note that the WAN appears to the MN as a L2 network; and is designed and optimized for the transmission of radio packets. Fodor, Persson, Westberg, Wiorek [Page 3] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks A simplified protocol stack of such a scenario is shown by Figure 1. +----+ +----+ |Appl|<----+ |Appl|<----+ | | V | | V +----+ +------+ +----+ +------+ |TCP/| |IP QoS| |TCP/| |IP QoS| |UDP | |Module| |UDP | |Module| +----+-+------+ +-----+ +----+-+------+ | IP |<---------------->| IP |<-------->| IP | +-------------+ ,---. +--+--+ +--+--+ ,---. +-------------+ | | | | | |IP|<>|IP| | | | | | | Radio L2 |<------->|R2+--+ +--+L2|<-------->+ L2 + | | | | | |T2|<>|T2| | | | | | +-------------+ | | +--+--+ +--+--+ | | +-------------+ | Radio L1 | \ / |R1|T1|<>|T1|L1| \ / | L1 | +-------------+ `-' +--+--+ +--+--+ `-' +-------------+ MN Radio RNC/ WAN IP Remote (mobile Network BSC GW Net User node) \ / -----v----- WAN Figure 1 Simplified end-to-end scenario with wireless access (WAN) The radio packet delivery service (associated with a specific set of traffic and QoS characteristics) that is provided by the WAN is often referred to as the radio access bearer service (RABS). For instance, a streaming RABS may mean a radio packet delivery service provided by the WAN that provides bounded delay and limited packet loss ratio. Typically, the terminal equipment has the responsibility for identifying the radio access bearers that it needs, and how it will use them. Thus, it is responsible for initiating the radio bearers between the MN and the WAN GW. Since the applications may use access technology agnostic QoS primitives (ie the applications do not have a direct interface towards the wireless resource manager), the MN must make use of the service information signaled at the IP level in determining the appropriate radio bearers to establish. The WAN provides RAB services in order to support the layer 2 connection between the MN and the WAN GW. The characteristics of these RAB services are dependent on the wireless mechanisms (see [1] for a detailed discussion on such mechanisms), and can be markedly different from bearer services in traditional wired networks. It is clear that different RAB services (in terms of the provided delay, bit error rate, etc.) can be provided, resulting in quite different characteristics of QoS, service costs and service behaviors. A consequence of this flexibility is that sufficient Fodor, Persson, Westberg, Wiorek [Page 4] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks detail about the applications' traffic and service requirements must be known in order to determine the appropriate parameter settings to enable service optimization. 3 QoS Requirements Imposed by the Wireless Link The wireless link provides the RAB service for IP packets over the air interface. The main characteristics of the wireless link may be summarized as follows: C1: Its resources (frequency, time slots, power, CDMA code, _wireless bandwidth_ etc.) are generally scarce (and therefore expensive). Moreover, some of its resources may change in time due to mobility or interference from other wireless users. C2: It imposes delay on the bits that are transmitted over it. This delay is typically orders of magnitude longer than that caused by fixed (e.g. optical) transmission links. Part of this delay is lower bounded by physical characteristics of radio wave propagation and is not possible to control by resource management techniques. C3: It causes bit-errors typically with orders of magnitude higher probability than that caused by fixed (e.g. optical) transmission links. It is possible to reduce the bit error probability associated with the wireless link to a _reasonably low_ level at the expense of consuming more wireless resources (e.g. power). Because of these main characteristics, the resources of the wireless link in most wireless technologies (and in most cellular systems) are carefully managed (as opposed to over-provisioning). More precisely, the objective of the wireless link resource management is twofold: O1. To provide the required QoS for the entity (e.g. an IP application running on a mobile node) that requests the wireless transport service over the air interface. In the case of an IP application, this transport service means the transport of the bits of the IP packets. The quality of this wireless transport service is typically characterized by quantifying C2 and C3 above. O2. To make as efficient use of the wireless resources as possible (because of C1 above). In order to meet these two objectives, wireless resource management protocols typically manage resources dynamically at a low level in the sense that they control the use of the wireless resources at a time scale on the order of the radio frames, which can be much shorter than the packet length. For instance, if the quality of a wireless link degrades because of increasing interference, the resource management technique may increase the transmission power over the wireless link. Fodor, Persson, Westberg, Wiorek [Page 5] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks From these two objectives (O1-O2) it follows that the radio resource management requires information about the required QoS and the characteristics of the traffic that is transported over the wireless link. These two pieces of information are required to exercise admission control and to allocate the necessary resources. In a scenario, where the wireless link provides the transport service for applications running over IP it is a challenging task to provide these two pieces of information, because the application may not be aware that it runs over a wireless link. Therefore, in such a scenario, the QoS and resource reservation mechanisms at the IP layer (e.g. resource reservation and/or signaling protocols, and the IP level bearer service, such as the set of integrated services) need to provide support for the resource management of the wireless link (i.e. the R2/R1 layers below IP in Figure 1). From this perspective, the basic requirements on an IP level signaling protocol and an IP level bearer service include support for the following: R1: Fine grained QoS and traffic characterization per user IP-flow There must be sufficient information available at the IP layer[1], (including parameters that characterize the required QoS and the actual traffic), which (through the access specific translation function) allows the wireless resource management to derive the required QoS parameters and obtain as precise information about the offered traffic as possible. These bearer parameters need to support diverse wireless access technologies. To this end [2] proposes an extension of the controlled load integrated service that meets this requirement. R2: QoS differentiation within a user IP-flow and/or within a user IP-packet Since the wireless resource management often operates on the bit- level, the wireless resource management can improve the provided QoS and the resource efficiency if it has knowledge about the structure of the IP packet payload. For instance, for some voice applications (codecs) some bits in the IP payload may be more important for the human ear than other bits, in which case bit- specific bit error rate is provided for these different voice sample bits within the packet. This technique is often referred to as unequal error protection (UEP) [3]. Another technique that may be used over the wireless link uses different drop levels for packets within the same user flow. This technique is similar to the unequal error protection but operates at the packet level. It can for instance allow a voice application (codec) in the mobile node to drop packets before sending over the wireless link and thereby to save wireless resources. Fodor, Persson, Westberg, Wiorek [Page 6] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks R3: Bi-directional asymmetric communication It is expected that the up-link (i.e. from mobile node toward the network) and the down-link (i.e. from the network towards the mobile node) can be asymmetric in terms of the required resources (e.g. allocated bandwidth). In such a case the wireless link must be configured accordingly in the respective directions, such that the necessary QoS and resource efficiency objectives (O1 and O2 above) are met in both directions. Therefore, there has to be support for asymmetric bearers that allow bi-directional communication between two mobile nodes (or between a mobile node and a server) over an IP network. R4: Local (as opposed to end-to-end) resource reservation The cellular network may provide access to a variety of IP networks, including the ones that provide best-effort differentiated or integrated services. Therefore, in some cases it is important that the resource reservation should apply within the access network only rather than end-to-end. Therefore, the QoS solution should provide support for non end-to-end resource reservations. R5: Multi-cast Multi-casting is an efficient way to save wireless link resources. A prime example on this is a scenario where multiple mobile nodes access a streaming server. In such a situation the wireless gateway node may broadcast the streaming data to the participating mobile nodes rather than using dedicated communication channels. Multi-casting (as opposed _multi-unicasting_) allows some of the wireless link resources to be re-used between the mobile nodes. Therefore, there needs to be a way to create and modify multi-cast trees of mobile nodes and wireless nodes. R6: Local Control In a typical wireless scenario where the mobile node (client) requires a service from a server (such as in a streaming or WWW services), it is the mobile node that is aware of the wireless link conditions including the available wireless bandwidth. Also, in such scenarios the wireless node is typically the charged party for both the up-link and down-link service. Therefore, it is important that the wireless node has some mechanism that allows it to control the bandwidth of the traffic that is transmitted over the wireless link for both directions. This control of the QoS service from the local access allows the wireless node to make efficient use of the wireless resources, and at the same time to control network charges. Fodor, Persson, Westberg, Wiorek [Page 7] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks R7: Rate Adaptivity Because the resources of the wireless link may fluctuate in time (see C1), the utilization of the wireless resources may be increased by allowing the mobile (receiver) node to dynamically control the sending rate of IP packets. In particular, the application running on the mobile node needs to be able to increase/decrease the sending rate of the streaming server depending on the available wireless bandwidth. This requirement is especially important in a situation where the mobile node executes a hand-over from a high bit-rate access network to a low bit-rate access network, where the available bit-rates may be different by several orders of magnitude. Therefore, in this sense, there has to be support for adaptivity that allows the wireless resource management to make best use of the (currently) available wireless resources. 4 Conclusions This draft considered applications that use IP level QoS primitives to communicate their QoS requests towards diverse access network types. These types of scenarios are expected to be important in future wireless systems, where users on the move access services of an IP backbone network over multiple accesses, including wireless and wired ones. A key requirement on such systems is that applications should be able to request QoS in an access agnostic manner. This is important because the one user application must be able to work over different interface types (eg high speed lan, dial-up modem, wireless access). Therefore, we have throughout assumed that applications use IP level QoS primitives to express their QoS requirements and different link layers use a link layer specific translation function to derive the access specific QoS parameters. We have also argued that in the case of wireless and especially cellular accesses, the end-to-end QoS is mainly influenced by the appropriate configuration of the wireless link. The wireless link configuration determines the resource efficiency and thereby it has a direct impact on the cost of the services that are offered by cellular operators. The set of requirements for an access agnostic IP level QoS primitives which shall control the QoS signaling at both the IP and access layer which is suitable for a cellular network is thus driven by the QoS requirements of the wireless link. These requirements include the support for: R1: fine granularity QoS parameters per user IP flow R2: QoS differentiation within a user flow and within a user packet R3: bi-directional asymmetric communication R4: local (as opposed to end-to-end) reservations R5: multi-casting Fodor, Persson, Westberg, Wiorek [Page 8] INTERNET DRAFT Requirements on QoS mechanisms Expires Aug 2002 over IP based cellular networks R6: local control R7: rate adaptivity. References [1] G. Fodor, F. Persson, B. Williams, _Application of Integrated Services over Wireless Accesses_ draft-fodor-intserv-wireless-issues- 01.txt, work in progress, January 2002. [2] G. Fodor, F. Persson, B. Williams, _Proposal on New Service Parameters (Wireless Hints)_ in the Controlled Load Integrated Service_, draft-fodor-intserv-wireless-params-01.txt, work in progress, January 2002. [3] Sjoberg, J, Westerlund, M, Lakaniemi, A, Xie, Q, _RTP Payload Format and file storage format for AMR and AMR-WB audio_ draft-ietf-avt-rtp-amr-11.txt, work in progress Fodor, Persson, Westberg, Wiorek [Page 9]