We investigate the throughput performance of networks interconnected through a bridge under real-time traffic environment. Throughput characteristics versus input traffic rate are shown with the variation of node spacing for fixed number of nodes in each of the rings. Certain buffer management algorithms which allocate the buffer space of the bridge dynamically are also studied. The algorithms are basically aimed at maximization of buffer usage. The networks are connected with each other through a bridge which is not in-j volved in correcting errors in the packets rather it simply acts as a store and forward device.

The standardization of FDDI has now reached a level of maturity. A key component that is specified within the suite of FDDI standards is the concentrator. This paper discusses how the concenirator fits into the intelligent wiring hub environment and explains how support for network management and the interface to multi-media cables is applicable for an intelligent FDDI concentrator.

The SAFENET II standard is a seven layer network profile which is being developed for use in Navy computer networks. SAFENET II employs the ANSI FDDI standard as its physical and data link layers. The profile also includes protocols from 051 and MAP as well as the emerging Xpress Transfer Protocol (XTP). The SAFENET II profile is broken down into three protocol suites: 051 lightweight and combined. Any of these suites may be implemented depending on the system requirements. The SAFENET II physical topology is adapted from the FDDI standard and includes the use of bypass switches for enhanced reliability. SAFENET II provides many network survivability features to the system integrator.

A 100-Mbit/s FDDI network interface unit (NIU) is described that supports real-time data, voice and video. Its high-speed interrupt-driven hardware architecture efficiently manages stream and packet data transfers to the FDDI network. Other enhancements include modular single-mode laser-dioce fiber optic links to maximize node spacing, optic bypass switches for increased fault tolerance, and a hardware performance monitor to gather real-time network diagnostics.

ECHO is on.

This paper addresses the advantages of using a concentrator for inserting or removing a station onto or from the Fiber Distributed Data Interface (FDDI) ring. It gives an overview of the architecture of a FDDI concentrator. It highlights major data and clock distribution problems in a backplane bus. The main focus is a design example of a small and large concentrator using the National Semiconductor FDDI chipset. Different timing contraints such as setup and hold data and clock flight time and flight time cancellation are discussed in detail.

The FDDI standard provides for a high speed optical channel for interconnections between mainframes and peripherals and for use as a backbone network between lower speed local area networks (LANs). As FDDI networks move out of the data center environment detailed engineering rules are required to construct and administer the dual ring architectures for building and campus applications. This paper reviews the generic building and campus layouts presented last year at the SPIE OE/FIBERS ''89 Conference and demonstrates the implementation of the logical rings within the standard physical star building wiring topology. New work is presented to simplify the engineering rules reported last year and to conform to the EIAII''IA Commercial Building Telecommunications Wiring Standard. The networks addressed range in size from those confined to a single closet to those spanning a single multi-floor building to larger networks encompassing a campus involving multiple buildings. The introduction of colorcoded fields and specialized administration labels results in simplified engineering rules for constructing networks of any size and configuration which conform to the FDDI dual ring standard.

The purpose of this paper is to discuss cost effective components for workstation interconnection as Fiber Distributed Data Interface networks transporting data at 100 Mbps proliferate within the next few years. The 1300 nm components called for in the draft standard are relatively expensive and will probably be used in the backbone. However 830 and 660 nm components based on AlGaAs/GaAs technology appear to be suitable and offer significant cost savings for workstation interconnections over 50 to 500 meter distances. The 830 nm interconnection still uses silica fiber while 660 nm interconnetion could be done with cheaper polymer fiber as the medium. The discussion is centered on a comparison of the costs and performance of these alternative components with the l300nm devices. The presently available 830 nm technology could carry the data over 500 meters while the 660 nm technology with polymer optical fiber can be used up to 50 meters. A recently developed receiver based on a monolithic GaAs preamplifier with a sensitivity of -39 dBm average at 125 Mbps and a wide dynamic range is also described here.

The Physical Layer Medium Dependent standard specifies the constraints faced by a designer of an FDDI cable plant. These include fiber type system bandwidth and attenuation and connector type. In designing an FDDI cable plant an understanding of the PMD and the application is required. There are three major types of FDDI networks: backbone backbone wiring horizontal wiring and data center wiring. Each has a logical and physical configuration.

Networks designed around the ANSI X3T9. 5 standard called Fiber Distributed Data Interface (FDDI) are limited in distance by the bandwidth not the attenuation of the fiber optic cable. The magnitude of this limitation is defined and potential options in sources and fibers are examined. Since FDDI is limited in distance by the bandwidth of the cable plant not the attenuation many end users have put a bandwidth specification in their fiber optic cable purchase specifications. Some have even specified testing for bandwidth. This has caused much controversy between vendors installers and users. As a test equipment vendor we have " refereed" quite a few of these already and we have found that more myths are believed than realities are understood. What Limits Fiber Bandwidth? Two things limit fiber bandwidth: modal dispersion and chromatic dispersion. Modal dispersion is caused by the fact that the graded index profile of the fiber isn''t perfect so all modes do not have exactly the same transit time through the fiber. Chromatic dispersion is caused by the fact that the velocity of light in the fiber is dependent on the wavelength of the fiber. Fibers are designed for a zero dispersion wavelength light of any other wavelength will have some dispersion. Sources of wide spectral width like LEDs will suffer from more dispersion than narrow spectral width sources like laser diodes. In most circumstances one can

Proper FDDI Physical Connection compliance testing requires careful attention to the jitter and loss budgets described in the FDDI Physical Layer Medium Dependent Standard (PMD). The optical system jitter and loss budgets are reviewed in detail to explain how these characteristics affect the performance of the FDDI optical system. Worst case optical test conditions for use in compliance testing are described.

The advantages of fiber optic data communications are well publicized. The system designer trying to solve a particular application problem is faced with many issues involving many technologies if they are to take advantage of fiber optic communication. The information to be transmitted is usually located in a memory or on a processor bus in the form of digital words (bytes) most often as 5 volt CMOS or TTL logic levels. To accomplish the transmission of this information from one system to another over optical fiber the following must be implemented. The data must be converted from parallel to bit serial format. More than likely it will be encoded to guarantee an edge density in the transmission media. It must also include some level of protocol for signaling purposes. These functions are most often implemented in silicon or GaAs integrated circuits. The data must now be amplified and shaped to drive a light source of a specific wave length. This is most often a III V compound semiconductor diode. The source must be critically aligned and mechanically secured with an optical fiber. In most cases an optical connector is involved. At the receiving end the same technologies and processes are used in the reverse direction. The fiber is aligned to a III V diode detector. The signal is amplified and timing is regenerated from the edges. The data is

This paper describes a complete 12 channel data link system designed for high speed parallel optical fiber communication at l. 3im wavelength. The system design is based on the technologies of 1x12 InGaAs photodiode and InGaAsP LED arrays and burst mode compatible receiver. The transmitter and receiver are linked by a 12 channel optical fiber ribbon. The system allows arbitrary data encoding protocols at any frequency between dc to 200 Mb/s/ch (total thru rate of 2. 2 Gb/s). Data transmission over I km has been achieved.

To interconnect multiple user stations in Local Area Networks (LANs) and/or Metropolitan Area Networks (MANs) environment a ShuffleNet multichannel multihop Lightwave Packet Communication Network concept has been recently proposed. The Wavelength Division Multiplexing (WDM) concept is applied to provide a total network capacity in the order of hundreds or thousands of Giga bits/second (Gbps) with each user station assigned a peak limited capacity of 1 Gbps. The performance analysis of a ShuffleNet multihop lightwave network under uniform and nonuniform input traffic load patterns are studied and the corresponding simulation results are presented in this paper. The input traffic load to the network consists of uniform as well as nonuniform traffic patterns. . A fewer ShuffleNet network topologies are simulated and A Fixed Routing algorithm as well as an Adaptive Routing algorithm are applied for routing the uniform and nonuniform input packet traffic requirements. The resultant average network throughput and packet delay performance characteristic curves are plotted and analyzed. The corresponding simulation results indicate that the adaptive routing algorithm provides better average network throughput and packet delay performance characteristics when compared to fixed routing algorithm.

Systematic relationships among codes code weights and code length have been developed for pseudo orthogonal codes. Special emphasis has been placed on identifying minimum code lengths applicable to code division multiple access systems for given codes and weights. These derived relationships are applied to temporal spatial and spectral CDMA systems to estimate their link budgets complexity and performance in various communication applications. The construction and benefits of a new class of hybrid pseudo orthogonal codes is discussed.

A novel coding scheme (DmB1M) proposed by Japanese researchers is analysed and compared with other coding techniques commonly adopted for optical communications. An experimental realization of the new code (D1OB1M m 10) using discrete components has been performed and the circuit test results for an information rate of 140 Mbps are presented. The design of speed converters and clock recovery circuits which are crucial for the code realization is described in the paper. The coding technique is suitable for optical transmission with data rates in the order of Gbps.

Development in Picture Archiving and Communications Systems for medical imaging applications requires high resolution real-time image transmission. In this paper we describe a 500 MHz baseband analog fiber optic system capable of transmitting 2K x 2K medical images for remote viewing applications. The system uses an intensity modulated laser diode in the transmitter and a germanium avalanche photodiode in the receiver. The system was tested in a field trial.

Some problems of applications of the fiber optic networks for mining seismology are presented. A structure of the network multiplexing and modulation techniques transmission quality and design aspects are discussed with respect to mining environmental conditions typical topologies and mining technical standards. Emphasis is laid on compliance with requirements of safe operation in mining environment. An example network is presented. The network has been given official approval for inmine practice. It serves for monitoring crump hazard in one of coal mines in Poland.

The proliferation of departmental LANs in campus environments has driven network technology to the point where construction of token ring fiber-optic backbone systems is now a cost-effective alternative. This article will discuss several successful real life case history applications of token ring fiber in a campus setting each with unique distance and load factor requirements. It is hoped that these examples will aid in the understanding planning and implementation of similar installations. It will also attempt to provide important information on the emerging Fiber Distributed Data Interface (FDDI) standard.

The application of an expert system for network management of a prototype optical fiber Local Area Network (LAN) is described. The logical bus physical star layout of the Lehigh University fiber LAN is designed for simultaneous 0. 83 jim and 1. 3 un operation using Wavelength Division Multiplexing (WDM) technology. The LAN application examined in this paper involves transmission at 0. 83 jim on multimode fiber with a 10 Mbps CSMA/CD MAC protocol (the 1. 3 im wavelength allows for future integration ofa 100 Mbps FDDI protocol). The Lehigh University LAN being described uses the NOVELL 386 operating system (OS) for network control. Network Management of this fiber LAN is provided by a prototype expert system which is coupled to the Netware 386 OS. The expert system uses the Netware data base in fault management The expert system also provides a detailed historical record of network operation. The expert system is designed to learn by adding new fault diagnosis rules as new types of faults occur and new causes for faults are defined. Early experiences with expert system network management of the 10 Mbps fiber LAN are discussed.

The pace of change in high-speed networking has been exceptionally fast in the last ten years. We discuss several general categories of contemporary solutions for high-speed data networking within a campus. The solutions are contrasted in several dimensions advantages and disadvantages are reviewed and guidelines to assist a campus in selecting the best approach for its individual needs are suggested. While FDDI and other emerging standards are the focus of media attention these days proprietary techniques optimized for real-time traffic still suit many applications best.

Case Western Reserve University is operating the first all fiber optic communications network on a university campus. When completely installed this system of some seven thousand outlets will interconnect all faculty offices student moms classrooms libraries and laboratories with computer data telephone audio video fax and image information resources. We refer to the system as CWRUnet and pronounce it " crewnet. " CWRUnet features a standard premise wire-once cabling architecture independence of cabling and optoelectronics and support for multimedia communications. Phase I of the network became operational in August 1989 for students in the University''s sixteen residence halls. Phase II connecting faculty and staff offices and other locations is completed in twelve more buildings with the remainder under construction or scheduled for 1990-91. CWRUnet accommodates a heterogeneous assortment of microcomputers servers and gateways to other networks telephones television equipment remotely operated surveillance devices and energy management controls. It is planned to incorporate this campus-area network into a metropolitan area network following the proposed IEEE 802. 6 standard. CWRUnet is steadily evolving into one of the earliest implementations of Broadband ISDN. The CWRUnet project has been carefully documented and precise costs are known for each element in the network. The data service is presently based on the familiar networking technologies of ethernet and TCP/IP and now offers a wide variety ofnetwork-based information services to the campus community. CWRUnet is an innovation because

The University of Massachusetts is planning to use fiber optic cable as the primary medium for a campus high speed data network. In the University environment it is critical to identify primary users and work cooperatively with them in the planning and implementation of the new network. With the installation of a new telecommunications system and associated new cableplant theUniversity plannedfor fiberoptic useby installing two one-inch innerducts specifically for fiberoptic cable in every conduitrun to ailmajor buildings on campus. The University commissioned a topological design forpulling fiber optic cable in a four phased plan that included all major academic administrative and residential buildings. Beginning with the first phase twelve to seventy-two strand 62. 5/125 micron fiber optic cable willbepulledto twenty-two buildings. A six node FDDI based network with Ethernet extensions will be implemented over this cable.

This paper discusses the strategic and technical aspects of upgrading a point-to-point fiber-optic Ethernet link to an active Ethernet hub for the purpose of networking MIS and academic services on a 220-acre campus.

Bridging among DQDB subnetworks and between LANs and DQDBs is discussed. Some fundamental issues encountered during the implementation and standardization of such bridges are examined. These issues include the level of interconnection the service requirements and performance measures the functionality needed in DQDB nodes dedicated to bridging addressing and address mapping routing through bridges error and flow control security of transmitted information reconfiguration and fault tolerance. Approaches to solve some of the identified problems are proposed and discussed.

The Distributed Queue Dual Bus network (DQDB) [1] is a candidate for the IEEE 802. 6 Metropolitan Area Network (MAN) standard. This network integrates two classes of traffic on the basis of an hybrid multiplexing technique. In this paper we propose a packet-switching multiplexing suitable for both asynchronous and synchronous traffic on DQDB. First we consider an extension of the Distributed Queueing protocol (DQ) adapted to real time constraints. We call it general Distributed Queueing protocol (g-DQ) [2] . The g-DQ protocol needs (as DQ does) a load control algorithm to be reliable under heavy load. Among several possible algorithms we investigate an adaptation of the Load-Controlled Scheduling of Traffic (LOCOST) [3] to g-DQ. Our simulations show the feasibility of such an adaptation. Finally we define the main characteristics of a new packet-switching multiplexer for DQDB which should facilitate the opening of this MAN towards the Broadband ISDN (B-ISDN). KEY-WORDS: Metropolitan Area Networks (MAN) High-Speed networks load control algorithm packetized traffic hybrid multiplexing Asynchronous Transfer Mode (ATM) B-ISDN.

This paper describes an adaptable-bit-rate video service concept experiment and its management in an experimental prototype of a public metropolitan area network (MAN). In the experiment the " service providers" supply their customers with a set of service management primitives to implement customer-defined management applications and provide users with a high level of flexibility in the service definition. The paper describes the architecture for an experimental service management system that includes customer controlled features for dynamic bandwidth allocation group addressing and address screening. 1

The design of optical switching systems built to serve practical metropolitan network requirements is described. Applications considered include route diversity switching for the protection of transmission systems against cable outages automatic fault sectionalization using loopback switching centralized remote testing of fiber optic cable systems optical cutover switching for transmission equipment relocation and node bypass switching for the protection of transmission systems against loss of intermediate terminals.

Switched Multi-megabit Data Service (SMDS) is a proposed high-speed packet-switched service which will support broadband applications such as Local Area Network (LAN) interconnections across a metropolitan area and beyond. This service is designed to take advantage of evolving Metropolitan Area Network (MAN) standards and technology which will provide customers with 45-mbps and 1 . 5-mbps access to high-speed public data communications networks. This paper will briefly discuss SMDS and review its architecture including the Subscriber Network Interface (SNI) and the SMDS Interface Protocol (SIP). It will review the fundamental features of SMDS such as address screening addressing scheme and access classes. Then it will describe the SMDS prototype system developed in-house by NYNEX Science Technology.

Deployment of Broadband ISDN is being influenced by both a market pull and a technology push. New broadband service opportunities exist in the business and residential sectors of the market place. It is envisioned that some customers will need connections directly to broadband switches because of the high bandwidth needed for their applications. At the same time Metropolitan Area Network (MAN) systems will serve those customers with bandwidth requirements less than or equal to 150 Mbps. A given MAN will have a geographical domain to serve where it will carry out the switching tasks within this domain. While MANs couldbe designed using differentarchitecturalconcepts the setofservices expected tobeprovidedby MANs could be equivalent to thelist ofservices thatwillbe supported by the targetbroadband network. This paperpositions MANs as a major building block for Broadband networks. It also examines the evolution process ofMANs as a needed step to assure the successful deployment of these new broadband services. 2. BISDN - OVERVIEW Broadband ISDN (BISDN) is being driven into existence by both a market pull as well as a technology push. Opportunities for new valueadded services are the prime market pull for future broadband networks. These services opportunities extend beyond simple voice and low speed data applications and cover both the residential and the business sectors of the market. It is noted for instance that business customers have growing needs for sophisticated telecommunication vehicles to support their

Fiber Channel (FC) is a new " physical interface" designed to carry the command sets associated with the existing SCSI IPI HIPPI and IBM Block Mux channels. FC supports multiple technologies to obtain different cost and performance levels - up to 102. 5 megabaud over 2 km. FC has a single " framing protocol" and uses the IBM 8B/1OB coding scheme. The background to the ANSI effort and the potential market will also be described.

This paper describes a serial link solution at 1. 25 Giga baud to extend the current distance of 25meters to up to 2 kilometers for the American National Standards Institute (ANSI)''s High Performance Parallel Interface (HIPPI) standard. The proposal uses Gazelle''s HOT ROD technology along with fiber optic links to serialize the HIPPI interface for data rates of 800 Megabits per second (Mbps). This scheme enables the use ofremote peripheral access interconnection of Local Area Network (LAN)s and Small Area Network (SAN) applications.

A High Speed Optical Data Transfer Network (HSODTN) experimental research prototype has been constructed for the Naval Sea Systems Command (NAVSEA)1. It was installed at the Naval Research Laboratory (NRL) to explore new high speed fiber optic network concepts and technology applications for broadband communications. The present status and objectives of the HSODTN are outlined. This experimental network utilizes a 1 . 7 Gb/s lightwave transmission system that serves as a fiber optic backbone. It can readily be upgraded to Asynchronous Transfer Mode (ATM)/Synchronous Optical Network (SONET) standards. The basic prototype hardware architecture consists of a network interface (i. e. Media Access Control (MAC) layer) and several independent modular units that interface with various lower bit rate services. At the MAC layer integrated services are supported by giving priority to those traffic types requiring bounded delay. The organization is modular allowing components and services to be added or improved while redesigning the minimum amount of hardware. The software network architecture includes a network controller to configure monitor and administer the network.

The Fiber Distributed Data Interface (FDDI ) will provide convenient high speed transmission ( 100Mb/s ) network among the various computers. For the optical data link which has been standardized for FDDI the jitter performance is one of the most important items from a view point of interoperability . We analyze the relation between the jitter characteristics and the circuit design of optical data links by using a computer simulations and compare it with the experimental data. And also we consider the performance of optical data link which is satisfies FDDI PMD requirments*. ( * FDDI jitter requirment of optical receiver : DCDDDJ 1. 4 nsec RJ 2. 13 nsec)

The purpose of this paper is to compare the cost and performance trade-offs of polymer optical fiber networks with glass fiber networks and networks on other media. The discussion also includes distance versus attenuation and bandwidth limits for higher data rate transmission over polymer fibers. Finally we describe the performance of a 10 Mb/s Ethernet system built and tested using a modular approach incorporating a prototype snap-lock connection system designed for low cost high volume applications. Low cost is achieved through the use of larger diameter polymer optical fiber which facilitates the use of molded plastic components and simple crimp-on fiber terminations integrated receivers and visible red LEDs.