A brief overview is given on background issues, concerning in particular the network technologies and services being developed to face the new market challenges. The paper focuses then on the current status of international standardization activities in the high speed arena. Established organizations such as CCITT, ISO, ETSI, IEEE have been working on these issues for the past few years; a number of interest groups (ESIG, EURESCOM, ATM Forum) have also been created, to answer the growing need for early implementation specifications, based on available standards and suitable for early field trials. Finally, the evolution trends in the high speed arena are dealt with, summarizing the expected progress and highlighting crucial areas where close attention will have to be paid to ensure the timely development of standards.
This paper describes the work done and results achieved in implementing a fault management prototype for a Metropolitan Area Network (MAN) switching system, the GMAN (GMS Fault Management Prototype for a MAN). The GMAN is implemented using the Generic Maintenance System (GMS) which has been developed by the RACE project AIM (R1006). The GMS is a model based expert system toolset (`2nd generation XPS') comprising components for knowledge acquisition, knowledge representation, model based reasoning, human machine interface, and dialogue design. The GMS supports automatic fault management from receiving and treating fault events to guiding an operator through a repair procedure.
This paper aims at giving a general description of major issues related to network management of IEEE/ETSI DQDB standard compliant Metropolitan Area Networks. The specifications defined within both the European standard-setting body (ETSI) and the European SMDS Interest Group (ESIG) are described and the characteristics of a prototype under development in the MANTIS European research and development project in the ESPRIT framework are illustrated.
The growing user demand for high speed data communication across wide area networks can now be satisfied by Frame Relay and Switched Multimegabit Data Service. The flexibility of a switching system based on metropolitan area network technology can cope well with their different requirements, and integrate FR, SMDS, and XA-SMDS on the same network. The network management system accommodating the various requirements coming out of the common switching functions and the different services provided, is addressed in this paper, where reference architecture and implemented functionalities are described.
ETSI has prepared a set of standards for a multi-service Metropolitan Area Network (MAN) which is based upon the IEEE 802.6 Distributed Queue Dual Bus Standard (DQDB). This paper commences with an overview of the ETSI MAN standards and then describes an ETSI compliant MAN node which is being developed as part of the Esprit III project 7314 `MANTIS,' and will subsequently be developed into a product. The node is based upon a segment switch operating at 155 Mbps, connected to a number of network access modules (NAMs) which provide connections to users, other nodes and other MANs. The node supports Telecommunications Management Network (TMN) compliant management. Novel features of the node include multicast capability in the switch, a flexible segment buffering strategy, and use of a 56 byte switched segment structure which unifies ATM and DQDB switching requirements.
This paper highlights the need for sophisticated multicast mechanisms to be provided by transfer protocols in order to support group communication. This need becomes even more evident in the light of the special QOS requirements imposed by multi-media applications. Today's protocols do not provide the broad range of functionality required by these upcoming new applications. From a number of sample applications we derive functionality and performance requirements which have to be provided by the transfer systems. A flexible implementation strategy being capable of dynamically coping with these demands is introduced. Taking the widely used XTP transfer protocol as an example, we show how some of the identified demands are met by this protocol. Furthermore, necessary enhancements are suggested.
The network, called ALADIN, is a modular network for industrial, domestic, and automotive applications which integrates isochronous and asynchronous services ranging from a few kb/s to several Mb/s. ALADIN is based on a star structure with a passive optical star coupler. The ALADIN control is decentralized for safety reasons. A master station which is dynamically determined at start-up, generates synchronous frames and the other stations include their data into these frames by superposition in the star coupler. The paper describes the concept and the building principles of the network, the protocols, and the routines for synchronization.
A growing diffusion of LANs and development of more complex computer applications is pushing requirements on network providers for a new kind of service able to offer connectivity and performance comparable to that of LANs but over a metropolitan area or better inside a geographical extension. B-ISDN is the future answer that standardization bodies have identified and PTOs chosen; but availability of B-ISDN is foreseen only for the late 90s. In the meantime SMDS, elected as the first service in the context of B-ISDN, seems to be the right solution for current users' needs, especially for LAN interconnection. In order to provide SMDS early, evolutionary towards B-ISDN but realizable solutions have to be used: SIP has identified MANs as a technology to offer an SMDS-like service in the near and medium term, and to gather traffic for future B-ISDN in the long term. In this view SIP is experimenting with Metropolitan Area Networks (MANs) on field, together with other technology like star- based geographical networks.
Based on a survey of 400 large European companies the paper discusses the potential for a MAN service. The focus here is on existing and future use of networks and applications. The survey indicates that growth in the use of existing applications and the general tendency to upgrade networks on a running basis will provide an increased traffic basis for a Metropolitan Area Network (MAN) based service. The service must provide access at 10 Mbps and support both asynchronous and synchronous traffic. But the overshadowing question is tariffs and no information is available on this subject in Europe yet. If prices are reasonable -- not favoring one or the other service -- the survey indicates that by the end of 1994 at best 250 companies will be using a MAN service, a figure that is likely to grow to 6,000 companies by 1997.
In 1987 the DQDB (Distributed Queue Dual Bus) technology for MANs (Metropolitan Area Networks) was standardized in IEEE 802.6. Today in Europe, over 20 MANs are in place. The results of the first trials were from an operational viewpoint very successful and a big step in transmission power. So now, just a short time after both bodies, Bellcore (for the USA) and ETSI (for Europe) decided to standardize on SMDS/CBDS (Switched Multimegabit Data Service/Connectionless Broadband Data Service) as the new high-speed services, based on MAN technology, for regional and supra-regional areas, these services are now being introduced in the USA and Europe as global, public services.
This paper, based on work within RACE project R1022 -- `Technology for ATD,' places the evolution of MANs towards ATM within the overall context of ATM introduction and evolution. Beginning with a major motivating factor, the need for LAN to LAN interconnection, a likely evolution of MANs is described, from first introduction, through interconnection via the ATM cross connect network, to the provision of connectionless data service over the ATM network. The introduction and evolution of ATM in general is then considered. This begins with a review of basic introductory approaches (substitution, island, overlay) and how they can be applied. An overview of the main customer groupings and their likely service requirements is provided as a precursor to discussing how fiber will be used in the access network for both business and residential customers. The paper concludes with an assessment of how MAN evolution towards ATM fits within the overall context of ATM introduction and the provision of broadband services.
The paper presents and analyzes two alternatives of interworking modes for interconnection of IEEE 802.6 based Metropolitan Area Networks (MANs) and ATM based B-ISDN. The two alternatives are the `cell-to-slot interworking' and the `frame interworking.' The cell-to-slot mode modifies ATM-cells to DQDB-slots and vice versa, whereas the frame interworking mode reassembles ATM-cells and converts the resulting AAL frames to DQDB IMPDU frames. In the paper, simulation results of a DQDB network are presented, where one of the DQDB stations represents an IWU getting aggregated traffic from B-ISDN. The aggregated traffic is modelled by a mixture of various constant bitrate and variable bitrate traffic sources representing different real-time sources and background traffic. It turns out that the cell-to-slot interworking mode is much more useful for keeping bandwidth requirements of real-time traffic than the frame interworking mode which causes additional burstiness due to reassembly. Based on these results one can conclude that the cell-to-slot interworking mode is much more appropriate for real-time applications than the frame interworking mode.
ImNet is a fiber-optic local area network, which has been developed for high speed image communication in Picture Archiving and Communication Systems (PACS). A comprehensive analysis of image communication requirements in hospitals led to the conclusion that there is a need for networks which are optimized for the transmission of large datafiles. ImNet is optimized for this application in contrast to current-state LANs. ImNet consists of two elements: a link module and a switch module. The point-to-point link module can be up to 4 km by using fiber optic cable. For short distances up to 100 m a cheaper module using shielded twisted pair cable is available. The link module works bi-directionally and handles all protocols up to OSI-Level 3. The data rate per link is up to 140 MBit/s (clock rate 175 MHz). The switch module consists of the control unit and the cross-point-switch array. The array has up to fourteen interfaces for link modules. Up to fourteen data transfers each with a maximal transfer rate of 400 MBit/s can be handled at the same time. Thereby the maximal throughput of a switch module is 5.6 GBit/s. Out of these modules a multi-star network can be built i.e., an arbitrary tree structure of stars. This topology allows multiple transmissions at the same time as long as they do not require identical links. Therefore the overall throughput of ImNet can be a multiple of the datarate per link.
The performance of a DQDB MAN supporting both connection oriented (CO) and connectionless (CL) services as observed at the point of attachment to an ATM network is the subject of this paper. The overall quality of service (QoS) preserved in the ATM-DATM-DQDB interworking is reflected in the traffic statistics of the multiplexed stream exiting the ATM- DQDB interworking unit (IWU). For the purpose of accurately generating this stream a simulation model is developed. This incorporates several source node traffic models for CO and CL services, the DQDB and IWU protocols as well as the signalling and metasignalling procedures activated for the establishment of end-to-end connections among communicating users. Through numerical evaluation of several access priority rules and protocol architectures we identify all those factors that are essential for the design of both the CO DQDB functions and the ATM-DQDB interworking components.
Simple channel transmission error arguments show how the size of an all-optical multihop network employing deflection routing is limited for a given optical bit rate. These limits are quantified here for non-regenerative all-optical mesh networks such as the Manhattan Street Network and ShuffleNet employing solitons. It is found that the node-to-node fiber span cannot exceed a few kilometers for network sizes greater than about 64 nodes when the optical bit rate is in the range of 100 Gb/s if the packet error rate is to be bounded below 10-6.
Ring networks present an attractive solution for optical, high speed local and metropolitan area networks due to the simplicity of network interfaces and access control. Two problems need to be overcome to obtain an all optical network. One, the limitation on power budget resulting from optical losses that occur when data passes through intermediate nodes. The other, a reduced network throughput related to the linearity of the ring topology. Recent progress in WDM techniques has opened the possibility of overcoming this problem by an optical multi- channel solution. WDM taps the large fiber bandwidth by using different portions of the optical spectrum to realize (omega) different channels on the same fiber. However, in extant electronic node based architectures, even though high bandwidth optical transmission can be used to propagate packets between the nodes, the electronic elaboration of data at each node creates a performance bottleneck for the whole communication system. This leads to network throughput that is a mere fraction of the optical bandwidth potential. This work presents an approach to obtaining a concurrently accessed multi-ring all-optical WDM network (CROWN) with a node architecture in which packets pass through the node without being converted into the electronic domain. Using a single high speed transmitter and receiver, CROWN allows the data to be maintained in optical format while resolving receiver contentions.
CRMA-II (Cyclic Reservation Multiple Access, Version II) is a media access control protocol for local and metropolitan area networks working in the Gbit/s range. The access scheme is based on direct access to a slotted medium combined with a reservation scheme guaranteeing both high utilization of network capacity and global fairness. The performance of CRMA-II is improved in comparison with the forerunner CRMA protocol by installing insertion buffers at each station. Network throughput is pushed to values far beyond the transmission capacity by spatial reuse of slots while buffer-insertion technique reduces media access time but increases round-trip delay. In this paper a new media access scheme for constant bitrate (CBR) traffic is proposed. For acknowledged CBR connections CBR slots have immediate access to the transmission media. While there is one insertion buffer for each priority of data traffic, CBR traffic is allowed to bypass insertion buffers to ensure minimum delay for constant bitrate services. In this paper a performance evaluation of the CRMA-II media access scheme and the new access mechanism for CBR traffic are presented. The evaluation was done using discrete event driven simulations.
Current metropolitan area networks are often a bottleneck for new, sophisticated multimedia services. Not even the IEEE 802.6 Distributed Queue Dual Bus (DQDB) manages to ensure these new services of adequate performance, especially when the communication channel bit rate is high. One way to achieve networks with a high throughput is to use multiple channels on optical fiber, by means of wavelength division multiplexing (WDM), and parallelize operations for access to the communication channel. In this paper the authors present a WDM Metropolitan Area Network based on the DQDB protocol in which the various user information segments are parallely inserted into the transmission queues. On the basis of the network architecture proposed, a management structure for multimedia applications is presented.
This paper deals with the implementation issues concerning efficient usage of optical Code Division Multiple Access (CDMA) Local Area Networks (LANs). It first discusses user requirements and how these determine the structure and design of LANs. It then goes on to introduce optical CDMA LANs and the various techniques that are used to provide an efficient communication media. The paper then continues with an investigation of the various problems concerned with efficient usage of this media and proposes methods that may be used to overcome these problems based on soliton logic and transmission. Issues such as electronic hardware limitations, PN sequence generation and synchronization are covered. The paper is concluded by a summary of the feasibility of the methods discussed. Proposals for methods of implementation are made and obstacles requiring solution before the techniques become attractive for wide use in LANs are highlighted.
The first part of the paper, after an overview of frame relay, SMDS/CBDS, and ATM experiments in U.S.A., Europe, and Japan, describes the high-speed network field trial carried out in Torino, Italy focusing on performance measurement aspects, application benchmarking, and users' feed-back. Then, the paper deals with the initiative concerning the deployment of a pilot ATM virtual path infrastructure across Europe in 1994, which aims at interconnecting the national ATM pilot network. Furthermore, the results of a feasibility study concerning the implementation of a Pilot ATM Network are pointed out. At the end, the focus is on evaluation and testing problems and on the advantages of the ATM solution versus user requirements.
The test and performance evaluation of high speed networks pose problems that cannot be solved by traditional equipment. In this paper, the concept of a powerful and versatile tool, capable of generating and detecting traffic to test and assess the performance of such networks, is presented.
SMDS is a new high-speed data service that is currently being implemented in the United States and will soon be available in Europe. The characteristics of SMDS and the protocols used to provide it are briefly explained. The protocol tester, a particular kind of telecommunication test equipment, is presented. Some typical test scenarios for SMDS are shown, and approaches used to implement an SMDS protocol tester are discussed. While SMDS is being introduced, another technology called B-ISDN is emerging as a worldwide standard. The goal of the B-ISDN network is to provide a wide range of high-speed communication services with near-universal connectivity, in the same way that the telephone network provides voice service around the globe today. The reliability and quality of service provided by the B-ISDN network will be extremely important. The requirements for a B-ISDN tester are discussed, and compared to the requirements for an SMDS tester.
Local Area Networks (LANs) have completed two generations of development (Ethernet and Token Ring the first, and FDDI and DQDB the second); the large volumes of traffic involved in the emerging multimedia applications, however, lead towards a third generation of LANs. This generation must provide real-time capabilities needed by new services and solve the problems of interworking with ATM-based B-ISDN. Moreover the possibility to vary the subscribed bandwidth with the B-ISDN will be given to the LAN interfaces. This paper focuses on an architecture for protocol testing of a Dynamic Bandwidth Allocation Protocol inserted in a LAN environment based on ATM technology. In fact, the technology of the third LAN generation will be the Asynchronous Transfer Mode solving every interface problem with the public B-ISDN. A testing and debugging environment which checks the implementation of the Dynamic Bandwidth Allocation Protocol at the interface host/LAN- ATM is discussed. The main concepts of the overall system architecture are analyzed, evidencing both software and hardware issues.
In passive optical networks using a time-division multiple access (TDMA) scheme, signal bursts from different remote transmitters are interleaved towards a common headend receiver in the upstream direction. We describe a receiver concept allowing correct detection of consecutive signal bursts which exhibit high nonuniformity of amplitude and offset. The time constant of the ac-coupled receiver is controlled by the digital circuitry of the headend station. During reception of the synchronization preambles, it is switched to a low value to allow fast settling of the receiver. Additionally, a power-control system with improved dynamic performance, which can also compensate for fluctuations of optical attenuation, is discussed. The structure of the transmitter and the receiver are described, and the results of simulations and experiments are presented.
In this article we present methods of amplitude and phase (clock) recovery for burst mode fiber optic receivers. Using these methods it is possible to recover amplitude and clock phase using an 8 bit preamble, which contains two guard bits, a field of three `1s' for amplitude recovery, and a `0 1 0' sequence for clock recovery. The methods shown are application independent and can be used with bursts of varying length, thus will be applicable to a number of LAN and MAN systems, as well as telephony over passive optical networks. Experimental results are presented.
The high performance characteristics of lenseless, tunable, all-fiber Fabry-Perot filters allow diverse optical signal processing necessary for a variety of wavelength division multiplex telecommunication systems and wavelength division multiple access data communication networks.