Challenges for Wireless Networks and OSI - The Eye Archive/v8n1pg50.pdfChallenges for Wireless Networks and OSI ... network layer orks physical layer Figure 1. The OSI reference model

Challenges for Wireless Networks and OSI

The integration of mobile data and multimedia services poses the need for a suitable protocol architecture. This paper describes the dlfferences between existing networks and the special problems to be solved for Open Systems Interconnection (OSI).

0. Spaniol A. Fasbender S. Hoff J. Kaltwasser J. Kassubek Aachen University of Technology Aachen, Germany

W ireless telecommunication services have become a daily routine in both the business and the private sector. Mobile networks and

services form an expanding field with a further boost being expected for Europe because of the deregulation of the European telecommunications market. Key issues of this process will be the introduction of mobile data and multimedia services and the development of suitable network and protocol architectures.

The integration of applications with heterogeneous communication requirements forces the merging of different communication worlds like Public Land Mobile Networks, Packet Mobile Radio and Wireless Local Area Networks (WLANs). This paper focuses on the problems emerging with the development of a suitable protocol architecture. We will describe the differences between existing networks and the special problems to be solved on the transport layer of the Open Systems Interconnection (OSI) reference model.

Wireless networking and the support of multimedia services are the two most important trend technologies in computer systems. The market introduction of notebook and palmtop computers has brought forth a

50 APPLIED MICROWAVE & WIRELESS WINTER 1996

new dimension of computer networking, emerging from the increasing acceptance of personal computing in the 1980s and the growing importance of local and wide area networks for industrial as well as private use in the 1990s

Employees in office environments, accustomed to having access to computational resources like file systems or information services anytime, today expect to carry these resources with them at the same level of service. However, distributing multimedia inforniation to the user's fingertips anytime and anywhere leads to stringent requirements on network resources and portable technology, and, as we will see, to application processes as well.

Future mobile data and multimedia applications will cover a vast field ranging from mobile extensions of today's fixed applications (mobile computing) to future wireless office scenarios including wireless multimedia services with a Quality-of-Service (QoS) comparable to fixed networks. However, the demand for mobile data services will also arise from totally different fields of applications, like telematics services for Road Transport Informatics (RTI) or Teleaction services for palmtop computers allowing applications like Teleshopping and Telebanking to go mobile.

The fusion of the telecommunication world of cellular systems and the world of Internations Organization for Standardization's (ISO's) Reference Model for Open Systems Interconnection (OSI R M ) will be of outstanding interest for the today's evolution of wireless networks and their applications. Intenvorking with and relation to the IS0 standards are important topics of current standardization efforts for communication and distributed systems. As the OSI reference model and its corresponding standards have proven their capability to realize the idea of open systems interconnection in real life, they have gained increasing importance as an abstract framework of how a communication protocol stack should be layered and designed.

increasingly makes use of OSI concepts and even OSI standards, the overall system architecture of telecommunication networks still does not comply with the OSI reference model in some crucial features.

In this paper we investigate the relationship between the rapidly growing and converging field of wireless technologies and the world of open systems interconnection. We will point out the role of the OSI reference model in different types of existing wireless networks as well as for future developments. The discussion of the role of the OSI protocol layers in wireless environments will show that wireless communication requires new protocol architectures, which differ from the classical OSI RM, and that especially the transport layer becomes a bottleneck (see Figure 1 ) . We discuss different schemes of QoS management, like proxy architectures, which are potential solutions for wireless systems.

multimedia

application layer

presentation layer

session layer

transport layer

network layer

orks

physical layer

Figure 1. The OSI reference model and its bottleneck.

Application-Oriented Layers

However, telecommunication especially is dominated up to now, mobile data services are offered in various by existing, mature systems like the Telephone Network ('STN), which

Switched existing networlcs: In GSM (global system for mobile communication) circuit switched data services and a grown for

decades with little or no relation to OSI. The importance and the enormous market potential of worldwide telecommunication systems as well as their inherent

that even new systems l i k e the Integrated Services Digital Network (ISDN) must be built

is a major requirement. Although subsystem design

packet-oriented short-message service are provided. The former extend the services available for fixed te~ecommunication networks (such as ISDN services like fax) for the mobile end user, while the latter is basically a paging service with the additional capability of mobile

medium-bandwidth traffic.

On top Of these systems, and to keep interoperability originated short messages. Both are not suited for bursty

52 APPLIED MICROWAVE &WIRELESS WINTER 1996

Wide area data networks like Mobile Data Communications (MODACOM) and Mobitex (trade- mark of Erricsson) reflect the need for packet-switched long haul communication, but are restricted to low- bandwidth applications like electronic mail or status reporting for fleet management. Wireless LANs provide a first step on the evolutionary path towards supporting medium- to high-bandwidth data services to mobile terminals. However, the supported user speeds are limited to a few miles per hour and mobility is restricted to private environments. Clearly, existing mobile data services are not sufficient to realize the mobile office scenario.

Maintaining reachability and a user’s desktop environment, such as connections to print and file servers, while in motion requires the integration of typical data applications known from wired networks into a mobile environment. Scenarios like remote login, remote operation and electronic mail require only slight modifications to handle problems not known in fixed networks. Short breaks in network connectivity due to the unreliability of the wireless media or gaps in the wireless coverage demand for an adaptation of transport and application time-out values to avoid frequent crashes of application processes. This so-called operational transparency may also be supported by a more stringent realization of OSI concepts like association and session management. An unsolved problem in this context is how an application may distinguish between a temporary and a permanent loss of network connectivity.

The scenarios listed above are transparent to the different performance characteristics of wireless and wired networks. Traffic is generated according to an irregular, packet-based and usually low-bandwidth pattern. A first example violating these characteristics is file transfer. Users are accustomed to a certain level of service when retrieving files from their local server and will expect to access data at comparable speeds when or even while moving to a different location. Another aspect is storage capacity and power limitations of the usually small and battery-powered portables, with demand for highly sophisticated power consumption and access time minimization algorithms. Possible solutions to these problems are addressed in lower layers.

An important aspect of future mobile communication systems will be the need to integrate multimedia applications like WWWW (Wireless World Wide Web) on the one hand and a number of different network architectures such as GSM, DECT (Digital European Cordless Telecommunications), WLANs and so forth

(see next section) on the other. This will engender demand for new application service elements able to supply information about potential links and their service quality at the user interface, to enable a user-initiated switching to the fas tedmost reliable/cheapest link available.

Also, it will be helpful for applications to determine actual location information (such as the nearest printer) in order to address a suitable server without explicit knowledge of the corresponding network address. This requires service directory concepts as momentarily proposed for Open Distributed Processing (ODP,[ 11). For WWWW, similar problems as for file transfer have to be addressed, which most likely will lead to distributed rather than centralized information and file systems[2].

This means bringing the information to the user, a concept commonly known to be addressed by Personal Communication Networks (PCN), where addresses refer to users and not terminals. Bandwidth consuming applications like video conferencing or joint editing additionally create demand for the fulfillment of real- time constraints. In the case of a short interrupt of network connectivity mechanisms like disconnected operation must be employed.

Another aspect is the presentation, for instance, of a video service on a personal digital assistant, which will require adaptation to rapidly changing bandwidth constraints within OSI layer 6 (such as switching from video to pure audio service when moving from an office high capacity wireless LAN into a GSM cell).

From the above discussion it becomes evident that application-oriented layers must be made mobility-aware and mobility related services must be provided by a mobile application programmer interface (API): Global network connectivity integrating various wireless bearers and different fixed backbones requires knowledge about potentially accessible resources also in higher layer protocols, to perform for example the mapping of applicationhser requirements on available networks and their current quality of service characteristics.

QoS handling is also a matter of growing interest in fixed networks like Asynchronous Transfer Mode (ATMs), and demands for network-dependent transfer protocols rather than one network-independent transport layer. Supporting user mobility transparent to application oriented layers (for example, as proposed for Mobile IP,[5]) contradicts the requirements of operational as well as performance transparency for application processes.

APPLIED MICROWAVE & WIRELESS WINTER 1996 55

outband signaling via a separate signaling network. The GSM protocol architecture for signaling consists of the protocols for the mobile station, the BSS, the MSC and the Signaling Network.

The higher layer protocols of GSM can be grouped into a third layer, This should not be confused with the OSI network layer, which is responsible for abstraction from physical network characteristics and for internetworking. The GSM layer 3 also includes functionality’s of higher OSI layers and OSI management, like connection management , subscriber identification and authentication

Figure 3. GXM system structure.

1) Mobile Station Protocol

As there is no complete physical layer recommendation for the OSI reference model, there is no severe restriction for that layer and conformance is not a crucial issue. At the data link layer the radio interface of the mobile station uses a LAPDm protocol, which is a modified version of the Link Access Procedure for the D channel (LAPD) protocol standardized by International Telecommunication Union (ITU) for the (ISDN) D- Channel. These protocols were developed from the I S 0 high level data link control (HDLC) protocoi. Conformance can be achieved by a simple protocol mapping.

2) Base Station Subsystem protocols

At the interface between Base Station and Mobile Switching Center, the lower layers are realized by the Message Transfer Part (MTP) of Signaling System No. 7 (SS7), which is not OSI compatible. It covers functionality’s of layers 1 and 2, and part of layer 3 of the OSI reference model. The MTP itself is layered into three so-called “levels”. The two lower levels can be mapped directly to the corresponding OSI Layers, and Level 3 covers the lower part of the OSI network layer.

The missing functionality of the higher part of the network layer is provided by the Signaling Connection Control Part (SCCP). The Base Station Subsystem Application Part (BSSAP) serves primarily as a bridge between the Radio Resource Management (RR) and the


In order to realize broadcast services in wireless environments, access points must provide forwarding functionality [4]. Moreover, the comparably high security level of wired media due to their closed nature does not hold for wireless systems. To enable identification of a communication partner, link level based authentication mechanisms must be employed. To guarantee privacy of data transmissions between authenticated instances, MAC level station invoked encryption procedures are mandatory. An optional implementation of these services within trusted environments may reduce the protocol overhead.

Wireless Local Area Networks (WLANs) aim at providing high speed data services within a relatively small area, and are directed towards supporting off-line rather than on-line and in-house rather than outdoors mobility. The increasing importance o f mobile communications, supported by a number of upcoming standards (for example, the European initiative HIgh PErformance Radio LAN (HIPERLAN) and DECT, the WLAN pendant in the telecommunication world), will most likely result in a wider deployment of WLANs.

However, it may be asked, whether the stringent OSI functional layering is not outdated considering the increasing importance of the (vertical) OSI management s tack and the splitting (or even redundant implementation) of services like authentication across multiple OSI layers.

Upon successful registration, the mobile host’s current address (address of the serving agent) is bounded to its home address. This enables the redirection of packets destined to the station’s home address to its actual location. Hence, routing of IP datagrams to mobile stations can be performed transparently to the transport layer and does not require modifications in existing Internet routers which do not provide mobility functions. However, this operational transparency is achieved at

the price of only suboptimal routing paths, since every message is first routed to the home network of the mobile host. Distributing the location information to specific location servers within the network may partially compensate this “triangle routing” effect.

Fansport Layer

The main service of the OSI network layer, as is in all communication networks, is the provision of an end-to- end communication service. Depending on the service quality and the properties of the communicating end- systems an additional protection of the end-to-end connection is required. This is the task of the transport layer. We will first introduce the problem of transport protocol design in wireless environments in the GSM case. Comparing the situation to WLAN based private networks, we will find that the transport layer in the classic OSI notion is the real bottleneck requiring new communication principles.

Public Wireless Data: The GSM Case Mobile IP

Providing global mobile connectivity requires additional functionality at the network layer. Existing routing protocols (for example, IP or ISO’s CLNP) utilize an hierarchical addressing scheme, with a station’s home address implicitly identifying its location and thus the endpoint of routing chains to this host. Any host migration will therefore result in a loss of network service, requiring the acquisition of a new network address, followed by numerous manual reconfigurations of name servers and local applications.

Within the Internet Society, the Mobile IP[5] group investigates on protocol enhancements necessary to route IP messages to hosts with varying points of network attachment. The proposed mechanisms are based on location registration and packet redirection. A mobile node changing its location must register with a local facility (such as an 802.11 access point), which then handshakes with the station’s home agent responsible for the location tracking.

Besides speech services, which are so far the most important and requested ones, an increasing number of data services are offered by GSM[14]. They can be divided into two classes: data services which are extensions of fixed network services, like circuit switched synchronous and asynchronous data, and services which are GSM specific, like the Short Message Service (SMS). The limited bandwidth does not permit bit rates above 9600 bit/s with acceptable error rates for circuit switched data services and for Packet Switched Public Data Network (PSPDN) access.

For these services the transparent mode establishes a pure circuit inside GSM without any means for retransmission. To increase quality for circuit switched date, the packet oriented Radio Link Protocol (RLP) was introduced. The peer entities for the RLP are the Terminal Adaptation Function (TAF) on the mobile station side (for data conversions from the terminal equipment to the Mobile Termination) and the Interworking Function (IWF), responsible for data


cornersions from the "GSM world" to the destination ne tnork . as shown in Figure 6. The end-to-end connection betn een terminals is usually protected by a transport protocol, especially in the transparent mode. In an OSI oriented environment additional protocols for session control, local syntax conversion and specific application support may be added (for example, to establish a Virtual Terminal connection according to IS0 9040 \ ia GSM).

a packed switched data service with a capacity around the raw data rate of 12 kbitis on a traffic channel, and will support broadcast, multicast and point-to-point transmissions. In contrast to the circuit switched data services the GPRS will not permanently allocate spare resources at the radio interface and on the GSM fixed network, reducing the waste of bandwidth in the case of bursty communication.

Mobile Station

...

I V.24

Fixed Station

OSI Transport Protocol

X.25 Network GSM

4

Radio Link Protocol - - I - -.

Mob i I e Term i n at i o n Generic GSM Transmission

Figure 6. OSI datu coriiriiuriicatiorz rising CS.M arid X25.

The Shoi-t Message Service (SMS) enables the point- to-point-transport of short messages between mobile stations and the Short Message Service Center (SM-SC), and it includes a cell broadcast service for the traiisniissioii of information with local relevance, such as traffic information. The SS7, currently the only element of the GSM infrastructure capable of packet switching, is used for the transport of short messages. The SMS capacity is limited to 640 bit/s with a maximal message size of 140 bytes. It was not designed as a general purpose packet oriented data service, but rather intended to be an extended paging service.

Since new telematics applications, especially from the area of Road Transport Informatics (RTI), like dynamic route guidance. and freight & fleet management, require higher capacities and a periodic transmission of small (or medium) volumes of data, the General Packet Radio Service (GPRS) will be introduced[l5]. This service is

In most cases GSM is used as a transfer network, the purpose for which it was intended. This means that the communication partner is usually located in a different network. Thus, two or even more networks are transparently passed.

The difficulty for the transport layer is to guarantee a certain QoS, which requires at least the knowledge of the characteristics of all intermediate networks. In the best case QoS can actively be negotiated between transport layer instances and the network. Once the GSM link has been established. the transport protocol in a mobile environment must react on the rapidly changing signal quality on the radio link and the resulting deviation of the packet delays between the transport layer peer entities.

This is a major problem, because in GSM the packet delay of the RLP ranges from a fraction of a second up to tens of seconds. This causes problems for transport protocols like ISO's TP4: Due to long packet delays, the transport layer may trigger packet retransmission even if the packets are just delayed, not lost. Another effect of

68 .APPLIED MICROWAVE & WIRELESS WINTER 1996

the packet delay deviation is an undesired side effect of the congestion control policy, whereby a short interruption of the radio link results in long response times[6]. In consequence, transport protocols like TP4 react too late, long after the error burst on the radio link has occurred.

The proxy agent is responsible for the balance between the errcineous low capacityconnectionon the mobile side and the high capacity connection on the fixed side. It will try to keep the fixed transport connection alive even if the mobile transport connection indicates a temporarily broken communication path, enabling intelligent caching algorithms and disconnected operation.

Network Access Network Access

Figure 7. Proxy architectuw for mobile environments.

Private Wireless Data: The Internet Case

To some extent the problem can be reduced to the long signaling distance between the transport layer instances. The solution is quite simple and intuitive: The single end-to-end transport connection is divided in two or more hop-by-hop connections. This results in an additional logical network on transport level as shown in Figure 7, wherein the transport connection is split up into two parts: the mobile transport part and the fixed transport part, linked together via a proxy agent. Instead of the network layer, now the transport layer is the lowest layer of the stack providing an end-to-end communication service to upper layers.

The mobile transport part can now react dynamically on the bursty behavior of the radio link and on the deviation of packet delays. In that way, the adaptive behavior of the transport protocol can be optimized for a specific mobile network, independent from the fixed network characteristics. Additionally, the reaction times of the protocol entities (for example, to determine retransmission timer values) are reduced.

Although principally related to the problems discussed in the last section, wireless communication in private networks (such as campus environments) leads to a different approach. In contrast to public mobile telecommunication networks like GSM, mobile internetting is not targeted primarily at providing global reachability anywhere and anytime, but more towards handling user off-line mobility (nomadic computing). A typical application will be a temporary move of a terminal to another location (for example, conference or hotel), where it reconnects to the Internet through a wireless or even wired network interface.

Hence, maintaining reachability is mainly a user- initiated task rather than a network service provided by a complex mobility management. It may be compared to the concept of Universal Personal Telecommunications (UPT,[ 12]), which will be introduced in fixed telecommunication networks (ISDN) but also in PCN. In UPT, the entities addressed are users instead of terminals, and the user can register at any terminal in the network by entering his Personal Identification Number (PIN). Subsequently, all calls will be routed to that particular terminal.

APPLIED MICROWAVE & WIRELESS WINTER 1996 7 1

It is revealing that the mobility oriented telecommunication service that is most closely related to the notion of nomadic computing originally was designed for fixed networks! A well known example from today's office environments is an E-Mail service, wherein a user connects to his mail server from time to time at different terminals, and requests delivery of his messages. In the same way, he will connect to a mobility server and, for instance, start a file transfer or remote login. Once a network connection is initiated, the server will be informed about his current point of network attachment and thus also be able to route incoming messages to the mobile station (see Mobile IP section). Handover is restricted to private environments and usually performed at the medium access control sublayer (for example, DFWMAC).

Reliable end-to-end communication between Internet applications is provided by transmission control protocol (TCP). If messages are lost on their way to the receiving peer entity, TCP retransmits them after a time-out and increases the time-out value. In wireless environments short breaks in network connectivity will occur more often than i n wired networks, leading to frequent adaptations of the back-off time. A s a result the retransmission window increases towards its maximum and thus limits the connection throughput[6].

Therefore, supporting station mobility transparent to the transport layer as proposed for Mobile IP or 802.1 1 leads to similar problems as with GSM, even though completely different communication principles are applied. An end-to-end connection between a mobile and a fixed terminal can be separated logically into wireless and wired part with totally different and in the first case highly dynamic link characteristics. I-TCP (Indirect TCP, [7]) is a protocol suggestion which retlects the obvious need to make the transport level aware of host migration.

I-TCP enables a mobile station to initiate a transport layer connection with a host in a fixed network via a special server, in Internet terminology usually called Mobile Support Router (MSR). The MSR establishes a regular TCP connection with the fixed host on behalf of the mobile station, thus location management is omitted in that step of connection setup. As long as the mobile station is able to communicate with the MSR (for example, via a WLAN), no further procedures are necessary. The advantage of this approach is the possibility of adapting the transport protocol at the wireless interface to the unreliable and low bandwidth communication link, for example, a proxy approach is used similar to the one described for the GSrd case.

If the mobile host moves to the area managed by another ' mobile support router during the lifetime of the I-TCP connection. the association must be handed over to the new MSR. To keep the TCP indirection hidden to the application processes of both mobile and fixed station, the same endpoint parameters are used. By this, the communication sockets on both sides need not be re- established, which drastically improves the transport level performance [7].

The two cases described above again show clearly the difference of mobile data networking in public and private networks. I n the first case, the network provider supplies the end user with permanent reachability, (nearly) independent of his location, and performs mobility and accounting management. In contrast, wireless data networking in private environments will not provide all-time reachability if the user leaves his local authority and moves to another. Instead, station- initiated relocation strategies must be employed to inform the network about his current position.

PIVXJ~ Concepts, QoS Management ilnd OSI

The problcms discussed in the previous section have not been considered in the seven layers of the OSI RM. The proxy architectures can of course be implemented upon OSI, but their concepts are still left out of OSI. The general tasks will be the introduction of

QoS negotiation and management schemes that are spread over the protocol layers and

flexible transfer systems based on a hierarchical structure enabling hop-by-hop transport connections.

Especially the mix of the very different communication characteristics of multimedia services (for example, real- time video and audio data streams mixed with ordinary text) will require these concepts not only for wireless environments: Even for high speed networks new transport protocols like express Transfer Protocol (XTP) have been introduced and QoS driven communication architectures are investigated[ 101. In general, these problems apply to data- as well as telecommunication networks in wireless and fixed environments, but the wireless part is still the most crucial point.


Proxy concepts are not restricted to the transport layer, they can also be realized on application layer. In this case dedicated intelligent agents are responsible for autonomous processing of specific tasks of an application on behalf of the application peer entity. For wireless environments, similar to transport layer proxies, the agents are placed between mobile and fixed network as an intelligent application oriented mediation device. In contrast to the transport layer approach, an intelligent application proxy agent looses its generality, but gains more knowledge of application requirements and enables application specific actions.

For example, for the stringent requirements of RTI applications like automatic fee collection, dynamic route guidance, and traffic & traveler information a dedicated wireless communication system has been designed, and is currently under standardization by ISO[9]. In these Dedicated Short-Range Communication (DSRC) networks, small (about 5 m), non-overlapping communicat ion zones are served by external communication devices (beacons) using microwaves at 5.8 GHz or at 63 GHz or infrared at 850 nm wave length. Beacons are not necessarily connected to other beacons or a master computer, so they often build separate communication areas.

They can be mounted at the road side or on gantries. DSRC offers accurate localization and real-time communication services, obtaining correlation between communication, position and vehicle detection and thus enabling to do without special localization equipment. The DSRC protocol stack is a typical example for a reduced protocol stack for real-time environments. It consists of a physical, a data link and an application layer. Hence an on-line dialogue between vehicle and infrastructure via the beacon. which would require network layer protocols, is omitted.

The DSRC application layer is based on OSI network management principles, offering a reduced Common Management Information Service (CMIS) enabling remote object manipulations via the DSRC link. For all communication partners the whole dialogue is terminated in the beacon, resulting in proxy-architectures on application layer and application oriented QoS negotiation schemes. The OSI network management. with some simplifications a“dedicated distributed system to manage open distributed systems,” permits layer independent remote operations on network resources represented by managed objects.

In that manner dynamic parameter setting of network connections can be performed to guarantee a certain level of QoS. In general, OSI’s network management paradigm is a proxy architecture by definition. ITU’s Te 1 e c o in m un i c at i on M an age m e n t N e t w o r k ( T MN ) standard is an extension of these OSI standards towards a management reference model. Due to the general scope of the management structure, the required “toolbox” for (wireless) application oriented quality of service management exists in principle in OSI. On the other hand. there may be a temptation to shift more and more functionality into the network management, leading to unbalanced architectures with ‘empty’ protocol layers and an overloaded management.

Regarding future public mobile networks, third generation mobile systems must cope with the various telecommunication and data communication services, as well as the migration of mobile and fixed networks into one global worldwide universal communication system. Such systems are currently under development and referred to as Universal Mobile Telecommunication System (UMTS)[ 161. In 1992, the WorldAdministrative Radio Conference assigned 230 MHz of spectrum around the 2000 MHz frequencq band to the system. Phase 1 is scheduled to be operational around the year 2000.

The network concept of UMTS is a hierarchical, mixed cell structure. This hierarchy of macro-, micro- and picocells will allow a network layout according to the steadily increasing number of subscribers with differing mobility and service profiles. Furthermore, UMTS aims at an integration of different services offered by fixed, cordless and mobile (cellular and satellite) networks improving speech quality and the variety of data services provided. It will offer at least Narrowband ISDN (N- ISDN) basic services. B-ISDN is considered as one option for the backbone structure for UMTS. Moreover, UMTS will have a satellite component in addition to the terrestrial network, to increase the service area, for example, for ships, aircraft or rural areas without c om m u n i c at i on in fr a s t ruc t ur e . In add i t i on , t h e introduction of high speed services and suitable air interfaces for UMTS is under investigation, for example, in the RACE 2067 project Mobile Broadband System (MBS)[ 1 1,171.

In contrast to GSM systems, the services of UMTS will be realized via the Intelligent Network (IN) concept [ 121. Although the Capability Set One (CS-1) of IN mainly contains call related services, the introduction of IN will

APPLIED MICROWAVE & WIRELESS WINTER 1996 73

in general provide the means for a variety of different service profiles and flexible service offers to the end users. Intelligent Networks are based on the Signaling System No. 7, and, with some simplification, may be interpreted as a distributed application environment for exchange and processing of all service/subscriber/call- related data. The major task of Intelligent Networks, for example, efficient and flexible switching and management of services, is not sufficiently covered by the OSI higher layer protocols and the OSI network management.

New concepts like the reference model for Open Distributed Processing (ODP), although designed for the field of distributed computer systems, may have the capability to overcome this gap. Nevertheless, the different views on service switching, service management and network management must converge before the idea of open systems interconnection will be able to succeed in the world of global tele- and data communication networks[8].

Third generation mobile networks will rely heavily on distributed processing and distributed databases, in contrast to, for example, the centralized database approach of the GSM mobility management. In UMTS, the QoS management gains a new dimension: For example, the current specification of UMTS has foreseen several handover schemes, for example, from picocells to an overspanning macrocell to permit uninterrupted communication when entering or leaving in-house subnetworks, resulting in a dynamic change of the complete communication characteristics.

To sum up, it has been shown how current networks, whether rooted in tele- or data communication, and the developing market of mobile applications must lead to a suitable redesign of protocol stacks. The paradigm of the distinction between transport (network independent) and network (network dependent) layer, or the view of end-to-end versus hop-to-hop communication will not meet the requirements of future data communication in a mobile environment. These data services must cope with all future mobile applications, for example, the wireless office.

The management of end-to-end connections in a conventional OSI oriented fashion will be the weak point of a mobile communication system. The discussed architectural redesign of using proxy agents to handle mobility or heterogeneous network characteristics, enables the system to cope for example with short disruptions of the communication link by providing

suitable caching algorithms and disconnected operation functionalities. Different mobile terminals with distinctive capability profiles and heterogeneous networks require intelligent QoS negotiation schemes.

Throughout this paper, public and private mobile networks have been distinguished. This distinction still holds for future networks of the next generation. It might be a desirable goal to establish only one system for all applications. However, the introduction of wireless LANs providing mobility within a more or less locally restricted and privately owned area and the ongoing extension of wide area cellular networks towards the provision of data communication services may prevent one single future system.

Public mobile networks do offer mobility and service management enabling permanent user reachability and service availability, as well as the desired distinction between service and network provider. However, these advantages cause restrictions on the available quality of service (bandwidth, packet delays, etc.) compared to high speed island solutions. Both systems will still hold their position, even in a deregulated telecommunication market.“

Thus, the wireless computing developments at the end have different targets. The OSI reference model as defined and even partially degenerated, regarding the increasing shift of responsibilities towards management strategies, has to be adopted to a wireless environment. Rigid layering structures do not fit with the emerging data communication requirements. For example throughout the course of the 4th framework of the EU, the investigation of suitable communication architectures for future mobile and wireless data communication is a major issue.

Author’s Note

This work is partially supported by the Deutsche Forschungsgemeinschaft under grant no. Sp 23019-2.

References

1 M Y Bearman. “ODP-Trader”, IFIP Kansacriom OIZ Open Di rtrihuted Procescrng II. North-Holland. 1994

2. M. Satyanaranyanan, J.J. Kistler, P. Kumar, M.E. Okasaki. E. Siegel. D.C. Steere, “Coda: A Highly Available File System for a Distributed Workstation Environment”. IEEE Tmzsactions on Conzputers, Vol. 39, 4, 1990.


3 , "\I'~reless L,4N," Draft StundarJlEEE 802 11, December 1994

3 . T. 0hsan.a. G.Q. Maguire. "Bridging Functionality for Medium Access Control Sublayer", Pimwdings oftlie j t h S~.riiposiuiii on Peixinul, Indoor and .Mobile Radio Coiiiriizinic~itiori (PI.URC 'Y4). pp. 880 - 884. The Hague, 1995.

5 , C. Perkins (Editor). "IP Mobility Support". Internet Draft. Internet Engineering Task Force. May 1995.

6. R. Caceres. L. Iftode, "Inipro\~iiig the Peformance of Reliable Transport Protocols in Mobile Computing Environments". IEEE Juiiimrl on Selected Areus in Coiiiriiiinicutiuns. June 1995.

7 A Brake. B Badrinath. "I-TCP IndirectTCP forklobile Hosts", 1 C ~ h i i i c ~ i I Repurt DCS-TR-314. Department of Computer Science. Rutgers U n n ersit). 1994

8. LZ. Kocltelmans. E. de Jong. "Overview of IK and TMK Hariiionization". IEEE Coininiinicution Magazine. pp. 62 - 66, March 1995,

9 . S. Hoff. J . Kassubek. "Road Transport Informatics: Neiv Applications for Communication and Navigation Systems". PI-oc. of .\'ui,igcirion & Coiiiiiiirriication f i l l . .Llciritirn B Land ~Wuhile .,~i)plic.citioii.s, Breinen. May 1995.

10. B. Heinrichs. \V, Reiiihardt. "Adaptive QoS-driven Commiinicatio~i Architectures '.. Pruc. of the EiiroArch 93. Springer. pp. 430. 1993.

I 1 L Fernandes. "De\elopmg a System Concept and Technologies for Mobile Bioadband ComiiiuIiications". IEEE Personal C ~ ~ ~ i i i i r i i i r i ~ t r ~ i i s . pp 54 - 59, February 1995

12. G . S. Lauer: "IN Architectures for Implementing Universal Personal Telecommunications". IEEE .Yer\i,oi.k .Llagazine, pp. 6-16. March April 1994.

I3 M Rahnema "O\ e n lev Of The GSM System and Protocol Aichitecture", IEEE C ~ i ~ i ~ i i i i ~ ~ i c n t i o i i ~ blngazirie. pp 92-1 00, 4pril 1993

14 J Jayapalan. M Burke "Cellular Data Seik icesArchitecture and Signaling". IEEE Pertonal C~iIiiiiiiinicatiOii1 Llagazine, pp 44- 5 5. Second Quarter 1994

15. P. Gilchrist. B. G. Gunther. "General Packet Radio Services on GSM". .blohi/e Coriiiiiioiicutioiis Inteixatiorial. pp. 50-54, Spring 1994.

16 J Rdpeli. "UPVITS Targets. System Concept. and Stdndardization in a Global Fraiiienork". IEEE Percoiial Coiiiiiiiiiii(i1troiti Uagcizrne. pp 20-28. Febiuai) 1995

17. H . .4rmbruster, "The Flexibility of ATM: Supporting Future hluititnedia and Mobile Coiniiiu~iicatioiis". IEEE Coiiiiiiiiiiicntiorls .Llagazine. pp. 76-84. February 1995.

18 0 Spdiiiol. 4 Fasbender, S Hoff. J Kaltnasser. and J Kassubek. ..\\ lieleas Uetn 01 ks and 01s Uev Challenges for Protocol Stack Deaigll. IEEL I994 PI llRC

Otto Spaiiiol I-eceived the Dipl.-.Uath. degwe in inatheniutics iii I968 and the Di: rei: nut. degi-ee in 1971. hoth,fi.oiii the Ciiiwsitj, of Slxurbriicken, Gerinanj: Since I984 he holds a chair for comniirnicotioii sj'steiiis at the conipiiter science departiiieiit of ..lachen C'ni\.ersit?, of techno log^: Since 1992 he is the Cliuirnian of the IFIP TC 6 "Data Coiiiii2iiiiicutio,i.s". His research interests incltide systems modeling and per-fomance aiia1j.s is of conniiiini- cation netwoiks and distribtited sj'stems.

Andreas Fa.shentler: Sinion HofJ JoseJ'Kalt>c.asser and Jiirgeii Kassiihek received their Dipl. -1nfbriii. degrees in computer. science fi.oin Aaclieri Ciiiversi[i, of' Technoloa: The]. are ivith the coiizputei' science departnient ut ilncheii Lniwrsity of Teclinolog~~ as reseurch assistants and, fiirni the mobile coriiriiiinicutioii.i i m k i i i g groirp o f the departnient. Their research interests include Wireless L.~.\'s, third genewtion cellular .s!'.stenis trnd R TliVHS sjsienis.

76 ..jPPLIED MICROWAVE & WIRELESS WINTER 1996

Challenges for Wireless Networks and OSI - The Eye Archive/v8n1pg50.pdfChallenges for Wireless Networks and OSI ... network layer orks physical layer Figure 1. The OSI reference model

Documents