Forward-thinking your capacity

At home, we use it to listen to our CDs, watch digital TV and listen to digital audio tapes. At the office, it helps hard drives accurately store information. Space exploration and satellite communications rely on it extensively to safely transmit high volumes of banwidth over extremely long distances. Cable TV companies are thinking of using it to help increase the number of channels they can stuff into a piece of coaxial cable for households.

It is forward error correction, a noise reduction technique that is used everywhere to help with the transmission of digital data. Most people are unaware of this technique.

One of the largest applications for FEC is in the telecommunications market. Its ability to correct information offers network planners and design engineers a variety of options for maximizing network performance and minimizing cost. FEC allows service providers to guarantee exceptional signal quality because of the extremely low bit error rate that FEC provides by effectively eliminating the residual bit error rate floor.

For network planners and designers who seek to improve their networks as capacity demands grow at a breakneck pace, FEC is a versatile and cost-effective way to overcome network limitations of the past.

FEC: A brief history

Whether using fiber optics or free space as transmission media, all communication systems are affected by a common problem: noise. When introduced into a transmission medium, noise and other physical elements cause transmitted digital data to be lost or changed. FEC techniques repair the signal to enhance the quality and accuracy of the received information, improving system performance.

The telecom industry has used FEC codes for more than 20 years to transmit digital data through different transmission media. Claude Shannon first introduced techniques for FEC in 1948. These error-correcting codes compensated for noise and other physical elements to allow for full data recovery. In 1960 Irving S. Reed and Gustave Solomon, staff members at MIT's Lincoln Laboratory, developed the Reed-Solomon code, which has become the most widely used algorithm for error correcting. The Reed-Solomon code is well understood, relatively easy to implement, provides a good tolerance to error bursts and is compatible with binary transmission systems. However, because technology was being developed at a breakthrough pace, hardware development had to catch up to support digital transmission.

The first applications of FEC in the telecom industry emerged in the early to mid-1980s with the introduction of digital microwave radios. The benefits of these correction techniques when compared with their predecessor, analog radio, were extremely noticeable. In the 1990s, out-of-band FEC (OOB-FEC) began to be used in optical transmission systems for transoceanic or submarine networks. These submarine network applications had two goals: limiting the number of amplifiers in an undersea span and reaching very long distances. In 1995 OOB-FEC was used in repeatered spans in the Asian Pacific Cable Network in southeast Asia to reduce price and increase span distances by improving the quality of transmission on a 5 Gb/s single wavelength network.

But 5 Gb/s was just not enough, so new applications with higher bandwidth capabilities had to be designed and implemented. The need for bandwidth led to using OOB-FEC to upgrade a 5 Gb/s system that did not use FEC to a 3 x 5 Gb/s system with OOB-FEC. This application was used on the TAT-12 transoceanic network, which runs two cables from Great Britain and France to the United States.

Today, OOB-FEC at 2.5 Gb/s is used in many dense wave division multiplexing (DWDM) submarine networks, such as SMW3, which runs from Germany to the United States, and Gemini, which runs from the United States to Europe. In these networks, OOB-FEC is used to evaluate the margin on each wavelength independently and to equalize the margin between all wavelengths by first measuring the corrected errors on each wavelength. Larger system capacities, such as 10 Gb/s systems, are now being designed for the submarine networks. The 32 x 10 Gb/s Flag network, which runs between the United States and Europe, consists of four fiber pairs. Installation of this network will begin in the second half of 2000.

Future applications in submarine networks include extremely long-distance, high-capacity routes. Examples include the United States-to-China network, which is a 12,500-km, 8 x 2.5 Gb/s system, and the United States-to-Japan network, which is a 9000-km, 16 x 10 Gb/s system.

Alcatel's work on the submarine networks helped lead the ITU-T FEC standards and recommendations. From 1993 to 1996, the ITU-T Study Group 15 developed the recommendation for FEC in submarine networks. The recommendation is outlined in G.975 and was approved in November 1996.

New applications

Most recently, terrestrial fiber optic transmission systems began implementing FEC for DWDM networks. There are two types of FEC in today's terrestrial networks: OOB-FEC and in-band FEC (IB-FEC). While each type of FEC has its own advantages and disadvantages, OOB-FEC outperforms IB-FEC in large network designs.

IB-FEC - which was introduced in field applications in 1997 - was introduced into fiber optic networks to improve span distances, while OOB-FEC was introduced in DWDM systems to allow tremendously improved performance of multiple channel networks. The same OOB-FEC technology that was developed for submarine networks is used in DWDM terrestrial systems today.

Figure 1 shows that by adding FEC to the optical channels, carriers can reduce the number of intermediate line amplifier sites. This application is currently being designed into new DWDM networks so that new sites are unnecessary and existing sites can be bypassed to reduce equipment costs and maintenance.

With IB-FEC, unused bytes of the Sonet overhead are used to add the correction bytes. Using these available bytes in the Sonet overhead - which includes payload information - allows IB-FEC to achieve an additional 1 to 3 dB of optical signal-to-noise ratio tolerance over other signals that are not using any FEC. Longer optical span distances can be achieved by reducing the required optical signal-to-noise ratio by 1 to 3 dB. IB-FEC is vulnerable to changes in the Sonet standards because it uses the unused bytes. If the Sonet standards changed to require some of the currently unused bytes for new information, then the correcting capabilities of IB-FEC would decrease because it does not have as many additional bytes to conduct the necessary calculations.

OOB-FEC has a stronger error correcting capability than IB-FEC. While IB-FEC only allows users to correct eight or so errors per frame, OOB-FEC allows up to 1024 error corrections per frame. OOB-FEC does not use any of the Sonet overhead bytes but instead uses additional overhead bits that are added to the transmitted Sonet signal.

Adding bits increases the bit rate of the transmitted signal by 7%. In addition, OOB-FEC is not susceptible to changes in the Sonet standards such as IB-FEC, because it spares the Sonet overhead. OOB-FEC decreases the required optical signal-to-noise ratio tolerance by up to 9 dB over signals that have no FEC. Reducing the optical signal-to-noise ratio tolerance gives DWDM systems a tremendous improvement in performance, including:

- Quadruple the number of channels.

- Quadruple the bit rate (from OC-48 to OC-192, for example).

- Two to four times longer span distances between electrical regenerators.

- 30% to 40% longer spans between optical line amplifier sites.

- Greatly improved tolerance to optical path degradations.

OOB-FEC can also provide real-time performance management for carriers by allowing them to see how "hard" FEC is working. If there is an increase in the error-correcting rate, the operator will know that there is a problem in the network and can take corrective action before customers ever see any problems.

In the United States, OOB-FEC is designed to address all of these applications. Today service providers look for systems that will grow as their demand for capacity increases because of data traffic demands. Decision-makers are recognizing the ease of increasing bit rate or increasing channel count without redesigning the route. Systems can be upgraded from 2.5 Gb/s to

10 Gb/s, or from 8 x 10 Gb/s to 32 x 10 Gb/s without redesigning routes or in-line amp and repeater spacing using OOB-FEC.

For long distances, eliminating a site or extending the distance between electrical regenerators could be cost effective and therefore, OOB-FEC maximizes these distances. Using the IB-FEC or OOB-FEC users can maximize the distance between the end terminals or electrical regenerators as seen in Figure 2. This application is being used quite extensively in remote areas such as mountain ranges, or across the plains, where service providers have no requirement to add or drop traffic at intermediate sites.

Figure 3 shows how networks are being designed with no FEC at 2.5 Gb/s; when service providers are required to increase their network capacity to 10 Gb/s, they simply implement the OOB-FEC on the 10 Gb/s signals. This can be seen in the figure on a 40-channel network. This application allows the service provider to upgrade one optical channel at a time with no site reconfiguration necessary. Only the end terminals are upgraded. Dispersion compensation devices allow dispersion management throughout the 10 Gb/s network.

Another popular application allows a service provider to upgrade an existing eight-channel, 2.5 Gb/s network that is not using any FEC to a 32-channel, 2.5 Gb/s network using OOB-FEC (Figure 4).

Future FEC technologies, such as enhanced OOB-FEC for 2.5 and 10 Gb/s, are around the corner. They will allow for longer spans and increased channel counts.

Network planners and designers can benefit by using these different types of FEC. For example, FEC can potentially save tremendous costs in some large network designs. Today's networks need to be designed for large traffic demands, and the most efficient way to use DWDM networks in the long-haul market is to fill all the optical channels with 2.5 and 10 Gb/s equipment equipped with FEC.

In particular, OOB-FEC provides an unprecedented increase in performance to terrestrial networks. DWDM networks can be deployed at minimal bit rates and with minimal channel counts today, then upgraded to quadruple the size and capacity in the future without affecting the route design. Network providers should take the time to examine the tremendous benefits of OOB-FEC before designing a DWDM network. Their economic futures could hinge on the technology benefits realized by designing a network that uses OOB-FEC.