Network Architectures for Evolving 3G LTE and Mobile WiMAX

Size: px

Start display at page:

Download "Network Architectures for Evolving 3G LTE and Mobile WiMAX"

Aubrey Spencer
5 years ago
Views:

1 Network Architectures for Evolving 3G LTE and Mobile WiMAX All-IP networking is the ultimate goal of 4G mobile networks, but 3G LTE and mobile WiMAX systems have designed semi all-ip network architectures due to their effectiveness in radio resource and mobility management. These semi all-ip networks separate layer 2 and layer 3 handoff operations by grouping many base stations (BSs) as a subnet, thus alleviating the handoff problem, while the pure all-ip networks provide a simple network platform at the cost of high handoff overhead. In this article, we compare the semi all-ip networks to the pure all-ip networks, and provide an overview to WiMAX access service networks and 3G LTE backhaul networks. We then illustrate advanced architectures for QoS provisioning: IP multimedia subsystems (IMS) and policy based management (PBM). Keywords: All-IP, LTE, Mobile WiMAX, Network Architecture, QoS Provisioning, IMS I. INTRODUCTION Fourth-generation (4G) mobile networks are expected to deploy a simplified network architecture based on all-ip [1]. The scenario of all-ip networking will alleviate the problem of third-generation (3G) access networks such as WCDMA and cdma2000, where there are many protocols to cover their complicated backhaul networks. While these 3G networks basically have evolved from a circuitswitched cellular network, 4G networks are expected to become an all-ip based packet-switched system where packets traverse across an access network and a backbone network without any protocol conversion. To set a goal for 4G networks, International Mobile Telecommunications (IMT) has defined IMT-Advanced of which requirements for supported data rates are 100 Mbps and 1 Gbps for high mobility and low mobility, respectively [2]. Alongside this effort, proposals such as IEEE m and 3G LTE (Long-Term Evolution) Advanced are on the table to develop new systems towards 4G networks. These proposals, despite the importance of all-ip networking, may not adopt pure all-ip due to some issues in terms of Radio Resource Management (RRM) and mobility management. The all-ip scenario may enforce each Base Station (BS) to trigger the change of an IP address, when a mobile station (MS) switches its serving BS. In reality, it is known that changing an IP address incurs a too long delay to provide a seamless service for the MS [3]. Therefore, a semi all-ip network is considered, where changing an IP address is not executed within a subnet (i.e., a group of BSs). In this article, we describe the difficulty in deploying all-ip networks for cellular systems by comparing pure all-ip and semi all-ip networks. In addition, we provide an overview of existing network architectures such as mobile WiMAX and 3G networks because there has not been much discussion yet about the network architectures for IEEE m and LTE Advanced. To reduce network complexity, IP networks operate in a way that the intelligence in the network is purposely located in the end nodes. This philosophy of network design helps in simplifying the network architecture at the

2 cost of the difficulty in QoS provisioning. This is not a serious problem to Best-Effort services such as Web page download. However, as many services with enhanced user experience which require real-time delivery of traffic are emerging, QoS support in IP networks has become an important issue. For this reason, we overview in this article the QoS provisioning mechanism for IP networks, especially the IMS which is expected to play a significant role in the next generation networks. II. OVERVIEW OF WIRELESS NETWORK ARCHITECTURES In existing cellular networks, an access network consists of many entities for supporting radio resource management and mobility management. For example, in 2G GSM/GPRS networks, the Base Station Subsystem (BSS) consists of the Base Transceiver System (BTS) that handles the physical layer and the Base Station Controller (BSC) that handles radio resource management and handoff. Also, the Mobile Service Center (MSC) fulfills upper layer functionality and acts as the Visitor Location Register (VLR) that is required to update the location of every MS for paging. Protocols defined in each layer in GSM systems are exhibited in Figure 1, where several protocols are defined for communication between any two entities. 4G networks, in contrast, will make such a complicated protocol stack much simpler, by enabling IP packets to traverse between a Base Station (BS) and a Mobile Station (MS). Each BS may need to perform all the functionalities required in BSS, BSC, and MSC. This makes the BS play a role of an Access Router (AR). This

3 architecture is shown in Figure 2. It incurs high overhead, however, especially when an MS configures a mobile IP (MIP) address for handoff. As it is known that it takes several seconds to run the MIP handoff [3], MIP hinders an MS from carrying out smooth handoff. In addition, the 4G network is expected to have a small cell radius due to use of high frequency band, which possibly results in short cell residence time. For this matter, reducing the latency in performing the MIP handoff is still a challenging issue. For instance, a fast handoff scheme [4] proposes to decrease the address resolution delay by pre-configuration. Another feature of such all-ip networks is their flat architecture. All the radio resource management and mobility management will be performed at each BS independently of the other BSs. Unlike traditional cellular networks of a hierarchical architecture, the flat all-ip network can be operated flexibly but at the cost of complexity in terms of intercell RRM (e.g., coordination among cells). There are increasing demands for intercell RRM for efficient network management; for example, fractional frequency planning for OFDMA wireless networks is needed to improve cell-edge performance. The upper entity such as the BSC in hierarchical cellular networks could be a good coordinator for such a scheme. To alleviate the difficulty in radio resource and mobility management of all-ip cellular networks, a semi (i.e., subnet-based) all-ip cellular network can be considered as shown in Figure 3, an example of a simple network where an AR manages several BSs. The functionality of an AR is separated from that of a BS in order that each undertakes L3 and L2 protocols, respectively. This relation is similar to that between BSC and BTS in GSM networks. Then, an MS moving within the subnet (i.e., changing BSs) performs L2 handoff without changing MIP attachment. The MS only needs to trigger L3 handoff, when it moves into another AR area.

4 Table 1 compares the network architecture of pure and subnet-based all-ip cellular networks. A main difference is that the former is decentralized while the latter is centralized. Since the pure all-ip network incurs a L3 protocol in the end access link, it requires long handoff latency and high signaling overhead. However, the architecture is simple and cost-efficient for implementation. On the other hand, the subnet-based all-ip network implements hierarchical architecture, so it is possible to fulfill efficient resource management in spite of its inflexibility. Both network architectures are being considered in WiMAX and 3G- LTE systems, which are described in the following. The WiMAX standard has defined three different profiles, Profile A, B, and C, for an Access Service Network (ASN) which consists of multiple BSs and an ASN gateway [5]. The relation between a BS and an ASN gateway is also similar to that between a BTS and a BSC in GSM systems. A hierarchical ASN is defined in Profile A and C, whereas a flat ASN is defined in Profile B. Profile A is a hierarchical structure that is similar to traditional cellular networks. As shown in Figure 4, the Radio Resource Controller (RRC) and the Radio Resource Agent (RRA) are implemented at the ASN gateway and the BS, respectively, so most radio resources are managed by the ASN gateway. In Profile B, the functionalities of a BS and an ASN gateway are co-located on the same platform/solution, which makes the architecture flat. That is, R6 defined for the link between an ASN gateway and a BS does not exist. In Profile C, the RRC is implemented at

each BS as shown in Figure 5, so all the RRM functions are performed at each BS as in a flat architecture, although it is still based on a hierarchical structure.

5 each BS as shown in Figure 5, so all the RRM functions are performed at each BS as in a flat architecture, although it is still based on a hierarchical structure. Thus, mobility can be managed by the ASN gateway or other upper entities. Thanks to the flat architecture, Profile C is becoming more popular than Profile A. The 3G LTE standard [6] has defined a simple network architecture of E-UTRAN (Evolved Universal Terrestrial Radio Access Network). The E-UTRAN consists of enbs (evolved Node Bs) which are interconnected with each other by the X2 interface. The User Equipment (UE) is usually served by a single enb. Each enb is connected with a S-GW (Serving Gateway) that terminates the S1 interface between an enb and the Mobility Management Entity (MME). The enb hosts the Physical transmission (PHY), Medium Access Control (MAC), Radio Link Control (RLC), and Packet Data Control Protocol (PDCP) layers that include the functionality of user-plane headercompression and encryption. Also, it offers the functions of RRM and dynamic resource allocation as in other BSs. The S-GW routes and forwards user data packets, while also acting as the mobility anchor for the user plane during inter-enb handovers and as the anchor for mobility between LTE and other 3GPP technologies; it manages and stores UE contexts, such as network internal routing information. It also performs replication of the user traffic in case of lawful interception. The MME is the key control node for the LTE access network. It is responsible for idle mode UE tracking and paging procedure including retransmissions. It is also responsible for authenticating the user by interacting with the Home Subscriber Server (HSS). The architecture of E-UTRAN is shown in Figure 6. It is similar to the Profile C of WiMAX ASN, since most RRM functions are fulfilled by the enb in a flat manner while some mobility functions are fulfilled by the S-GW in a hierarchical manner. 3G LTE Advanced is expected to evolve towards flat All IP architecture eventually. WLAN will form a part of 4G networks too. Still it is a controversial issue how to implement the access network in the IEEE wireless LAN systems. A subnet is composed of an AR and APs, where the hierarchical structure is also similar to a cellular network. Three types of APs are considered according to the role assigned to the AP [7], and they are Fat AP, Thin AP, and Fit AP. The Fat AP provides router-like functions, so there is no backhauling of traffic. This scenario is very close to the all-ip networking. In contrast, the Thin AP is close to the BS in the WiMAX Profile A. The primary role of Thin

6 APs is to receive and transmit wireless traffic, but in this case, a group of APs are managed by a centralized access controller which acts as an ASN gateway in the WiMAX ASN. In the Fit AP architecture, MAC functions are split between the AP and the access controller, so this architecture is compromised between the Fat AP and the Thin AP models. 3. QUALITY-OF-SERVICE PROVISIONING ARCHITECTURES IN ALL-IP NETWORKS Along with the architectural evolution towards all-ip network, one of the most salient trends for future network design is emerging in the form of Fixed Mobile Convergence (FMC). The integration of wireline and wireless technologies and services realized by FMC is expected to offer benefits to both operators and consumers by delivering enhanced user experience over a unified framework. In this section, we survey the QoS provisioning architectures in all-ip networks which are currently being researched in academia and industry. IMS lies at the heart of this network convergence. It is a framework that provides a variety of IP based services. This framework enables wireline, wireless and cable operators to offer rich multimedia services across both legacy circuit switched and new packet switched network infrastructures [8]. IMS was first introduced by 3GPP in Release 5, and thereafter has been standardized by 3GPP and considered as a key aspect for 4G mobile networks. IMS is independent of IP-CAN (IP Connectivity Access Network) which provides the user terminal with IP connectivity towards the IMS core. For the example of GPRS (General Packet Radio Service) in 3GPP, the IMS architecture is shown in Figure 7, but IMS can be employed with any access technologies such as LTE, WiMAX, and DSL (Digital Subscriber Line). CSCFs (Call Session Control Functions) play an essential role in IMS. There are three types of CSCF: P- CSCF (Proxy-CSCF), S-CSCF (Serving-CSCF) and I-

7 CSCF (Interrogating-CSCF). The P-CSCF is the first contact point for an IMS user, and functions as a proxy server for the UE. The connection between the P-CSCF and the UE is established using IPSec (IP Security). Also, the P-CSCF is in charge of SIP (Session Initiation Protocol) compression/decompression which is to reduce signaling overhead. The S-CSCF performs session control and registration services for the UE. During the registration process, it downloads user information and service-related data from the HSS. Based on the subscription information, it determines whether the received SIP request or response needs to be sent to an AS (Application Server) for further processing. In this way, service providers can be separated from the network operator. The I-CSCF is another IMS entity that locates at the edge of an administrative domain. As the name implies, it interrogates the HSS to determine the assigned S-CSCF for the UE. If no S-CSCF is assigned yet, the I-CSCF assigns the UE an S-CSCF based on the UE's capabilities received from the HSS. Also, it publishes its IP address to a DNS (Domain Name Server) as a representative contact point of the whole IMS. The MRFC (Media Resource Function Controller) together with the MRFP (Media Resource Function Processor) handles the bearer-related services such as mixing and transcoding of the incoming media streams. Upon receiving a signaling message from the S-CSCF, the MRFC instructs the MRFP to perform the corresponding operations. The BGCF (Breakout Gateway Control Function) is used for handling a call from the IMS to a circuit-switched network such as PSTN (Public Switched Telephone Network) or PLMN (Public Land Mobile Network). The breakout can be either in the same network or towards another network depending on which network the caller and callee reside in. If both are in the same network, the BGCF selects an MGCF (Media Gateway Control Function) for further handling of the session. Otherwise, it forwards the session to the BGCF of the target network. The MGCF and IM-MGW (IP Multimedia Media Gateway) is a controller-processor pair like the MRFC and MRFP. The MGCF converts a signaling message between the IMS and the circuit-switched domain, while the IM- MGW links the user planes of the two domains. To achieve the goal of network convergence, a large part of the protocols used in IMS is borrowed from the existing Internet standards whenever possible. This includes SIP, SDP (Session Description Protocol), RTP (Real-Time Protocol), and so on. However, this unification framework based on IP supports the Best-Effort nature of IP network for QoS provisioning. This particularly becomes a serious problem for IMS since it is crucial to guarantee end-to-end QoS in order to support the different types of real-time services envisaged in IMS. Flexible QoS establishment needs tight coordination between control and transport plane because different

8 operators in different domains may configure their network elements in different ways to support the required QoS [9]. Policy-based management (PBM) has been adopted to provide this coordination between control and transport plane. The advantage of such architecture is that the resources in transport plane will be reserved according to the QoS parameters indicated in signaling messages. Due to its flexibility and adaptability, 3GPP adopted PBM as a QoS support mechanism for IMS. Figure 8 depicts the entities in PBM and the relation between themselves. A policy is defined by one or more rules which are composed of Condition and Action. It is stored in the Policy Repository (PR) and managed through the Policy Management Tool (PMT). The Policy Decision Point (PDP) is responsible for making decisions for policy requests sent by the Policy Enforcement Point (PEP). PEP takes actions based on the decision made by PDP. Figure 9 illustrates how PBM is applied to IMS at session establishment stage [10]. In this example, UE A which is an IMS-registered device tries to establish a session with UE B which is also IMS-registered (we explain the process using GPRS terminology). Initially, UE A sends a SIP INVITE message which in turn traverses the IMS core (i.e., P-CSCF, I-CSCF, S-CSCF) and arrives at UE B. This INVITE message contains SDP media description (e.g., supported media codecs, bandwidth requirement) in its body. Upon receiving the message, UE B answers with SESSION PROGRESS message which contains the list of media codecs supported at this end. Until now, the required resource is not

9 reserved at both sides since it is still not clear which media codec(s) will be used for the session. When UE A receives the SESSION PROGRESS message, it selects an adequate codec in the list and returns a PRACK (provisional acknowledgement) message to UE B. And, it starts the resource reservation process on its side. That is, it maps the chosen SDP information to GPRS QoS parameters, and sends a PDP (Packet Data Protocol) context activation request to the GGSN. Then, the GGSN in which the PEP functionality resides asks the PDF (Policy Decision Function, 3GPP term for Policy Decision Point) to authorize the request. As a response to the authorization request, the PDF informs the GGSN of the decision result containing IP QoS parameters related to the PDP context. Finally, the GGSN translates the IP QoS parameters to GPRS QoS parameters, and compares with those conveyed in the PDP context activation request from UE A. If they are met, the PDP context activation request will be accepted. Likewise, UE B starts its own resource reservation process upon receiving the PRACK message. When the resource reservation process is completed and a UPDATE message is received, UE B will start alerting the user (i.e., ringing). In order to enable end-to-end QoS provisioning, not only the operator's network but also the external IP networks the session goes through should have QoS support mechanisms. IETF (Internet Engineering Task Force) has been working on standardizing the related specifications for years. IntServ [11] and DiffServ [12] are the two earliest works which specify fine-grained and coarse-grained QoS systems, respectively. In IntServ systems, an application that requires some kind of QoS guarantees has to make an individual reservation using RSVP (Resource Reservation Protocol). However, it can only work on a small-scale network due to the difficulty for keeping track of all the reservations. Contrary to IntServ which is a flow-based mechanism, DiffServ supports QoS provisioning through a class-based mechanism. Edge routers in a DiffServ system are responsible for classifying packets into several classes and marking the classification code in the IP header. Then, core routers relay the packets to the next-hop router based on the marked code at the edge routers, its own scheduling and queueing policy. Due to its scalability and low management complexity, it has received a lot of research attention recently especially in the form of DiffServ-aware MPLS TE (Multi-Protocol Label Switching Traffic Engineering). While the abovementioned mechanisms deal with only intra-domain QoS support, end-to-end QoS provisioning requires the resource reservation in all the domains along the path. SLA (Service Level Agreement) negotiations among different network providers can be made for this purpose. An SLA is a formal agreement between two or more entities that is reached after a negotiating activity with the scope to assess service characteristics, responsibilities and priorities of every part [13]. Using these intra- and inter-domain QoS support mechanisms altogether, the required level of service quality can be provided to users in different administrative domains. IV. CONCLUSIONS This article has considered the network architectures evolving into 4G mobile networks. Eventually the architecture will follow the all-ip scenario, but to tackle the current problem of mobility management in a given time constraint, it is expected to have a form of semi all- IP. We discussed the basic architectures of all-ip and semi all-ip, and briefly overviewed the architectures of 3G LTE and mobile WiMAX. Additionally, we presented some core components of QoS supporting architectures that would be considered in 4G networks. [1] ITU-R, ''Framework and Overall Objectives of the Future Development of IMT-2000 and Systems Beyond IMT-2000,'' Recommendation ITU-R M. 1645, [2] ITU-R, ''Principles for the Process of Development of IMT-Advanced,'' Resolution ITU-R 57, [3] H. Yokota, A. Idoue, T. Hasegawa, and T. Kato, ''Link Layer Assisted Mobile IP Fast Handoff Method over Wireless LAN Networks,'' in Proc. ACM MOBICOM, Sep. 2002, pp [4] R. Koodli, ''Fast Handovers for Mobile IPv6,'' draft-ietfmipshop-fast-mipv6-03.txt, [5] WiMAX Forum, ''WiMAX Forum Network Architecture,'' Release 1, version 1.2, [6] 3GPP Release 8, ''Overview of 3GPP Release 8,'' V0.0.4, [7] T. Sridhar, ''Wireless LAN Switches -- Functions and Deployment,'' The Internet Protocol Journal (Cisco), Vol. 9, No. 3, 2006, pp [8] M. Poikselka, A. Niemi, H. Khartabil, and G. Mayer, ''The IMS: IP Multimedia Concepts and Services in the Mobile Domain,'' WILEY, [9] M. Mani and N. Crespi, ''Inter-Domain QoS Control

Mechanism in IMS based Horizontally Converged Networks,'' in Proc. IEEE International Conference on Networking and Services, 2007, pp. 82-86. [10] M.

Shenker, ''Integrated Services in the Internet Architecture: an Overview,'' RFC 1633, 1994. [12] S. Blake, D. Black, M. Carlson, E. Davies, Z.

10 Mechanism in IMS based Horizontally Converged Networks,'' in Proc. IEEE International Conference on Networking and Services, 2007, pp [10] M. Sauter, ''Beyond 3G: Bringing networks, terminals, and the web together,'' WILEY, 2009, pp [11] R. Braden, D. Clark, and S. Shenker, ''Integrated Services in the Internet Architecture: an Overview,'' RFC 1633, [12] S. Blake, D. Black, M. Carlson, E. Davies, Z. Wang, and W. Weiss, ''An Architecture for Differentiated Service,'' RFC 2475, [13] ITU-T, ''Framework of a service level agreement,'' Recommendation ITU-T E. 860, 2002.

DAY 2. HSPA Systems Architecture and Protocols

DAY 2. HSPA Systems Architecture and Protocols DAY 2 HSPA Systems Architecture and Protocols 1 LTE Basic Reference Model UE: User Equipment S-GW: Serving Gateway P-GW: PDN Gateway MME : Mobility Management Entity enb: evolved Node B HSS: Home Subscriber