CS 650 - Networks: Project

The Synchronous Optical Network/Synchronous Digital Hierarchy (SONET/SDH) Protocol

The demands on network capacity have grown significantly as the principal uses have expanded from simple character based terminal emulation and occasional file transfer to current efforts to provide real-time video and audio. In response to these demands, optic fiber is used increasingly in preference to copper cable. Optic fiber has the potential to carry considerably more data with less loss than copper cable. However, standards are needed to insure that various networks can communicate. This need is met by the Synchronous Optical Network (SONET) standard. SONET is used by telephone companies to provide both voice and data services, by cable television companies to provide interactive TV and by LANs as a base for high speed Asynchronous Transfer Mode (ATM) networks.

This paper will explore how SONET works, how it is able to meet such diverse needs, its limitations and future potential and the differences from the highly similar Synchronous Digital Hierarchy (SDH) standard.

SONET/SDH most important characteristics are probably its ability to bind together signals of many different protocols and speeds into a single signal and its scalability in terms of transmission speeds and technology. These give SONET/SDH a present and a future. Other factors contributing to its success as a standard include built-in survivability and network management capabilities. SONET provides the underlying physical layer standard supporting many of the newer high-speed networking standards being proposed today. An understanding of this important standard will be helpful in understanding the higher layer protocols that will power the networks of tomorrow.

Research and publications on the SONET/SDH protocol have been extensive. Only a small fraction of the available material is referenced here.

SONET/SDH owes it beginnings to the breakup of AT&T in 1982. This created a situation in which a large number of networks operated by competing companies were required to interoperate with each other. Seven regional Bell companies were created which were not allowed to transport signals outside of their Local Access and Transport Areas (LATA). Transport between LATAs was to be provided by inter-exchange companies like AT&T, Sprint and MCI. The regional Bells must offer all inter-exchange companies equal access to their LATAs. As a result there were far more interfaces involving many different and competing companies. None of these companies were willing to allow AT&T to dictate the terms on which their signals would be transported. In England, Japan and Europe similar situations were developing [BERT97]

Berthold [BERT97] cites three technologies which provided the necessary prerequisites for the development of SONET/SDH. Primary, of course, has been the rapid growth in optical transmission performance which has grown exponentially over the last several decades. Also significant is the improvements in the Very Large Scale Integration (VLSI) that allow critical processing to take place at the chip level. Microprocessors allow distribution of control and remote management.

The original SONET standard was proposed in 1984 by Bellcore [CHIN93] on behalf of the regional Bell companies and other interested companies. [NORT97] The proposal had the following objectives:

• Compatibility among the various vendors equipment (the "mid-span meet").
• Synchronous networking.
• Improved network Operations, Administration, Maintenance and Provisioning (OAM&P).
• Better Add/Drop Multiplexing (ADM).
• Survivable network rings based on standards.
• Capability to handle new services, e.g., ATM.

[NORT97]

The original SONET definition was not acceptable to European telecommunications companies. The Europeans had developed their DS-1 equivalent later and consequently were able to achieve higher transmission rates because of improvements in technology. As a result the European base rate (E-1) contained 32 digital voice channels for a combined rate of 2.048 Mbps vs. the North American standard of 24 channels at 1.54 Mbps. [NORT97][]

In order to accommodate various regional requirements, particularly North America, Europe and Japan, a compromise was reached that established SDH as an umbrella standard with local variations. [CHIN93] The SONET base rate was increased to 51.84 Mbps and the SDH base rate was established at three times that, 155.52 Mbps. In addition, the multiplexing strategy for SONET was changed from bit-interleaving to byte-interleaving.[NORT97]

The principle difference between SONET and SDH relates to the bit rates defined for each. SONET is defined for bit rates as low as 51.84 Mbps while SDH begins at 155.52 Mbps. [HALS96] The two standards converge at STS-3/STM-1. [NORT97] The table below, taken from [HALS96] illustrates the comparative rates of each standard.

Note the difference in terminology. SONET transmission rates are categorized as STS-n or OC-n where STS stands for Synchronous Transport Signal and OC is Optical Carrier. STS defines the electrical equivalent of the OC signal. SDH speeds are categorized as STM-n where STM stands for Synchronous Transport Module. In each case, n is an n-multiple of the base rate (STS-1, OC-1 or STM-1). [NORT97]

The SONET frame structure is strongly influenced by the major users of the technology—the telephone companies. Its roots go back to earlier digital technology, digital voice. The digitization of voice is achieved by a process called pulse coded modulation (PCM). PCM consists of a series of amplitude samples taken at regular intervals. The sampling interval is based on an analog voice telephony bandwidth of 4kHz and the Nyquist sampling theorem, which states that a sample rate equal to or greater than twice the bandwidth of the analog signal will allow the signal to be accurately reconstructed. This implies a sample rate of 8000 times per second which produces an eight bit sample every 125 µsec. So it should come as no surprise that telephone companies would come up with a digital protocol based on frames of 125 µsec duration. [HALL96][OMID93]

The basic STS-1 frame structure of SONET is based on filling each 125 µsec frame with 9 segments of 90 bytes/octets each. Each segment is divided into 3 header bytes and 87 payload bytes. However, the first byte of the payload is reserved for path overhead. [BLAC97] The basic SDH frame structure is STM-1 which is 9 segments of 270 bytes of which 9 are header and 261 are payload. [HALS96] The key to SONET's flexibility, however, is the fact that the payload floats within a Virtual Container in the payload area. The frame contains a pointer to the Virtual Container. The pointer plus the Virtual Container is called an Administrative Unit. [OMID93]

SONET/SDH frames are generally depicted as tables of rows and columns with each segment beginning with its header data represented as a row. Header data is represented by columns on the left side of the table. Actual transmission sequence is left to right, row by row.

The basic bit rate for SONET STS-1 is determined using the frame structure:

Other transmission rates are multiples of the basic STS-1 rate. For example, STM-1 is equivalent to STS-3, 155.520 Mbps. Multiple rates normally retain all of the information in the STS-1 overhead. However, there is a provision for concatenated rates which retain a single path overhead and are able to carry larger payloads. Such lines are indicated with a 'c', e.g., STS-3c. [MINO93]

Transmission rates below STS-1 are transported in a Virtual Tributary structure. There are predefined mappings for DS3, DS1 and DS0. Other Digital rates are supported as well. DS1 signals are mapped to a Virtual Tributary of 9 X 3 octets which are entered into the 9 X 87 synchronous payload envelope (SPE). Therefore 28 DS1 signals could be transported over a single STS-1 line. [MINO93]

Three levels of Virtual Tributaries are defined: VT-6, VT-2 and VT-1.5 corresponding to DS-2, E-1 and DS-1 respectively. The STS SPE is divided into Virtual Tributary groups, each of which may contain one VT-6, three VT-2s or 4 VT-1.5s but not a mixture. On the other hand, the STS SPE can contain various types of VT groups. [NORT97]

Some of the key overhead data found in the headers and path overhead include [MINO93]:

• Section overhead

• Framing bytes - constant start of frame data (F628 Hex).
• STS-1 Identification - identifies a specific STS-1 signal within an STS-N signal.
• Section BIP-8 - bit-interleaved parity 8 code - even parity for error monitoring.
• Orderwire - provides communications between regenerators, hubs and remote terminal locations.
• Section User Channel - Network provider's use, terminated at section level equipment.
• Section Data Comm Channel - Communications between section terminating equipment.

• Line overhead

• Pointer - Payload offset pointer.
• Pointer Action Octet - buffering.
• Line BIP-8 - Line error monitoring.
• APS Channel - Automatic Protection Switching.
• Line Data Comm Channel - Network administration communications.
• Orderwire - Communications between line entities.

• Path overhead

• STS Path Trace - Constant string that allows path terminating equipment to verify connection.
• Path BIP-8 - Path error monitoring.
• STS Path Signal Label - Information about the construction of the SPE.
• Path Status - Information back to originating PTE.
• Path User Channel - Network provider use.
• Multiframe - Frame status.

Key concepts underlying SONET are Time-Division Multiplexing (TDM) and Add-Drop Multiplexing (ADM), Digital Cross-connect switching (DCS) and byte interleaving. TDM enables multiple signals to be carried in the same frame. Because of byte interleaving, the ADM, can extract or add individual signals without having to multiplex/demultiplex the entire frame. [BLAC97]

SONET networks consist of a number of elements:

Add-Drop Multiplexers (ADM) - ADM may be used at intermediate or terminal sites to add or drop signals as required. Only the signals required to be processed by the site are affected. Other signals pass through untouched.

Broadband digital cross-connect - This element performs STS-1 level switching. The difference between this cross-connect and an add-drop multiplexer is the much larger number of STS-1 signals that the cross-connect can interconnect. These switches are used in SONET hubs for broadband traffic management and grooming of STS-1 signals. Grooming is the process of segregating and consolidating signals to improve network efficiency.

Wideband digital cross-connect - This element performs Virtual Tributary level switching similar to the broadband digital cross connect. Switches of this type are used for DS-1 level grooming. Similar other SONET elements, only the affected tributaries are accessed and switched by the wideband digital cross-connect.

Digital loop carrier terminal - This element provides remote connection between users in a Carrier Service Area (CSA) and a SONET network. The CSA is served by copper pairs.

Switch interface - This element provides switching at the DS-0 level.

[NORT97]

According to Berthold [BERT97] the major architectures for SONET networks are point-to-point, chains, self-healing rings and digital cross-connect switch.

Point-to-point links. This is a linear connection between two points requiring a dedicated channel. The overall circuit is called the path and the equipment on either end that sends or receives a signal is called path-terminating equipment (PTE). The portion of the network occurring between two regenerators is called the regenerator section, or just the section. The portion between multiplexers is the multiplex section, or more commonly, the line. [BERT97] As can be seen from the overhead description above, each of these subdivisions has its own control information built into the frame structure.

Point-to-point topologies imply the termination of the entire SONET payload at either end and are typically used to provide a single system/single route solution. Point-to-point configurations can be made more survivable by adding a geographically diverse protection path. The reach of a point-to-point system can be extended by the use of regenerators or optical amplifiers. [NORT96]

Chain. This is similar to the point-to-point, but with the addition of intermediate Add/Drop Multiplexes This allows intermediate nodes on the lines. [BERT97] This enables additional traffic to be added or dropped off along the way without having to multiplex/demultiplex traffic that is just passing through. This configuration can also achieve increased survivability through the use of geographically diverse protection paths. Additionally, a "path protected linear ADM route" can be created by combining the linear chain with a subtending ring (see below). [NORT96]

Self-Healing Rings. SONET rings are able to respond to breaks and node failures by rerouting network traffic. [BERT97] This capability is a major selling point for SONET [MORR97] but it is not limited to SONET. [WU94] SONET has three basic types of rings: unidirectional path switched ring, two-fiber bi-directional line switched ring or four-fiber bi-directional line switched ring. [NORT96]

Unidirectional path switched rings are a closed-loop transport architecture typically used in access networks. Each node is connected to adjacent nodes by a pair of optic fibers. On one fiber, the working signal travels in a clockwise direction around the ring, while on the other, a protection path signal travels in a counter-clockwise direction. Thus the working traffic is unidirectional. If a line or node failure is detected, the ring reverts to a bi-directional linear add-drop topology. [NORT96]

Unidirectional path switched rings are frequently connected to other rings using matched node gateways. They also support virtual ring configurations in which other ring types are treated as nodes. [NORT96]

Two-fiber bi-directional line switched rings are made up of network elements connected in a closed loop of bi-directional two fiber spans. Half of the available bandwidth on each fiber is used for working traffic and half for protection. Traffic entering at any node can travel in either direction around the ring. Once the traffic is delivered, the channel can be reused. If a node or line failure is detected, the automatic ring protection switching loops the signal back over the protection path. If service is restored, the routing reverts to normal. [NORT96]

The reusable bandwidth feature of the bi-directional line switched ring leads to capacity advantages over the unidirectional path switched ring. Depending on the ring configuration, capacity can be over 300 percent greater for the same line rate. [NORT96]

Four-fiber bi-directional line switched rings are similar to two-fiber bi-directional line switched rings, but use four fibers to connect adjacent nodes. The additional fiber provides two advantages: double the capacity and additional protection switching modes. One pair of fiber transports the working signal and one pair provides protection. Each pair is bi-directional. As a result of this configuration, the response to a line failure can be a span switch similar to that performed by a point-to-point connection, or, in a more serious breach, a loop-back similar to that performed by the two-fiber bi-directional line switched ring. [NORT96]

Subtending rings are a dual ring configuration in which a secondary ring is fed by a tributary node. The subtending ring may employ any of the switching topologies described above. [NORT96]

Folded rings enable the functionality of unidirectional path switched rings and bi-directional line switched rings where the existing fiber will not support true geographical diversity. They consist of a series of rings with either side of the ring sharing a common conduit. Folded rings allow a ring structure to be adopted before all of the needed fiber is in place. [NORT96]

Matched node gateways provide survivable connections between adjacent rings. This configuration consists of a primary and a secondary gateway with redundant routing. [NORT96]

Digital Cross-Connect Switch. This is an arbitrary pattern of switches which can support a number of SONET signal rates. [BERT97]

One of the key features of SONET is the provision of bandwidth in the frame overhead structure to support management of the network. Although, not all of the standards for this capability have been established, SONET has the advantage over asynchronous networks that where ever the SONET signal is passed, overhead communications are maintained. [NORT97]

SONET uses a single highly accurate Primary Reference Source (PROS) clock. Currently, most networks in the United States use Building Integrated Timing Supply (BITS). BITS provides a bit clock and a byte clock in the form of a 64/8 composite clock. Both clocks must be synchronized if a signal is to be read correctly.[BLAC97]

The internal clock of one terminal in a SONET network derives it timing from the BITS, which is also used by switching systems and other equipment. The outgoing OC-n signal provides timing to the other terminals. This is called "loop timing." [NORT97]

ATM and SONET are both complementary and competing protocols. ATM provides superior switching capability, relegating SONET to the role of large scale permanent circuit type connection. [BERT97] But many organizations are finding that the combination of the flexibility and intelligence of ATM and the speed and survivability of SONET are perfect for consolidated data, voice and video wide area networks. [MORR97] The reusable bandwidth feature of SONET bi-directional line switched rings allows networks to take advantage of ATM's cell structure and switching capability. [NORT96]

According to Federal Computer Week, the Department of Defense is in the process of building the "world's largest private SONET network." The Defense Information Systems Network will use ATM to integrate data, video and voice from existing networks. Advantages attributed to SONET include: bandwidth, scalability, reliability and quick recovery. Advantages attributed to ATM were: dynamic bandwidth allocation and visibility of the whole network. [MORR97]

On a smaller scale, Time Warner Cable in Florida is experimenting with a concept they call Full Service Network (FSN). This service offers video, shopping and news on demand as well as regular cable services. This system uses SONET over optic fiber to the neighborhood node and then coaxial cable to the subscriber's home. ATM is used throughout. [BLAN95]

Organizations whose primary need is for data transport might consider Cisco Systems, Inc., new Gigabit Switch Router to support IP over SONET and skip the ATM layer. This approach avoids the ATM overhead, but loses the flexibility. [MORR97]

Byte Magazine reports that Media One plans to use IP over SONET to provide a backbone network to support its advanced cable services. Best Internet, an ISP, dropped ATM in favor of Border Gateway Protocol, a TCP/IP routing protocol. However, Best's SONET link is point-to-point and does not take advantage of ATM's switching capability. [MACE97]

RFC 1619 [SIMP94] describes a method for running Point-to-Point Protocol (PPP) over SONET/SDH. As a baseline, PPP over SONET runs on an STS-3c line with 149.760 Mbps of bandwidth available for payload. Lower bit rates are required to use Virtual Tributary mapping. Higher bit rates should use SDH STM rates to simplify multiplexing and integration.

One promising advance in optical communications is wavelength-division multiplexing (WDM). By allocating different wavelengths to various sets of signals, each wavelength will have the same data capacity as a single SONET signal now. In other words, an STM-1 signal with three wavelengths would have a bandwidth of 466.560 Mbps. [WILL97]

WDM can be unidirectional or bi-directional, with signals going in both directions on a single fiber. There is also wideband, narrowband and dense WDM. Wideband and narrowband WDM double the channel capacity, while dense WDM can increase it as much as eight times. [NORT96]

This is secondary research that was conducted using a regular library search aided by an internet bibliography search and a pure internet search. The library revealed several texts on fiber optic networks and a relatively complete collection of ACM and IEEE publications. A number of ACM and IEEE publications were also available for download over the internet, and in some cases these were used even though the same publication was available in the library. Searches on 'sonet' using Galileo, The Collection of Computer Science Bibliographies and Hypatia each turned up over fifty entries - many of which were either over five years old or not available locally. A search on 'sonet' using HotBot turned up several handbooks by NORTEL which proved very useful. Since there seemed to be a significant amount of reasonably new material available, I did not pursue difficult sources.

SONET is a powerful protocol which is extensively used for large and high performance networks. The cost appears to match its power. It is not something you will find running in a local insurance agency or doctor's office. It is, however, the solution chosen by the Department of Defense to run the DISN, a large wide area network for data, voice and video and on a smaller scale it is the network chosen by Time Warner, Inc. to implement their "Full Service Network". SONET's compatibility with ATM, its network management capabilities, and its ability to support survivable topologies make the future importance of SONET as a data transport likely.

[BERT97] Berthold, Joseph E. "SONET and ATM" in Optical Fiber Telecommunications IIIA. San Diego: Academic Press, 1997.

[BLAC97] Black, Uyless and Waters, Sharleen. SONET & T1: Architectures for Digital Transport Networks. Upper Saddle River, NJ: Prentice Hall, 1997.

[BLAN95] Blank, Christine. "The FSN challenge: Large-scale interactive television," Computer, 28:5 (9-12), May 1995.

[CARP97] Carpenter, Tamra J., Steven Cosares and Iraj Saniee. "Demand Routing and Slotting on Ring Networks," DIMACS Technical Report 97-02, January 1997.

[CHIN93] Ching, Yau-Chau and Say, H. Sabit. "SONET Implementation: Does the status of SONET deployment meet the original expectations of the systems's developers?" IEEE Communications Magazine, 31:9 (34-40), September 1993.

[CYPH93] Cypher, David and Shukri Wakid. "Standardization for ATM and Related B-ISDN Technologies," StandardView, 1:1 (40-47), September 1993.

[FURH95] Furht, Borko, Deven Kalra, Fredrick L. Kitson, Arturo A. Rodriguez and William E. Wall. "Design Issues for Interactive Television Systems," Computer, 28:5 (25-39), May 1995.

[HALS96] Halsall, Fred. Data Communications, Computer Networks and Open Systems. Harlow, England: Addison-Wesley, 1996.

[MACE97] Mace, Scott. "ATM's shrinking role," Byte, 22:10 (59-62), October 1997.

[MINO93] Minoli, Daniel. Enterprise Networking: Fractional T1 to SONET, Frame Relay to BISDN. Boston: Artech House, 1993.

[MORR97] Morrissey, Jane. "ATM rides high atop Sonet" in A Supplement to Federal Computer Week. November 1997.

[NORT97] NORTEL. "SONET 101: An Introduction to Synchronous Optical Networks," October 1997.

[NORT96] NORTEL. "Introduction to SONET Networking," October 1996.

[OKAM77] Okamoto, Satoru, Kimio Oguchi and Ken-ichi Sato. "Network Architecture for Optical Path Transport Networks," IEEE Transactions on Communications 45:8 (968-977), August 1977.

[OMID93] Omidyar, Cambyse Guy and Aldridge, Anne. "Introduction to SDH/SONET," IEEE Communications Magazine 31:9 (30-33), September 1993.

[SIMP94] Simpson, W. "PPP over SONET/SDH," RFC 1619, Network Working Group, May 1994.

[VETT95] Vetter, Ronald J. "ATM Concepts, Architectures, and Protocols," Communications of the ACM 38:2 (30-38, 109), February 1995.

[WANG97] Wang, Chorng-kuang and Po-Chiun Huang. "An Automatic Gain Control Architecture for SONET OC-3 VLSI," IEEE Transactions on Circuits and Systems—II: Analog and Digital Signal Processing, 44:9 (779-783), September 1997.

[WILL97] Willner, Alan Eli. "Mining the optical bandwidth for a terabit per second," IEEE Spectrum, 34:15 (32-41), April 1997.

[WORS97] Worsley, Debra J. and Tokunbo Ogunfunmi. "Isochronous Ethernet - An ATM Bridge for Multimedia Networking," IEEE Multimedia, January-March 1997.

[WU94] Wu, Tsong-Ho. "A Passive Protected Self-Healing Mesh Network Architecture and Applications," IEEE/ACM Transactions on Networking, 2:1 (40-52), February 1994.