An Integration Architecture to 4TH Generation Wireless ... - IEEE Xplore

3 downloads 22324 Views 192KB Size Report
implementing the 4G network to provide the Always Best. Connected services. This paper analysis the current communication business model, and get a ...
An Integration Architecture to 4TH Generation Wireless Networks Xu Yang

John Bigham

MPI-QMUL Information Systems Research Centre, Macau Polytechnec Institute

Department of Electronic Engineering , Queen Mary University of London

Macau SAR, P. R. China

London, U.K.

E-mail: xu.yang @mpi-qmul.org

E-mail: [email protected]

Abstract: Integration architecture will play a key role in implementing the 4G network to provide the Always Best Connected services. This paper analysis the current communication business model, and get a conclusion that the Mobile Virtual Network Operator (MVNO) can be a good candidate to integrate the multiple radio access networks. The integration can be realized step by step to adapt the user’s gradually increased service requirements. We analysis the characters of the MVNO, summarize the required management functions and propose an integration architecture that allows the MVNO to support real time resource management and seamless service handover in the heterogeneous wireless networks. The integration architecture, which is a variation of hybrid network model, requires the least modification of the current access networks.

1. INTRODUCTION 4G networks are envisioned as the integration of different existing wireless network technologies (e.g. WPAN, WLAN, WMAN, MANET, 2G, 2.5G, 3G, satellite, DAB, DVB-T) and future wireless network technologies in an optimum way, in order to ensure seamless handovers from one technology to another, thus continuously providing the available best service to the user anywhere, anytime, anyhow. [1] The Always Best Connected concept allows a person connectivity to applications using the devices and access technologies that best suit his or her needs. This concept refers to being not only always connected, but also being connected in the best possible way, combining, for instance, the worldwide coverage of cellular systems with the high bandwidth of WLAN hotspots. [2] To deploy 4G and providing the ABC services, various issues related to the network, the terminal, and the user need to be resolved [1, 2], such as the user terminal needs to be capable of communicating with different wireless access network technologies, and wireless network selection, Authentication Authorization and Accounting (AAA), Mobility Support, User Profile handling, Security, and QoS guarantee for end to end services. Additionally, network architectures will play a key role in implementing the features required to address these issues. There are several architectures integrating multiple different radio access networks (RAN)[3].

Tunnelled networks: Different access network technologies work independently. Based on some policy, the optimal network for the requested service is selected. The Integration layer tunnels the traffic across the Internet and the selected access network to the mobile host. This system requires no modification to existing access networks, however the services latency increment because of functionality duplication and lack of integration in the lower layers. Hybrid Network. In this model there is a hybrid core that interfaces directly between the RANs and the Internet. The RAN implement the network layer and below. Advantages are that in this model there will be less duplicate functions. Heterogeneous Network. In this model the integration core layer deals with all network functionality and operates as a single network to the upper layers. Different RANs handles only those functions specifically related to a distinct radio access technology. A major challenge of this model in that the different RANs should converge, which requires a standardization effort and business commitment to support it. The advantage of Heterogeneous Network is obvious: this solution can meet all the technical requirements mentioned above, and give a complete scheme to integrate the different network technologies. All the network technologies can work highly cooperatively and complement each other. Additionally this architecture is very flexible and open, capable of supporting various types of networks, terminals and applications. One example is as MIRAI [3]. However the disadvantages from the business view cannot be ignored. Firstly, implementation of the Integration Core Layer is difficult and may cost too much because of the complexity of standardization for different network technologies. Secondly this revolution cannot be achieved step-by-step adapting to the users’ gradually increasing requirements. This may involve the risks of investment failure. Thirdly, this architecture may decrease or remove the competition between these operators. In this case, high QoS service with lower prices may not be guaranteed. In this paper we present a different integration architecture that can avoid these disadvantages. The article is organized as follows. First, it is an introduction on the current

communication business model to explain that the MVNO can be a good candidate to integrate the different RANs. Then we analysis the management functions required by the MVNOs to manage the heterogeneous networks, and present the integration architecture, which is a variation of hybrid network, using agent technology. A work procedure for setup connection is to show the interactions between the agents. Finally we give a conclusion. 2. THE CURRENT COMMUNICATION BUSINESS MODEL Before considering the integration, we need to look at the current business model of the commercial wireless networks, as any kind of integration should be built above the basis of the business model. In current commercial communication networks, there are three main business entities: Mobile Virtual Network Operators (MVNOs), Mobile Phone Operators (MPOs), and users. MVNOs mediate between endusers and different MPOs (Figure 1). MVNOs Offer Services to end to end users User

User

SLA Network Elements

Access Points

IP BackBone

IP BackBone

Interconnection Network Operator 1 UMTS

Network Operator 2 WLAN

Figure 1: Relationship between MVNOs and MPOs A MVNO is a company that does not own any frequency spectrum, but resells wireless services under their own brand name, using the network of another MPO.[4] A MVNO has Service Level Agreements (SLA) with end users, and it guarantees the Quality of Service (QoS) for its customers. For the end users, the use of different network technologies is transparent. On the other hand, the MVNO does not own any GSM, CDMA or other wireless infrastructure, such as Mobile Switching Center (MSC) or a radio access network. Some may own their own Home Location Register (HLR), which allows more flexibility since multiple host networks could be used, and the MVNO appears as a roaming partner. [4]Additionally, it also has a SLA with each MPO, and both of them are responsible for meeting the agreement . It is the MPOs that manage the transmission through the IP or non-IP based backbone; an interconnection agreement may be established between different MPOs to ensure connectivity between networks. There are currently approximately 200 planned or operational MVNOs world-wide.

In this business model MVNOs offer at least three advantages to MPOs. First, it allows MPOs to capitalise on spare radio capacity while incurring little extra cost. Particularly if MVNOs attract customers that the MPOs do not normally have access to. For example, Virgin Mobile in the U.K. uses the Virgin Brand to attract the youth segment. Second, they enable operators to devolve marketing, sales, billing, customer relations and related "front" and "back” office functions to specialist companies that may provide those functions more efficiently than can the operators themselves [5]. Additionally, well-diversified independent MVNOs can offer a product mix that normal mobile operators cannot match. Due to the role of MVNO, it can be a good candidate to integrate the different network technologies in a common platform in an optimum way in the future 4G networks. Firstly, the MVNOs can interact with different MPOs in order to offer a wider choice of services and applications at a lower price to end users in a large geographic area, thus they can ensure a more efficient use of the spectrum. Secondly under the operation of MVNOs, different MPOs not only compete with each other to carry users’ transmissions but also work cooperatively and complement each other. Thirdly, with the gradually increasing requirements of users for high quality multimedia services, the MVNOs can make the best choice on when and how to do integration and adapt users’ requirements, which can minimise the possibility of failure of investment. Additionally, the user sign one contract and get services from many different network technologies with one billing system and the MVNO can profile its customers to provide Personalized Services in selecting specific services from a diversity of mobile service offerings and adjusting selected services to their individual needs. Later section will analysis the management functions, and present novel integration architecture with purpose to provides a virtual common network management platform for the MVNOs to manage the different network technologies. 3. MANAGEMENT FUNCTION To respond to the user’s requirements, a MVNO needs to communicate with user’s terminal wherever the user is located. On the other hand, the MVNO cannot manage the real traffic flows across different network technologies; it only monitors the parameters of QoS to select a network technology which has the most appropriate QoS, and perform CAC function to distribute the users’ requests to available and appropriate network elements. Additionally, it cannot manage and control the QoS of transmissions inside each network, which means it cannot perform the routing functions. Furthermore, the only thing a MVNO can do to affect QoS is to perform degrades and upgrades of the requested bandwidth of transmissions it manages. Therefore the IP backbone integration can’t be managed by the

MVNOs to guarantee the QoS issues. It is the MVNOs lead the integration of the different access network technologies; and the MPOs will decide when and how to integrate the backbones and have the responsibility to guarantee the QoS. The management functions of MVNO are defined as follows: Authentication Authorization and Accounting (AAA), security, call Admission control (CAC), radio resource management and handover management across and within the heterogeneous networks, monitoring and predicting QoS for each technology, configuration and fault management, and finally users’ profile management. To fulfil these functions, the MVNO needs not only to perform real time processing such as to control traffic flow, but also needs to monitor and predict the QoS for each network technologies in order to guarantee QoS for users. 4.INTEGRATION ARCHITECTURE To accomplish the virtual common management platform with the above management functions, the management framework for the MVNO must be flexible, scalable, provide interoperability and attain high performance and by making good resource management decisions quickly. To achieve this, a distributed processing approach is required rather than the traditional centralized approach. Intelligent agents have been used in the network management of telecommunications system to address the problems of distributed and decentralised resource control and management [6].

In a whole MVNO domain, there are several CSNs. They could back up each other, and exchange information to create a view of the whole network management. Each CSN can communicate with hundreds of GRA nodes. A CSN contains a Configuration Server and a QoS Server. These two agents provide management information that is shared by the agents in different GRA nodes; the Configuration Server performs configuration and fault management, and manages the users’ database and profile. The QoS GRA monitor the QoS for individual MPOs, and assist the Local GRA nodes to do the CAC. The Local GRA node is logically located in each network node (such as a base station, WiFi access point), but normally does not reside physically at the network node. For example, for a cellular network it could be physically collocated with the RNC. The Local GRA node performs four management functions with four local GRAs. 1.

Setup GRA performs CAC and network selection. Normally two Setup GRAs control one transmission, one is represent the caller, and another one is for the receiver. Setup GRA gets the technical QoS parameters for the different network technologies from Control GRA. If the selected network element cannot carry the user’s request, Setup GRA may inform Resource Management GRA to perform resource management function to release some bandwidth.

2.

Handover GRA performs seamless horizontal and vertical handover. Handover GRA monitors all the live connections’ QoS, if observed QoS decrease to unbearable level, it can initiate and perform horizontal and vertical handover.

3.

Resource Management GRA (RM GRA) manages bandwidth resource and executes QoS degrades and upgrades to ensure important communications are given as much quality of service they can be offered.

4.

Control GRA manages the GRA node. It has several management functions: firstly it performs the fault management of individual GRA in a Node, if any other GRAs are broken and cannot work, the Control GRA can create a new instance for it; secondly Control GRA has a database for all the alive connections’ information, which can be used by the Setup GRA, Handover GRA and RM GRA; thirdly Control GRA also perform security functions of prioritized traffic (it can give extra encryption and decryption function on the selected calls to make them safe even when processed in the backbone.

4.1 Agent Architecture Our Integration Architecture is a deviation of Hybrid Network [7], like Figure 2. The management architecture is divided into two levels: a set of Central Server Nodes (CSNs) and local Gateway Resource Agent nodes (GRA nodes). The MVNOs operates an intranet, attached to the Internet backbone that hosts the execution of systems. Central Server

Central Server

Internet

GRA node GRA node

T NRA

T NRA

GRA node T NRA

Multi air-interface user devices are supported

User terminal with UA

Figure 2 Integration architecture

There is also a User Agent logically executes in a user’s mobile. It has four responsibilities: firstly, to intelligently select a Setup GRA to handle the user’s request based on predefined rules; secondly, to provide security functions to the user’s transmissions; thirdly, it can initiate the vertical or horizontal handovers with negotiation with the handover

GRA; finally user can update its profile through the User Agent for access network operator selection and service preference. In this architecture, the Network Operator Agent (NRA) is used to represent the MPO. Each network element node logically has one Transmission NRA (T NRA) and one QoS NRA. The Transmission NRA is responsible for responding to set-up, upgrade, and degrade requests from the GRAs. The QoS NRA provides information on the QoS performance to the QoS GRA . 4.2 Agents Communication Each GRA or NRA has a different IP address, and they could communicate with each other through the IP core network wherever their locations are. The current wireless data networks (e.g. GPRS) and future 3G and 4G networks are backboned by all-IP networks, IPv6 based backbone network is capable of advanced traffic engineering abilities such as prioritising certain traffic. Communications between MVNO elements are most important traffics for MVNOs thus they are highest prioritised for most reliability and least latency. Carefully designed communication messages and protocols are also ensuring only necessary data will be transferred to ensure fastest response between those MVNO elements. Compares to agent platform approach. The main aim of the IP based architecture is to achieve faster and more reliable communications. Agent communications can be implemented either by agent platform (e.g. FIPA) or self-designed protocols over IP. These two approaches both have advantages and disadvantages: •

By agent platform: all communications are implemented by Agent Communication Language (ACL) messages, which are semantic level languages, ACL messages are delivered from agent to agent by a protocol stack: namely, Agent Communication Chanel (ACC), Object Request Broker (ORB), Internet Inter-ORB Protocol (IIOP) and then TCP/IP. Communications implemented over agent platform are easy to design, maintain but slow due to several protocol stacks over IP, redundancy headers are added to messages and extra processing power is needed. [7]



Directly implemented over IP: this approach needs a full set of communication protocol designed directly over IP. Because the messages are transported over network layer, carefully designed protocols (minimum header, more efficient handshakes etc.) can guarantee minimum message transport time, which is crucial to call setup, hand off procedures. IPv6 are further engineered by QoS so that agent communications over IPv6 are categorised by highest priority class to guarantee most reliable and fast delivery. 4.3 Agents Physical Location

To carry this agent architecture into execution, we need to consider where we could put GRA Nodes and NRAs in the different networks, because the location will affect the communication speed between agents. As the core backbone network in 3G and GPRS is an IP based network, the central server can be concentrated in a single server with different IP addresses. GRA nodes and NRAs are software elements in the radio system, for the sake of fast reaction times, GRA nodes and NRAs should be located on the same sub-net as their corresponding network element. A User Agent can exchange information with a Setup GRA physically through the control channel, such as paging channels in GSM. The following text gives a proposed configuration of GRA node and NRAs in different radio access networks. •

The configuration for WLAN systems: NRAs logically correspond to local access points. The NRAs also need to link to an AAA (authentication, authorization, and accounting) server to retrieve user information. Except for the link between 802.11 stations and 802.11 Access Point (AP), all other connections are usually connected by wired local area network. Thus the software GRA node and NRAs can be physically located in a single server and representing each 802.11 AP’s coverage area. The NRAs can also be physically resided in the control unit of 802.11 AP as embedded software. In this situation the communication between GRA Nodes is relayed by the wired local area network.



The configuration for GPRA or UMTS: Since GPRS and UMTS based systems are converging to common core network architecture, only the radio access network differs: the UTRAN in the case of UMTS and GREAN for GPRS. Hence some of the terms can be reflected with each other. The Transmission NRA and QoS NRA will be closely linked to the RNC (BSC in GPRS or GSM network) because the RRC (Radio Resource Control) and RRM (Radio Resource Management) functions reside in the RNC, not within each Node B. The NRAs are also linked with HSS (home subscriber server) or HLR (home location register) in order to retrieve user information. The location of GRA node should be same as the NRAs.



For satellite systems, NRAs and GRA node can be located in the ground control station that has access to the satellite control unit and user registration database.

4.4 An example: work procedure for handling a new setup request As an example of interaction between the agents, a Setup Procedure is described in Figure 3: 1.

A UA (User Agent) chooses a reachable Setup GRA and sends it a request for a connection. Such information

exchange is made physically through the mobile’s control channel; 2.

The Setup GRA extracts the User Priority class from the request message, reads the user’s personal profile from the central server, and authenticates the user.

3.

If the user is an authenticated user, then the Central Server, which can locate the receiver with the help from the different MPOs, will active one reachable Setup GRA of the receiver;

4.

The Central Server inform the caller’s Setup GRA the IP address of receiver’s Setup GRA;

5.

Then the Setup GRAs for both of the caller and receiver perform Network Selection, which is to choose access network elements to physically carry the connection over its radio network. The selected access network elements maybe belong to different MPOs, and should have plenty of bandwidth resource to carry the request network service. The Network Selection is according to the predefined rules or the current QoS parameters for the different MPOs.

6.

The Setup GRA of the receiver informs the Setup GRA of the caller that the address of the selected network element for the receiver.

7.

The Setup GRA of the caller relays the user’s request and the address of the receiver’s selected network element to the Transmission NRA of its chosen MPO. The Transmission NRA then informs the selected MPO to setup and route the request call.

8.

When the call has been physically setup, the T NRA then inform the Setup GRA of the caller;

9.

The Setup GRA of the caller then informs the Setup GRA of the receiver that the call has been setup.

User Agent

Setup GRA

Central Server

T NRA

Setup GRA

User Agent

1. a setup request 2. authenticate the user 3. locate the receiver and its setup GRA 4. receiver’s IP address 5.network selection

5.network selection

6. receiver’s selected accesses network element

7. ask selected MPO to setup and route the call 8. inform that the call is setup 9. inform that the call is setup

Figure 3 Agent Interactions for setup a connection When a new connection is set up, Setup GRA will send the new connection’s information to the RM GRA. If there is no enough bandwidth to carry a request service, the Setup GRA may inform RM GRA to consider a QoS degrade.

The agent architecture and basic functionalities have been implemented in a computer simulator to demo the agents’ interactions. 5. CONCLUSION In current commercial wireless networks the MVNO mediate between end users and MPOs, the MVNO can be a good candidate to integrate the different network technologies in a flexible and expandable platform. Because the MVNO doesn’t own any existing frequency spectrum, it can lead the integration of different access network elements, and the MPOs have the responsibility to perform the IP backbone integration. We analysis the character of the MVNO, summarize the required management functions, and propose an agent based integration architecture, which is a variation of hybrid network model, that allows the MVNO to support real time resource management and seamless service handover in the heterogeneous wireless networks. The advantages of our approach are: 1. The MVNOs can select any kind of MPO to start integration, more MPOs can be added to the architecture step by step to adapt the users’ gradually increased services requirements, which can avoid the investment failure; 2. It only requires least modification on the current network technologies towards 4G. REFERENCE [1] I.Ganchev, M.O’Droma, H. Chaouchi, I. Armuelles, M. Siebert, N. Houssos, Requirements for an Integrated System and Service 4G Architecture, 2004 IEEE. [2] Eva Gustafsson and Annika Jonsson, “Always Best Connected”, Erisson Research, IEEE Wireless Communications, 10(1):49-55, February 2003. [3] Gang Wu and Mitsuhiko Mizuno, Paul J.M.Havinga, “MIRAI Architecture for Heterogeneous Network”, IEEE Communications Magazine, February 2002. Issue [4] http://www.shosteck.com/news/aug03.htm, #70 August, 2003,published by The Shosteck Group, http://www.shosteck.com [5] http://en.wikipedia.org/wiki/Mobile_Virtual_Network_Opera tor; [6] A.L.G.Hayzelden & J.Bigham, 1999, “Agent Technology in Communication Systems: An Overview”, KERJ, J4:4. [7] Suguri,H “A standardization effort for agent technologies: The Foundation for Intelligent Physical Agents and its activities” System Sciences, 1999. HICSS-32. Proceedings of the 32nd Annual Hawaii International Conference on Volume Track8, 5-8 Jan. 1999 Page(s):10 pp.