The long and short of fiber links
For long- and short-haul applications, forward error correction can help provide the performance carriers require.
SIMON KEETON, Vitesse Semiconductor
To stay competitive, today's carriers must be able to accommodate a wide variety of protocols and data rates over their backbone optical-fiber networks. Today's customers want transport for ATM or SONET/SDH frames and even packet-based streams like Internet protocol (IP) and Gigabit Ethernet.
Despite auspicious beginnings and a well-entrenched following for optical backbone transport, SONET/SDH has endured a number of transformations and additions, making it a bit burdensome for packet-based data-transport or optical-networking needs. Instead of trying to fit everything neatly into SONET/ SDH frames, carriers need a different way to transport various signal types over their networks. What's required is a universal framing and formatting structure, one that is protocol- and rate-independent.
Carriers and equipment OEMs alike are looking toward future deployment of the universal optical transport network, and to date, the most appealing approach from both cost and backward-compatibility standpoints is a proposed time-division multiplexing "digital wrapper" protocol.
Figure 1. Forward-error-correction (FEC) frame structure as defined by ITU G.975 using Reed Solomon (255,239) algorithm. This G.975 frame is composed of three blocks-framing structure (overhead), data (payload), and redundant data (FEC check bytes). This frame can be logically organized into 128 subframes consisting of 1 bit of overhead, 238 bits of payload, and 16 check bits. The frame is transmitted column by column, from top to bottom and left to right as indicated. The result is 16 bytes of overhead, 3,808 bytes of payload, and 256 FEC check bytes.
Architectural details of the optical transport network are still being defined, but much of the groundwork has already been completed and is contained in the current version of the ITU-T G.872 recommendation. Some implementation details are contained in ITU-T G.709-a working draft that defines the "Network Node Interface for the OTN [optical transport network]." Essentially, the working draft represents the optical network-to-network interface for short-reach links up to 40 km. The interface is slated to use Reed-Solomon (255,239) error-correction coding. This forward-error-correction (FEC) mechanism, which is both protocol- and rate-independent, provides the required overhead bytes for a digital wrapper. The work represents the critical first step toward implementation of a standardized, universal optical transport technology.
The initial proposals for the digital-wrapper standard roughly mirror the ITU G.975 frame (see Figure 1), which employs FEC based on Reed-Solomon (255,239) error-correction coding. This algorithm results in a 16x255-byte optical-channel frame that incorporates 16x16 redundant FEC check bytes to accompany every 16x238 bytes of payload, plus an additional 16 bytes of unused overhead. Taking that a step further, four 16-row "subframes" are grouped together to form a "super-frame," which provides 64x238 bytes of payload information, 64x16 FEC bytes, and 64 bytes of additional overhead (see Figure 2).
The proposed digital wrapper overhead will impart many SONET/SDH-like features but will be more universal in nature, allowing carriers to easily accommodate a wide range of "speeds and feeds." Digital wrapper in fact may represent the veritable "Holy Grail" of protocol-independent transport mechanisms for equipment vendors or carriers trying to win in the short-haul transport space. Unfortunately, the Reed-Solomon (255,239) algorithm isn't a be-all, end-all solution for long-haul applications, where squeezing out the longest distance between transponders at minimal cost is key.
The same functions and operations that occur in a SONET/SDH overhead must occur in the optical transport network. In the SONET/SDH world, transport overhead is essential to providing operations, administration, maintenance, and provisioning, including things like framing, bit-interleave parity calculations, trace identifiers, and protection switching. The 64 bytes of unused overhead resulting from FEC based on Reed-Solomon (255,239), dubbed the "digital wrapper," are used to implement these kinds of SONET/SDH-like functions.
As data signals move from short-haul metropolitan networks to long-haul networks (or vice versa), performance monitoring becomes a crucial capability. Carriers must be able to assess the incoming signal's integrity as well as what's being sent out, and to do that, the performance of the link must be monitored-a well-worn function in SONET/SDH. But without SONET/SDH, another mechanism must be used to monitor the performance or quality of the signal. FEC provides this ability.
When a signal is FEC encoded and then later decoded at the other end of the link, a count is tabulated for the number of digital ones and zeros corrected in the output signal. In essence, this count corresponds to the protocol-independent bit-error rate (BER) or degradation factor of a given transport channel. So regardless of the protocol or data rate of the signal being transported, FEC enables link performance to be monitored. In this way, signals like HDTV, IP, or Gigabit Ethernet can use FEC's detection and correction capability to flag the number of errors generated by the link, determine relative BER performance, and assess signal integrity.
Further, FEC enables performance monitoring to be carried out with or without the incorporation of a standardized digital-wrapper technology. Since FEC requires adding a constant number of overhead bytes per given amount of transport, some unused overhead invariably is freed up for other purposes. This unused overhead can be used for anything, including proprietary digital wrappers or signaling mechanisms.
Since FEC is applied on a per-channel basis, every channel can then be provided with its own independent signaling mechanism. While this notion leads to the development of proprietary protocols, with long-haul backbones, where transponder-to-transponder link length and low cost are primary objectives, it may be the only means of developing the optimal solution.
FEC is a superb way to provide overhead for a digital wrapper, and for the short-haul, it is sometimes applied merely for its framing and formatting structure. On the other hand, for long-haul applications, FEC's most important benefit is its detection and correction capability, which improves performance by effectively increasing the optical gain of the link (see sidebar, "Optical transport fundamentals: Distance, capacity, and cost").
For the long haul, standards take a back seat to the primary goal, which is transporting signals from one point to another with maximum spacing between costly repeaters or amplifiers. For this reason, operators of long-haul applications such as submarine or long-distance terrestrial links are less likely to be interested in digital wrapper as their short-haul brethren.
For those who see FEC as simply a means to increase link performance, digital wrapper's Reed-Solomon (255,239) is but one of many possible algorithms. In fact, algorithms other than Reed-Solomon (255,239), which expends about 7% of the channel's data-carrying capacity as overhead, can improve the effective optical gain and maximum link transport distance. The tradeoff for this higher performance is a higher cost in overhead, often as much as 15% or more. Long-range applications generally still carve out a portion of overhead for signaling and performance monitoring, but performance of the FEC code becomes paramount.
Figure 2. Lambda frame structure for short-haul Inter-Domain Interface using the Reed Solomon (255,239) FEC algorithm. This lambda frame consists of three blocks-overhead, payload, and FEC. Rows 1 to 4 of columns 1 to 16 of the frame are reserved for the transport of overhead, while rows 1 to 4 of columns 17 to 3,824 are for payload, and rows 1 to 4 of columns 3,825 to 4,080 are for FEC. The structure of the lambda frame is derived from the G.975 framing structure.
Robust FEC codes handle lower-quality input signals (higher BER) much more effectively. For example, some FEC codes can easily handle an input BER greater than 1.2x10-2, yet provide exceptional output quality on the order of 10-15 or less. The better the FEC code, the more errors the input data stream can have and the farther the signal can be sent before repeaters or amplifiers are needed. Often, performance may be improved by simply increasing the amount of overhead available to a given error-correction coding scheme.
At OC-48 speeds (2.5 Gbits/sec unencoded), FEC is not an absolute requirement. But with dropping hardware prices, carriers now see great economic benefit in moving to OC-192 speeds (10 Gbits/sec unencoded). At these data rates, FEC becomes a critical capability to maintain acceptable signal-to-noise levels-that is, if currently available technologies are employed. At OC-192, effective signal energy is reduced by a factor of four versus OC-48.
In addition to being an active participant in furthering the standards effort for digital wrapper, Vitesse is actively working on products that utilize advanced FEC coding algorithms that greatly surpass the strength of the ITU's G.975 algorithm's defined frames, to provide the improved optical gain demanded for future very-high-speed, long-haul applications.
Short-haul networks can take advantage of the digital-wrapper standards. As the number of protocols and data rates continue to grow, short-haul metropolitan carriers need transparent management solutions, and digital wrapper provides that solution. Adoption of digital wrapper as a standard for the long haul may be slower in coming, at least as long as minimizing costs and maximizing link distances are key concerns.
In the near term, there's likely to be a proliferation of proprietary implementations to protocol-independent transport. On the other hand, for equipment OEMs and carriers that want a standardized, universal approach, the optical transport network's digital wrapper may be the cure.
As the optical transport network hastens toward reality, the critical enabler to moving traffic between different transport domains (e.g., from a short haul to long haul, or between different carrier's networks) will be the ability to monitor signal or channel quality. In lieu of available SONET/SDH performance monitoring, FEC represents a viable approach that is both transparent and protocol-independent, with the added benefit of increased optical gain for longer transport distances.
From a systems standpoint, the application of FEC facilitates the use of multiple carriers' networks for end-to-end transport over long distances. Customers demand link performance information as their signal traverses the network, as well as accurate predictions about quality at the far end. Also, carriers along the network path will need to know ingress signal quality to make guarantees about the integrity of outgoing signals. Performance monitoring represents the key enabler to realizing the emergence of future optical transport networks.
Simon Keeton is a product-line director at Vitesse Semiconductor Corp. (Camarillo, CA).
- International Telecommunication Union (ITU-T) "Series G: Transmission Systems and Media, Digital Systems and Networks: Forward error correction for submarine systems," ITU-T Recommendation G.975 (1996).
- Brungard, Deborah A., "Draft ITU-T G.709 Network Node Interface for the OTN," AT&T (2000).
- Schmitt, Andrew, "Forward error correction advances optical-network performance," Lightwave (1999).
- Schmitt, Andrew, "New components enable management functions in optical network," Lightwave (1998).
This article appeared in the July 2000 issue of Integrated Communications Design, a sister publication.
Just as speed-power product is popular for gauging IC performance, distance versus capacity provides a useful baseline for comparing optical-networking hardware. Combining distance-capacity product with costs provides an excellent means of understanding the economics within the telecommunications arena.
Optical link degradation is largely the result of two effects: optical attenuation and dispersion. Optical attenuation describes the loss of optical energy as a signal passes through a fiber. Essentially, the light gets "darker" as it passes through the fiber. On the other hand, dispersion is the deformation of the tight pulse of light wavelengths transmitted by an optical laser-typically, stretching it out and making it harder to resolve.
Both attenuation and dispersion are well understood and can be managed effectively by applying current technology to the design of links. Attenuation is deterred by adding regenerators, which are limited by the sensitivity of the receiver. Dispersion can be combated by using dispersion-compensating fiber, but attenuation is also increased. Recent developments like the erbium-doped fiber amplifier (EDFA) have shifted focus away from regenerators. EDFAs are a primary enabler for WDM technology but generate some noise themselves, such as amplified spontaneous emission noise.
Other factors come into play as wavelength spacing gets tighter and bit rates increase. Four-wave mixing is particularly troublesome for WDM systems, while other nonlinear effects such as polarization-mode dispersion, self-phase modulation, and scattering effects can further complicate link engineering. Unfortunately, compensating for one factor inevitably leads to increasing the effects of another and increases the cost of the system.
Optical signal-to-noise ratio (OSNR) provides an excellent abstraction of all optical noise effects. OSNR is the ratio between the received signal and combined noise added by all factors on the optical link.
A direct relationship exists between the OSNR and bit-error rate (BER) of a given link-higher OSNR leads to a lower BER, and vice versa. For obvious reasons, optical-network engineers want a low BER, typically on the order of 10-12 to 10-15, and the required bit rate can be extrapolated directly to provide the necessary OSNR.
If the measured OSNR is higher than required, excess margin exists on the link, which can be used to transmit more wavelengths, go a longer distance, or increase the bit rate of transmitted data-all of which improve the distance-capacity product. Since OSNR margin can be "spent" to improve distance-capacity product, great competitive benefits are to be had by improving OSNR-that is, if it can be done in an economical fashion.
Fiber-enhancement technologies and improved amplification techniques represent an enormous effort to improve the problem, albeit at a relatively high cost. An alternative, more cost-effective solution is FEC, which provides improved BER through the use of mathematics.
Lightwave is a monthly international publication focusing on fiber optics and optoelectronics, the technologies driving the growth, convergence, and improved performance of telephony, computer communications, and video. Lightwave pro vides technology news as well as applications and product information for corporate and technical managers and staff engineers. Lightwave's editors emphasize analysis and interpretation in their reports on the technological impact of fiber-optic components, systems, and networks in these markets.