All fiber optic networks involve the same basic principle: information can be sent to, shared with, passed on, or bypassed within a number of computer stations (nodes) and a master computer (server). In addition to various topologies for networks, a number of standards and protocols have been developed, each with their own advantages, topologies, and medium requirements. This article discusses these standards and protocols, including: ATM, Ethernet, FDDI, Fibre Channel, ISDN, and SONET.
Asynchronous Transfer Mode (ATM over Fiber)
Asynchronous transfer mode (ATM) is widely deployed as a network backbone technology. This technology integrates easily with other technologies, and offers sophisticated network management features that allow signal carriers to guarantee quality of service (QOS). ATM may also be referred to as cell relay because the network uses short, fixed length packets or cells for data transport. The information is divided into different cells, transmitted, and re-assembled at the receive end. Each cell contains 48 bytes of data payload as well as a 5-byte cell header. This fixed size ensures that time critical voice or video data will not be adversely affected by long data frames or packets. ATM organizes different types of data into separate cells, allowing network users and the network itself to determine how bandwidth is allocated. This approach works especially well with networks handling burst data transmissions. Data streams are then multiplexed and transmitted between end user and network server and between network switches. These data streams can be transmitted to many different destinations, reducing the requirement for network interfaces and network facilities, and ultimately, overall cost of the network itself. Connections for ATM networks include virtual path connections (VPCs), which contain multiple virtual circuit connections (VCCs). Virtual circuits are nothing more than end-to-end connections with defined endpoints and routes, but no defined bandwidth allocation. Bandwidth is allocated on demand as required by the network. VCCs carry a single stream of contiguous data cells from user to user. VCCs may be configured as static, permanent virtual connections (PVCs) or as dynamically controlled switched virtual circuits (SVCs). When VCCs are combined into VPCs, all cells in the VPC are routed the same way, allowing for faster recovery of the network in the event of a major failure. While ATM still dominates WAN backbone configurations, an emerging technology, gigabit Ethernet, may soon replace ATM in some network scenarios, especially in LAN and desktop scenarios. A discussion of Ethernet follows.
Ethernet over Fiber
Ethernet began as a laboratory experiment for Xerox Corporation in the 1970′s. Designers intended Ethernet to become a part of the “office of the future” which would include personal computer workstations. By 1980, formal Ethernet specifications had been devised by a multi-vendor consortium. Widely used in today’s LANs, Ethernet transmits at 10 Mb/s using twisted-pair coax cable and/or optical fiber. Fast Ethernet, transmits at 100 Mb/s, and the latest developing standard, gigabit Ethernet, transmits at 1,000 Mb/s or 1 Gb/s. Figure 1 illustrates the basic layout of an Ethernet network.
Figure 1 — Basic Layout of an Ethernet Network
The formal Ethernet standard known as IEEE.802.3 uses a protocol called carrier sense multiple access with collision detection (CSMA/CD). This protocol describes the function of the three basic parts of an Ethernet system: the physical medium that carries the signal, the medium access control rules, and the Ethernet frame, which consists of a standardized set of bits used to carry the signal. Ethernet, fast Ethernet, and gigabit Ethernet all use the same platform and frame structure. Ethernet users have three choices for physical medium. At 1 to 10 Mb/s, the network may transmit over thick coaxial cable, twisted-pair coax cable or optical fiber. Fast 100 Mb/s Ethernet will not transmit over thick coax, but can use twisted pair or optical fiber as well. Gigabit Ethernet, with greater data rate and longer transmission distance, uses optical fiber links for the long spans, but can also use twisted-pair for short connections. CSMA/CD represents the second element, the access control rules. In this protocol, all stations must remain quiet for a time to verify no station in the network is transmitting before beginning a transmission. If another station begins to signal, the remaining stations will sense the presence of the signal carrier and remain quiet. All stations share this multiple access protocol. However, because not all stations will receive a transmission simultaneously, it is possible for a station to begin signaling at the same time another station does. This causes a collision of signals, which is detected by the station speaking out of turn, causing the station to become quiet until access is awarded, at which time the data frame is resent over the network. The final element, the Ethernet frame, delivers data between workstations based on a 48-bit source and destination address field. The Ethernet frame also includes a data field, which varies in size depending on the transmission, and an error-checking field which verifies the integrity of the received data. As a frame is sent, each workstation Ethernet interface reads enough of the frame to learn the 48-bit address field and compares it with its own address. If the addresses match, the workstation reads the entire frame, but if the addresses do not match, the interface stops reading the frame. Ethernet at all data rates has become a widely installed networks for LAN, MAN, and WAN applications. Its ability to interface with SONET and ATM networks will continue to support this popular network. In LANs, Ethernet links offer a scalable backbone, and a high speed campus data center backbone with inter-switch extensions. As a metro backbone in MANs, gigabit Ethernet will interface in DWDM systems, allowing long-haul, high speed broadband communications networks. Finally, Ethernet supports all types of data traffic including data, voice, and video over IP. Figure 2 illustrates a typical Ethernet deployment scenario.
Figure 2 — Switched, Routed Gigabit Ethernet Network
Gigabit Ethernet has emerged as a cost-effective alternative to ATM network structures. ATM has a greater cost, and he standards and products used to transmit ATM are still in flux, unlike the proven paradigm of Ethernet. In addition, system complexity is reduced in gigabit Ethernet, and because it works with existing Ethernet formats, the system does not require emulation software to act as a gateway between an Ethernet LAN and an ATM network. Table 1 outlines how Ethernet and gigabit Ethernet offer the same benefits of ATM.
Fiber Distributed Data Interface (FDDI)
FDDI usually finds placement as a high-speed backbone for mission-critical or high traffic LANs, MANs or WANs. Operating at a data rate of 100 Mb/s, FDDI was originally designed for optical fiber transmission. An unbroken FDDI network can run to 100 km with nodes up to 2 km apart on multimode fiber, and 10 km apart on single-mode fiber. However, a copper standard exists, known as a copper distributed data interface, or CDDI, although it is restricted to distances of only 100 m. Any one ring, copper or fiber, may contain as many as 500 nodes. FDDI’s niche is high reliability, the result of its counter-rotating ring topology illustrated in Figure 3. A dual-attached station connects the two paths via Port A, the primary path, and Port B, the secondary path. Port A may also have a number of M ports which attach to single-attached stations such as computer workstations. Information is passed around the FDDI ring via a token generated by the main station. The token moves around the ring until a requires access to the network. When a station needs to transmit information, it takes control of the token, and transmits in an FDDI frame, after which it releases the token, signaling that it has completed its transmission. Each FDDI frame contains the address of the station or stations that need to receive this frame. All nodes read the frame, but only to verify this address. If the node address and the FDDI frame address match, the station extracts the data from the frame and then retransmits it to the next node on the ring. When the frame returns to the originating station, that station strips the frame, and the network remains quiet until a node captures the token. A second generation network, FDDI-2 currently under development, supports the transmission of voice and video information as well as data. It uses a circuit-switched configuration in which a physical path is obtained for and dedicated to a single connection between two end-points in the network for the duration of the connection. In addition, another variation of FDDI, called FDDI full duplex technology (FFDT) uses the same network infrastructure but can potentially double data rates. If the secondary ring is not needed for backup, it can also carry data, extending the network’s capacity to 200 Mb/s. Work is underway to connect FDDI networks to the developing synchronous optical network (SONET).
Figure 3 — Dual Counter-rotating Ring Topology
Fibre channel, originally developed in the United Kingdom, was designed to provide high bandwidth (100 Mb/s), long distance connectivity (over several kilometers), and flexible topologies that allow the use of the same physical interface and media as existing channel and networking protocols. In fact, fibre channel was an attempt to combine the benefits of both channel and network topologies. Channels are closed, direct, structured and predictable mechanisms for data transmission. No decision making is required, allowing for a high speed, hardware intensive environment. Channels connect peripheral devices such as disk drives and printers, to a work station using protocols such as HIPPI or SCSI. By contrast, networks are unstructured and unpredictable in that much decision making is required to correctly route the data from one point to another.Fibre channel’s biggest impact has been made on storage devices, using an upper layer SCSI protocol. This gives fiber channel the ability to access mass storage devices more quickly and from a greater distance. Three main topologies include a point-to-point configurations, an arbitrated loop topology or a fabric topology. As a result, the most common of these three is the arbitrated loop, illustrated in Figure 4. In a fibre channel arbitrated loop (FC-AL), when a device is ready to transmit to the rest of the network, it first arbitrate for control of the loop. This is done via an arbitrate primitive signal (APBx), and each device in the network has its own APBx. It submits the signal to the network control, and the signal is looped around until the originating device receives its APBx, its signal that it has control and may begin transmitting. An open primitive signal allows the device to communicate with other devices in the loop by creating, essentially, a point-to-point connection between the two devices. All other devices in the loop simply repeat the data. A fabric topology represents the costliest configuration, because it requires a cross-point switch to connect multiple devices in a switched configuration. The benefit of this topology is that many devices can communicate at the same time; the media is not shared.
Figure 4 — Arbitrated Loop
Integrated Services Digital Network (ISDN)
ISDN has been designed to replace the standard telephone system and provide greater numbers of digital services to telephone customers, such as digital audio, interactive information services, fax, e-mail, and digital video. ISDN uses asynchronous transfer mode which can handle data transmission in both connection-oriented and packet schemes. As with regular telephone lines, the user must pay a fee for use of the line. Basic rate ISDN or BRI offers two simultaneous 64 kb/s data channels as well as a 16 kb/s carrier channel for signaling and control information. The combined data rate, 128 kb/s, allows for videoconferencing capabilities. Multiple ISDN-B connections further increase the data rate and the transmission quality. Primary rate ISDN (PRI) offers 30 channels (of 64 kb/s each), giving a total of 1920 kb/s. As with BRI, each channel can be connected to a different destination, or they can be combined to give a larger bandwidth. These channels, known as “bearer” or “B” channels, give ISDN tremendous flexibility. The original version of ISDN employs baseband transmission. Another version, called B-ISDN, uses broadband transmission and is able to support transmission rates of 1.5 Mb/s. B-ISDN requires fiber optic cables and is not yet widely available.
Synchronous Optical Network (SONET)
SONET is the American National Standards Institute (ANSI) standard for synchronous data transmission on optical media. The international equivalent of SONET is synchronous digital hierarchy (SDH). SONET provides standards for a number of line rates up to the maximum line rate of 39.808 gigabits per second and beyond. SONET is considered to be the foundation for the physical layer of the broadband ISDN (B-ISDN). Asynchronous transfer mode runs as a layer on top of SONET as well as on top of other technologies. The network defines optical carrier levels and their electrical equivalents, called synchronous transport signals (STS) for fiber optic transmission. The first step in the process involves multiplexing multiple signals by generating the lowest level or base signal, called STS-1. Its optical carrier counterpart is called OC-1, and it transmits at 51,480 Mb/s. Other levels operate from 155 Mb/s up to 40 Gb/s. The basic network elements include the terminal multiplexer (PTE), a regenerator (as needed for long distance transmissions), an add-drop multiplexer (ADM), for use in point-to-multipoint configurations, wideband digital cross-connects (W-DCS), broadband digital cross-connects, and the digital loop carrier. Together, these elements may be used in a point-to-point, point-to-multipoint (hub), or ring network configuration. Figure 5 illustrates a typical hub network configuration.
Figure 5 — SONET Hub Network
SONET provides a number of benefits over asynchronous systems. Its multiplexing technique allows simplified synchronous clocking and reduced back-to-back multiplexing, which reduces circuit complexity and cost. SONET’s optical interconnections meet a number of vendor requirements. The hub configuration adds greater flexibility to the system, allowing the convergence of a number of types of network protocols, ATM, Internet protocol, etc.