100G Metro Network
Overlaying flourishing 10 Gb/s services with additional co-propagating 10 Gb/s channels in different colours, or wavelengths, is already common practice. Network operators eager to increase capacity have begun overlaying 40 Gb/s channels onto available fiber without leasing more dark fibers as the best pragmatic approach. With the advent of 100G technology, the interest is now shifting from 40G to 100G installations. This paper provides a primer in 100G technology developments and examines important parameters and prerequisites supporting practicable 100G metro network solutions. An example of complementing the existing 10 Gb/s services with new 100 Gb/s WDM channels is also discussed.
Direct or Coherent Detection: Which one is for Metro?
The transmit signal strength is limited by laser heat dissipation and power consumption. Consequently, significant research and development efforts have been undertaken to improve the sensitivity of the receiver. Two technology alternatives exist: direct detection or coherent detection.
Predominantly suited to trans-oceanic submarine or terrestrial long haul applications, the performance of coherent detection is undeniably superior to that of direct detection. In the case of coherent detection, the opto-electric conversion process is linear. Thus the phase information embedded in the optical signal is preserved permitting the straight forward electrical compensation of fiber linear effects including chromatic dispersion (CD) and polarisation mode dispersion (PMD). However, the hardware required to perform coherent detection is somewhat more elaborate and comprises a local oscillator, a 90 hybrid module necessary to discriminate the phase quadrature of the received optical signal, and four balanced photodiodes to detect the signal from a single polarisation as seen in Fig. 2.
A direct detection technique, on the other hand, requires only a delay interferometer and two single photodiodes translating into a device of much lower cost and complexity. The advantage of longer transmission distances (in the range of several 1000 km) enabled using a coherent approach is overkill for deployments of much shorter reaches - ranges of several 100 km. The direct detection technique is thus seen as an attractive alternative for metro networks, enterprise and datacenters where economic viability plays a vital role.
Taking the minimalist approach, direct detection works without additional equipment in most cases. When the application range approaches the limit of direct detection technology, the simple addition of dispersion compensation and amplification may be used to augment signal detection and extend reach.
Similar to 10G multiplexed networks, 100G networks begin with the use of 100G transceivers and dense wavelength division multiplexing (DWDM) multiplexers or de-multiplexers. Depending on the application scenario, 100G networks may require optical amplifier, dispersion compensation modules and FEC enabled transponder cards. In this section, we delve deeper into the details of these components.
The 100G DWDM-CFP transceiver is currently specified for applications associated with distances of up to 20 km. The 100G CFP uses four lasers, each tunable over the DWDM grid, such that the resulting link realises 4 x 25 Gb/s with each 25 Gb/s travelling in a separate single mode duplex fiber.
The DWDM multiplexer or de-multiplexer is the core technology providing a platform for adding essentially unlimited co-propagating channels to existing 10G and 40G channels with the 100G upgrades and eventually making it possible to upgrade all channels to 100G. In DWDM, multiple wavelengths are used to carry the optical signals through the same fiber leading to higher spectral efficiency per fiber. Pass band bandwidth, pass band flatness, stop band rejection ratio, transition slope from pass band to stop band and insertion loss are the key specifications determining selection of multiplexers and de-multiplexers.
Traditional thin film optical filters based DWDM multiplexers provide excellent wavelength selectivity. However, since these are free-space optical devices, much care must be taken to precisely align individual optical sub-components to achieve perfect collimation of the optical beams. Additionally, complexity increases with an increase in channel count.
Thus, arrayed waveguide gratings (AWG) based on planar light wave circuit (PLC) technology, are widely used for multiplexing large numbers of channels. The AWGs are composed of integrated multiple waveguide delay lines to render multipath optical interference. As depicted in Fig. 4, an AWG consists of input and output waveguides, two star couplers and an array of waveguides inter-connecting the two star couplers. Each waveguide within the array manifests a slightly different optical length, leading to different interference at the output when the differing paths recombine. The AWG can be built into a 19" rack type installation.
Depending on the transmission power budget available from the transceiver (e.g., current DWDM 100G CFP are specified to a power budget equivalent to about 20km), an optical amplifier can be introduced to overcome the additional loss encountered from the fiber attenuation, optical power splitting, and other factors. For a DWDM application, an erbium-doped fiber amplifier (EDFA) is the best choice as it provides high gain, high optical signal power, low crosstalk between wavelength channels and convenient optical coupling from and into optical fibers.
The EDFA is composed of an erbium-doped fiber, a pump laser, an optical isolator, and a wavelength-division multiplexer. The amplifier needs to have a wide optical bandwidth and flat optical gain within the WDM bandwidth. Depending on the application and length, the gain of the amplifier has to be adjusted so that the power at the receiver does not exceed the receiver sensitivity limits and destroy the transceiver.
The phenomenon of temporal broadening of the signal while the signal travels through the optical fiber is termed dispersion. A DWDM 100 Gb/s CFP specifies a typical dispersion tolerance of 400ps/nm which typically permits a reach of 20 km without compensation. This optical phenomena causes overlapping of pulses called inter-symbol interference (ISI). Due to ISI, the detector cannot detect the data correctly, increasing the number of error bits and consequently limiting the transmission distance feasible.
The dispersion compensating fiber (DCF) module is widely used to overcome the obstacle of ISI. A DCF module consists of a special fiber inducing a large value of negative (normal) dispersion at the 1550 nm wavelength window. The advantage of DCF for DWDM applications is that the DCF may be applied over a wide wavelength window and allows tailoring the slope of chromatic dispersion to match with SMF-28 for C-band channels.
Forward error correction (FEC) is a means of mitigating errors in data transmission to effectively provide OSNR enhancements and thereby extending the transmission reach of high data rate optical systems. Redundancy is added into the signal in terms of additional control bits which are used at the receiver to identify and then correct errors in the received signal. FEC coding can be implemented in the transceiver itself or in the transponder line card hosting the transceivers.
Complementing the existing 10 Gb/s system with 100 Gb/s upgrades
The phenomenal growth in new applications such as cloud services, telemedicine, video on demand etc., has ushered in the unprecedented outburst in data rate migration encompassing the entire optical network. Core networks have deployed DWDM to quench this bandwidth explosion. The application of DWDM in metro is not deployed in exactly the same way as in core networks since with metro distances impairments tend to be much less severe. DWDM-based architecture cost-effectively fulfils the bandwidth boost requirements with the flexibility to continuously scale according to evolving market needs. Thus the question for service providers is not if they employ DWDM metro networks but how and when.
A typical metro network scenario portrays an existing infrastructure often relying on multiple 10 Gb/s or 40 Gb/s services multiplexed onto a single fiber pair. Complementing these existing services with higher data rate services requires careful network planning. The optimal network solution is constructed using the building blocks presented in this paper and the architecture depends on the particular geographic location, traffic and expected network growth.
For example consider a legacy system with fully functioning DWDM 10G services in a 100 GHz frequency grid. The 10G services from the line card are aggregated with the aid of a 100GHz DWDM multiplexer as shown in Fig. 7 (upper part).
To maximise the bandwidth utilisation and prevent higher cost, one may opt to add one or more 100G services into the same fiber via the same DWDM multiplexer by deploying the new service via the unused wavelengths. However, due to the distance limitation of currently available 100G transceivers, to enhance the transmission reach to several 100km, the optimal network architecture would be as depicted in Fig. 7 (lower part).
Since the 100G services are more susceptible to dispersion, they would require extra dispersion compensation and optical power boost. Thus an extra 100GHz DWDM multiplexer is first used to combine all the 100G services together followed by a combined dispersion compensation and amplification stage. The grouped 10G and 100G services can then be bundled together with the help of a 50/100 GHz interleaver. Such an architecture makes bridging distances greater than 100 km possible given that the existing 10G channels already supported that distance. Nonetheless, the exact distance depends on the amplifier gain, amount of dispersion compensated and the transceiver performance. To further guarantee a decent transmission distance for this system, FEC can be enabled in the 100G switches (transponders) or transceivers.
Additionally, this architecture conveniently supports the ‘pay-as-you-grow’ model for service providers. If and when the bandwidth is exhausted, the existing legacy 10G channels may be seamlessly interchanged with 100G services. The same remaining components can even be reused to extend the data rate up to 2.4 Tb/s.
This scenario would require 24 differently coloured 100 Gb/s DWDM CFP transceivers deployed together with the already existing 48 channel 100 GHz DWDM and interleaver as shown Fig. 8. All the 100G services are first multiplexed together such that only one dispersion compensation and amplification stage suffices. Clearly, such a network architecture provides higher density with capability to salvage existing infrastructure with flexibility while remaining cost friendly.
The rapidly increasing traffic demand will soon make the operators face the age old problem: lease more dark fibers or find a way to increase capacity of the existing fibers. With the growing developments in the 100G transceiver technology and the rise in interest for installing 100G products, the metamorphosis of current network architecture to 100G systems is inevitable and only a matter of time. This paper, thus, delves into the 100G technologies and network solutions supporting seamless evolution to 100G while easily coexisting with previous 10G services. However, as outlined, care must be taken to limit disruption of the existing 10G services while maximizing reuse of existing equipment.