A number of different industry surveys indicate that total internet demand is growing at about 40% per year. This growth is driven mainly by increasing video traffic in the network—Netflix now takes up to 30% of the internet’s bandwidth at peak hours, and new competitors like Amazon, Hulu, Youku, and the BBC iPlayer are growing rapidly. This growth is now further accelerated by mobile access, with video clients shipping on an all smart phones and tablets, enabling video to be consumed more conveniently via network connections anywhere, anytime.
Furthermore, the video and rich media files are migrating into the network, stored in cloud architecture such as Microsoft Azure or Apple’s iCloud, which tend to produce significant replication of data for resilience and performance purposes, and in which every user experience, whether consumer or enterprise, accesses the network. This provides an opportunity and a threat for service providers. Those that can provide the best user experience in this dynamic and mediarich mobile environment can capture market share. But they must be able to scale their networks dramatically, and do so while lowering capital and operational costs per gigabit per second. The place to start is with the Transport Network, which forms the foundation of long distance internet communication. It is clear that in addition to a move toward larger, more powerful transport switches, the mechanisms of DWDM optical transmission have to change too. A new approach to DWDM capacity—the coherent super-channel—promises an effective solution to the challenges posed by internet growth.
Figure 1: Video, Mobility and Cloud are helping to drive around 40% annual growth in internet demand
WHAT IS A SUPER-CHANNEL?
Many in the industry view the “state of the art” in DWDM in 2012 and 2013 to be 100 Gb/s. However, the growth in the internet is demanding additional bandwidth scale without increasing operational complexity, and asking the question, “What is beyond 100 Gb/s?” One answer is a super-channel, an evolution in DWDM in which several optical carriers are combined to create a composite line side signal of the desired capacity, and which is provisioned in one operational cycle.
WHAT PROBLEMS WILL SUPER-CHANNELS ADDRESS?
Super-channels address three fundamental issues:
• Scaling bandwidth without scaling operational procedures
• Optimizing DWDM capacity and reach
• Supporting the next generation of high speed services
Moreover, it must be possible to implement super-channel technology using existing DWDM engineering techniques, without requiring longer term technology advances.
The most recent innovation that enables a practical increase in bandwidth with long haul reach is coherent detection —allowing 40G and 100G carriers to operate over the same or even greater reach compared to 10G Intensity Modulation Direct Detection (IMDD, also known as Intensity Non-Return to Zero (NRZ) or On/Off Keying. Super-channels will help move beyond 100 Gb/s but must do so while maintaining long haul reach, and the multi-carrier approach appears to be the obvious choice. A flexible coherent (FlexCoherentTM) modulation capability will be essential in achieving these goals.
The first generation “split spectrum” super-channels (See: “Implementing Super-Channels”) will offer around 8 Tb/s in the C-band—between five and 10 times the spectral efficiency offered by 10G IMDD systems that use a traditional 50 GHz grid. When service providers are ready to move outside the current ITU DWDM fixed grid spacing (defined in ITU G.694.1), the next generation of “gridless” super-channels will offer even higher spectral efficiency, enabling a gain of up to 25% in net fiber capacity through more efficient spectrum use.
At this stage it’s not clear what that data rate will be (the two main candidates being 400 Gb/s and 1 Tb/s), so super-channels must be flexible enough to deal with this challenge when it arrives.
Figure 2 shows a proposed architecture of 400-Gb/s Ethernet transceiver. It supports reaches up to 10 km, is referred to as 400GE-LR16, and will be specified by the IEEE. It uses the same LAN WDM optical grid as 100GE-LR4, but spans 70nm. The components extend 100-Gb/s photonic integration technology to sixteen channels. The SOA is used to close the 10km link budget due to higher SMF propagation losses and penalties, and higher optical Mux and DeMux losses.
Figure. 2. 400Gb/s 10km WDM Transceiver.
There are two obvious implementation options for developing single-carrier transponders that operate at data rates above 100 Gb/s. One is to transmit more modulation symbols per second and the other is to encode more bits into a modulation symbol (or some combination of the two). Super-channel technology adds a third option—the ability to treat multiple carriers as a single operational unit.
We assume that a 1Tb/s unit of capacity is required.
A. Transmit More Symbols per Second
Figure 3: A single, 1Tb/s carrier is impractical because the electronics will not exist for another decade. In addition, the single carrier solution would experience significantly greater fiber impairments compared to a super-channel solution.
With single carrier there are two drawbacks:
1. The electronics that drive the interface would also need to operate at 320 GBaud—and this level of integrated electronics performance will probably not be available for another decade.
2. The high symbol rate will encounter significantly higher implementation penalty with today’s electro-optic technology for the same modulation type.
Implementing a 1 Tb/s super-channel using 10 carriers (shown to the right of Figure 3) divides both the required electronics performance and the symbol rate on the fiber by a factor of 10, and so 32 GBaud electronics is all that is required. Moreover, because the symbol rate of each carrier is (in this example) the same as a 100G PM-QPSK transponder, the optical performance is more than adequate for most terrestrial long haul and ultra-long haul links.
A key point is that both single carrier and super-channel implementations have about the same spectral efficiency, but the super-channel has far better optical performance and it is possible to build it using technology that will be available in the near future.
B. Encode More Bits per Symbol
PM-QPSK encodes four bits per symbol—which is four times more than conventional IMDD modulation. The combination of this encoding efficiency, coherent detection and high-gain FEC technology allows a 100G signal to have the same or even better reach compared to 10G IMDD.
Higher order modulation will certainly be a useful tool to allow service providers to optimize the total spectral efficiency for certain routes. But the penalty for this is reduced reach.
Figure 4: Adding more bits to a symbol increases spectral efficiency, but the total power per symbol (before non-linear threshold is reached) is indicated by the thick black circle.
A pair of yellow circles in Figure 4 represents an encoded bit—and the more of these that are put into the black circle, the less optical power is available per bit. Table 1 shows this numerically. In simple terms, modulation techniques like 16QAM may be limited to regional network use.
Table 1: Reach vs Total Capacity for a selection of phase modulation types. Note this table is illustrative only—not all modulation types shown are practical, and other modulation types may be available in final products.
THE IMPORTANCE OF PHOTONIC INTEGRATION
Super-channels allow a Terabit of DWDM capacity to be turned up in a single operational cycle, without any penalty in terms of spectral efficiency and with the same optical reach as today’s generation of 100G coherent transponders.
Figure 5: A super-channel built with a PIC. The PIC enables hundreds of optical functions to be collapsed into two small chips the size of a fingernail enabling, in this example, 10 x 100 Gb/s in a single line card.
It is clear that a 10 carrier super-channel requires 10 sets of optical components per line card. Implementing such an interface using discrete optical components would seem totally unrealistic, and Figure 5 shows the scale of the problem. On the left we see 10 individual 100G transponders. These will contain in total around 600 optical functions that are probably implemented in discrete optical components.
On the right of Figure 5 is a Terabit super-channel line card. All of the major optical functions on all 10 100G line cards have been integrated into a single pair of Photonic Integrated Circuits (PICs)—one to transmit and one to receive. All 10 carriers can now be implemented in a compact line card where the super-channel is brought into service in one operational cycle, consuming far less power than 10 discrete transponders and resulting in far greater service reliability.
FLEXIBILITY IS THE KEY TO SUPER-CHANNEL SUCCESS
In the real world super-channels will have to be extremely flexible in a number of properties:
• What type of modulation should be used?
• What is the best way to optimize spectral efficiency and reach?
• What spacing will be used between the carriers?
• What is the total width of the super-channel?
Figure 6: The ideal super-channel line card would offer flexible modulation to allow the service provider to trade reach for capacity without making spares management more complex.
SUPER-CHANNEL IMPLEMENTATION PROGRESS
In 2011 Infinera announced a series of super-channel trials, each one designed to showcase a specific aspect of PIC-based super-channel technology.
NOVEMBER 2011: THE TELIASONERA TERABIT SUPER-CHANNEL TRIAL
In November 2011 TeliaSonera International Carrier successfully completed the world’s first Terabit optical transmission based on two 500 Gb/s super-channels. The trial was over a 1,105 kilometer route from Los Angeles to San Jose, California. This was a production fiber, and in addition to the Terabit of super-channel capacity, there was 300 Gb/s of 10G IMDD traffic on the route. This shows that a split-spectrum super-channel can operate very effectively with grid-based IMDD traffic.
MIGRATING TO SUPER-CHANNELS: WHEN TO GO GRIDLESS?
The ITU frequency grid defined in G.694.1 has been used in most of the world’s DWDM networks for many years. Multi-carrier super-channels do not require such rigid 50 GHz guard bands between carriers, and this fiber spectrum could be reclaimed by going “gridless”.
Figure 8: Split-Spectrum vs Contiguous Super-Channels. By removing the inter-channel “guard bands” a contiguous spectrum super-channel could occupy about 25% less fiber spectrum. A split-spectrum super-channel would be more compatible with existing grid-based transmission systems.
This is shown in Figure 8, in which about 25% of the fiber spectrum could potentially be reclaimed by a contiguous carrier super-channel that ignores the current ITU grid. The ITU is now working on a new “flex-grid” spacing (based on multiples of 12.5GHz) that would enable support for contiguous, or “gridless”, multicarrier super-channels.
A split-spectrum super-channel trial was recently completed by TeliaSonera International Carrier (see sidebar), in which interoperability with existing 10G IMDD, grid-based carriers was demonstrated.
White Paper ---- Super-Channels: DWDM Transmission at 100Gb/s and Beyond, Infinera.