Horizontal Layering –

an Essential Aspect in Modern Networking

Denis Duka, Lovre Hribar and Damir Buric

Ericsson Nikola Tesla

Poljicka cesta 39c, Split, Croatia

Phone: +385 21 434820 Fax: +385 21 434834 E-mail:

Abstract - Coupled with the growing interest in the third generation mobile communication system as a standard for future mobile communications, the need for a set of functions to effectively support advanced telecommunication services in such an environment is also increasing.

The separation of the network functionality into independent layers is a key principle in modern networking in the tele- and datacom industry today. The layered thinking is also a very fundamental and visible aspect in a number of standardization initiatives and industry forums, such as the Multiservices Switching Forum, led by several of the largest operators and manufacturers. This papers presents the short overview of horizontal network model as well as the migration strategies toward the new network architecture.

I.INTRODUCTION

Today's tele- and data communications environment consists of a variety of networks. Most of these networks are highly specialized, designed and optimized to serve a specific purpose as illustrated in Figure 1 (left part). To a large extent these networks can also be described as vertically integrated in the sense that they combine very different functionality (for example, transport, control, services and so on) in one and the same network element.

For historical reasons the networks in this vertically integrated multi-network world have evolved independently of one another and therefore differ in many ways. This, of course, reduces the possibilities for operators to create synergies among their networks as a means to reduce costs, provide service portability and so on. The architecture in the right part of Figure 1 illustrates a way to evolve this multi-network situation in order to overcome some of its deficiencies. The solution is based on a horizontal structure of the network into a number of independent layers, which gives a more flexible system. The open architecture of the UMTS network ensures a smooth migration from the existing systems to the new technologies[1].

Figure 1: Vertically versus horizontally integrated networks

II.THE LAYERED NETWORK ARCHITECTURE

The UMTS network is divided into three layers:

  • Connectivity layer,
  • Control layer,
  • Application layer.

The Core network includes the control and connectivity layer, which are separated and on top of these two layers there is a Service Network, the application layer (see Figure 2).

The connectivity layer handles the transport and manipulation of user and control data. Manipulation includes coding/decoding of the user plane and protocol conversion in the control plane. This layer comprises transport backbone elements and Media Gateways.

The control layer (seen in the middle Figure 2) is where the service intelligence resides. The service intelligence is unique and specific to each service type. The nodes found in the control layer are generically referred to as control servers.

The top layer is the application layer. Applications in the application layer are as generic as possible, enabling their use for all types of services - these applications would typically be Internet applications, Intelligent Network (IN) applications and so on.

Figure 2: The architecture split

All services will use the same transport network, which forms the connectivity layer. Here, everything that has to do with the transport and manipulation of user data is handled. The control layer provides control for specific services, for example call control and session management. In the service layer, the generic applications that may be used by all types of services are located [2].

Special Media Gateways, controlled by specific network servers, adapt and connect different access types to a common backbone network (see Figure 2).

Key benefits of the layered architecture

The 3rd generation mobile communication system is designed in a way which offers a number of advantages and possibilities. Special attention has been put on meeting requirements for time to market, low cost of ownership, openness and future evolution towards an “all-IP” solution.

The key characteristics are the following:

  • Offers an open and versatile architecture capable of meeting the current and future demands in a fast changing telecommunications environment. Specifically the layered architecture with its transport flexibility constitutes the perfect platform for evolving the network towards an “all-IP” solution.
  • Provides an inherent flexibility for coping with growth and/or changing traffic patterns and traffic mixes (circuit and packet)
  • Independence between layers allows each layer to evolve independently, for example, as a result of advances in different technology areas
  • Provides great transport flexibility and allows different transport technologies, both existing and new ones, to be deployed without impacting the control or services/application layers
  • Allows common transport arrangements for multiservice networks, that is, several service networks sharing the same transport network
  • Allows access independent and seamless services through a common service/application layer
  • Provides efficient use of network resources, that is:

– Placing the codec on the edge of the network. This will result in a much more efficient utilization of transmission resources:

– Centralizing devices, that is, forming larger pools of devices

– Sharing user plane resources between the packet and circuit mode communication services

  • Relies on proven and stable protocols and design
  • Allows a very flexible re-use of investments in the GSM infrastructure [3].

III.ADOPTING THE HORIZONTAL LAYERING IN THE GSM/UMTS CORE NETWORK – GENERAL PRINCIPLES

The layered Core Network architecture is derived from the current standards reference model by separating the control plane functions in the MSC from their user plane functions, thus turning these nodes into Servers and Media Gateways as illustrated in Figure 2.

The existing MSCs and GSNs will retain their current 2G roles but they can be integrated into a combined 2G/3G infrastructure in different ways in order to optimize the overall network solution. Such an optimization would target different networking aspects and could include:

  • Provision of full 3G functionality in the existing 2G nodes
  • Integration of 2G nodes into a common 2G/3G transport infrastructure
  • Support for 2G/3G handover.

It is possible to apply these ‘integration levels’ differently in different nodes (for example, only some nodes need to be upgraded to support 3G functionality). Certain modifications (hardware and/or software) of the 2G equipment may be necessary depending on the level of integration required.

A migration package is developed for the installed 2G MSC and GSN nodes which will include the functionality necessary to operate these MSCs and GSNs as full fledged 3G nodes, to handle inter-system handover and to support common transport network solutions.

The migration to UMTS will normally be done in two steps for a GSM operator. The first step involves a logical separation of the functions in the existing 2G MSC nodes (the internal structure of today’s 2G products is already prepared for this), into a Server (control) and a Media Gateway (connectivity) part. Both the functions are still integrated into one node, which is based on the AXE platform. New software and hardware is needed in the node. The server part will then in fact be identical to the server in the new UMTS product line. This creates great synergies and allows an updated MSC to function both as a Server and a Media Gateway for UMTS, or to act as one fully integrated network entity in the existing GSM network. In the second step, a new hardware is installed in a stand-alone MGW and it is controlled by the server function in the upgraded 2G node (or by a new standalone server). This is called a physical separation and the MGW is based on the CPP platform (see Figure 3).

All the devices that manipulate the media stream are placed in the MGW and the server parts of the MSC now form the MSC server [4].

Figure 3: The migration steps

IV. CORE NETWORK ELEMENT OVERVIEW

The main innovation in the core network is a split of call control and user plane handling, which leads to many new requirements for the traditional MSC HW platform. The traditional MSC will be divided into a control node, named MSC server and a Media Gateway (MGW) node, which provides bearer adaptation for the user plane from the PLMN core network to external networks like PSTN, ATM or the Internet. The MGW will also provide the bearer adaptation/conversion for the signaling plane. To realize this concept a migration of the existing network based on AXE 10 and a new hardware based on CPP must be developed.

MSC/VLR server

The MSC/VLR Server (see Figure 4) is an MSC/VLR with only signaling connections to other network nodes.

Figure 4: AXE10 MSC server and MGW

The MSC/VLR Server is based on AXE10 technology and shall be seen as a further development of the present GSM MSC/VLR. The MSC/VLR Server controls the user plane in a remote or collocated Media Gateway via the GCP protocol. The MSC/VLR Server is able to use any remote Media Gateway in the network for a specific call. The Media Gateway used for a specific call is typically the one where the user plane enters the PLMN (mobile terminated call) or where it leaves the PLMN (based on B-number, mobile originated call). For PLMN internal calls the Media Gateway is normally selected in order to minimize the transmission path and it could be based on a roaming number (B-subscriber location). For special purposes other selection mechanisms may be used, such as selecting a specific Media Gateway when certain specialized HW is needed, for example a modem pool for data calls.

The MSC/VLR Server terminates the RANAP and BSSAP signaling. Call control between the MSC/VLR Server and a Transit Server or a GMSC Server is done with BICC signaling. The BICC signaling is routed to a Transit Server and the user plane to the Media Gateway associated with the Transit Server. The MSC/VLR Server performs charging of calls and other transactions.

The MSC/VLR Server is independent of the bearer used for the user plane in the network. The bearer used for the signaling links would normally be the same as used for the user plane. This is however not strictly necessary, as Signaling Gateways can be used for converting between different bearers.

Media Gateway functionality within MSC/MGW

The Media Gateway function (shown on Figure 5) operates on the user plane in order to enable interworking between different transport domains. The MSC Server controls the Media Gateway function.

Figure 5: MGW function within MSC/MGW

The Media Gateway function has two functional layers:

  • Media Streamprovides the devices needed to manipulate the user data according to the circuit-based service being used
  • Transportprovides the switching of the user plane and bearer control for interworking between different transport domains.

CPP MGW

Figure 6: CPP MGW

The CPP MGW (shown on Figure 6) can contain a full set of speech and data resources for performing manipulation and additions to the connectivity layer. It also contains transport resources for performing protocol and connectivity layer conversions between different networks and it provides Signaling Gateway functionality for performing conversions of lower layer control protocols. An incoming connection on a physical line interface with a standardized bearer protocol is connected to the appropriate function. On the outgoing side, the connection is connected to an outgoing standardized bearer. Therefore, an incoming bearer is switched to an outgoing bearer even if the stream is modified and the bearers are changed. As part of this process, conversion between different bearers and formats can be made. This conversion can, for example, mean converting a compressed voice to a non-compressed format and changing the bearer from ATM to STM, or terminating the packet data traffic received from the Gn interface in the GTP

tunnels and re-tunneling the IP packets into IPsec or L2TP tunnels towards external IP networks (Gi interface). All MGW application functions during a connection are initiated from the Gateway Control Protocol that provides a direct link to the controlling MSC server(s). The platform provides the physical line interfaces and most of the bearer layer interfaces including bearer control signaling, for example, AAL2 with Q.AAL2.

For packet based traffic functions such as tunneling (GTP, IPSec and L2TP), re-tunneling (of GTP packets to IPSec encapsulation and vice versa), translation of IP address to IMSI address, volume based charging, security functions QoS handling is performed in the Media Gateway. The external control interface is the Gateway Control Protocolused by the MSC Server to request the Media Gateway to add and remove media stream functions into a speech and data connection. The Gateway Control Protocol also includes commands to establish and release a through connection with the requested media stream functions included. In the process to establish and release a connection to other network nodes the relevant bearer signaling is included. A Media Gateway can lend its resources to any MSC Server and an MSC Server can use the resources of any Media Gateway. That way UMTS Core Network architecture allows an m:n relation between Servers and Media Gateways.

V.CONCLUSION

The layered architecture is being deployed in third-generation mobile networks – that is universal mobile telecommunication system (UMTS). This enhanced protocol architecture, today generally pursued by most standardization forums, provides an inherent flexibility, which allows operators to build scalable and cost effective multi-services solutions in the new telecom world. At the same time the layered architecture offers a pragmatic way to rationalize legacy networks, allowing them to run over state-of-the- art, cost effective transport solutions.

The modular nature of MGW built on CPP platform means it is possible to create nodes with different configuration, functionality, capacity, cost, reliability and performance level. MGW based on CPP represents the base for supporting the multimedia services associated with UMTS making the convergence of telecom and data in mobile communications possible.

ABBREVIATIONS

3G – Third Generation (of Mobile Telephony)

3GPP – Third Generation Partnership Project

AAL – ATM Adaptation Layer

ATM – Asynchronous Transfer Mode

BICC – Bearer Independent Call Control

BSSAP – Base Station System Application Protocol

CPP – Connectivity Packet Platform

GCP – Gateway Control Protocol

GGSN – Gateway GPRS Support Node

GPRS – General Packet Radio Service

GSM – Global System for Mobile Communications

GTP – GPRS Tunneling Protocol

IP – Internet Protocol

MGW – Media Gateway

MSC – Mobile Switching Centre

PLMN – PublicLand Mobile Network

PSTN – Public Switched Telephone Network

QoS – Quality of Service

RANAP –Radio Access Network Applicat. Protocol

SGSN – Serving GPRS Support Node

STM – Synchronous Transfer Mode

TCP – Transport Control Protocol

UMTS – Universal Mobile Telecomm. System

VLR – Visitor Location Register

WAP – Wireless Application Protocol

REFERENCES

[1] J.S. DaSilva, D. Ikonomou and H. Erben, European R & D programs on third-generation mobile communication systems, IEEE Personal Commun. (February 1997)

[2] Gene Robinson, Communication Networkswith Layered Architecture, IEEE802 N-West Standards meeting for Broadband Wireless Access Systems (March 1999)

[3] D. Medhi, S. Jain, T. Srinivasa Rao, D. Shenoy, M. Saddi, andF. Summa. A Network Management Framework for Multi-Layered Network Survivability. Technical Report, Computer Science Telecommunications, University of Missouri–Kansas City (July 1999)

[4] Fyro, Heikkinen, Petersen and Wiss: Media gateway for mobile networks, Ericsson Review No.4,2000, pp. 216-223.