The emerging role of electronics in optical networks

By Saeid Aramideh

Overview

As carriers seek to evolve their networks to 40 and 100 Gbps, new cost-effective ways to deal with the resulting signal impairment issues and the longer transmission requirements must be found.

The ever-increasing need for more bandwidth—from the user to the long-haul network— is creating the requirement to upgrade data rates from 10 Gbps to 40 and even 100 Gbps. Carriers are seeking innovative ways to perform these upgrades in ways that will not require major overhauls and will allow them to reuse their existing optical fiber.

However, before taking an existing DWDM optical network from 10 to 40 Gbps and beyond, there are signal impairment issues that must be addressed—primarily chromatic dispersion (CD) and polarization-mode dispersion (PMD). This will require deployment of intelligent electronics in the core of the network to bring existing optical infrastructure in line with this heavy bandwidth demand.

To 40 Gbps and beyond

The proposed optical transmission systems that promise 40 Gbps per wavelength fall into two broad categories. The first uses amplitude modulation, typically using binary or on-off keying. These formats basically include non-return-to zero (NRZ) and return-to-zero (RZ) to support one bit per interval, making the interval rate equal to the bit rate.

The second category carries information in the optical phase using phase-shift keying, such as differential binary phase-shift keying (DPSK) or differential quadrature phase-shift keying (DQPSK). Both of these schemes are considered differential schemes since they convey information by viewing the change in the phase of the signal.

A superior modulation scheme that has shown excellent performance for upgrades from 10 Gbps to 40 Gbps for existing overlay networks combines QPSK with coherent detection. This coherent polarization-multiplexed QPSK (CP-QPSK) modulation format seamlessly works with intermixed 10- and 40-Gbps links.

Three basic technologies make up 40-Gbps CP-QPSK. First, polarization multiplexing, or dual multiplexing, enables the transmission of a signal by decomposing it into two signals of the same frequency that are polarized by 90º. Each polarization allows the signal to have one bit per symbol. Combining them together effectively allows for the modulation of two bits per symbol.

Second, QPSK modulation supports two bits per interval or symbol because the symbol can exist in one of four different phases. By combining these first two technologies, four bits per symbol can be achieved, which lowers the symbol rate to that of 10-Gbps transmission and eases the bandwidth requirements on optoelectronic components.

The third technology is the use of coherent detection with digital signal processing (DSP) at the receiver. DSP recombines the data in the four optical analog signals by applying analog-to-digital conversion (ADC), clock recovery, equalization, carrier phase estimation (CPE), and recovery. Figures 1 and 2 provide the physical perspective on QPSK modulation with polarization multiplexing and coherent detection, respectively.

In Fig. 1, the transmitter side employs an advanced integrated tunable laser assembly (ITLA) that supports the full C-band with 50-GHz-spaced wavelengths. The ITLA output is fed to the polarization beamsplitter (PBS) and split into two orthogonal polarization signals. These are each fed into the input of the QPSK modulator where two Mach-Zehnder interferometer (MZI) structures generate four polarization-diverse, in-phase, quadrature data streams—each supporting a 10-Gbps data rate. The outputs are recombined with a PBS to form one polarization-multiplexed signal. For the ease of implementation, all these parts are integrated into one Mach-Zehnder modulator.

At the receiver side (see Fig. 2), the four polarization-multiplexed 10-Gbps signals propagate through a PBS into the inputs of a polarization-diversified 90º optical hybrid. The optical hybrid converts and feeds complete amplitude, phase, and polarization information of the optical signal to the photodetectors. Finally, the electrical outputs of the four detectors are sent to a multipurpose DSP that performs analog-to-digital conversion, carrier recovery, equalization, differential decoding, and G.709 frame detection for recovery of 4×10-Gbps signals before the final SFI-5 electrical interface.

Comparing all the options

Several other options exist for 40-Gbps phase modulation. The two that have achieved some level of success on the market are DPSK and return-to-zero DQPSK (RZ-DQPSK).

As mentioned earlier, DPSK only supports two possible phases or one bit per symbol. The two different phases of the DPSK signal can only be at 0° and 180° or, in binary terms, encoded as zero or one. Therefore, the symbol rate of DPSK is equal to its bit rate of 40 Gbps.

The capacity of RZ-DQPSK for four possible phases enables it to be encoded by two bits per symbol. Thus, for 40-Gbps applications, its symbol rate is still only half the bit rate, or 20 Gbps. This precludes the ability to mix 10- and 40-Gbps transmissions for migrating an existing 10-Gbps network to 40 Gbps without link engineering modifications.

As shown earlier, CP-QPSK combines several technologies, including polarization multiplexing, to yield a symbol rate of 10 Gbps. This enables carriers to reuse their existing 10-Gbps optical infrastructure for a phased migration to 40 Gbps. Additionally, the lower symbol rate and the additional capabilities of coherent detection result in a narrower optical spectrum, improved optical signal-to-noise (OSNR) sensitivity, and a much higher tolerance to CD and PMD.

Table 1 shows the optical performance comparison of the different 40-Gbps phase-modulated technologies. Direct detection technologies, such as NRZ, optical duobinary (ODB), RZ-DQPSK, and DPSK, all fit into a certain performance level—from 6 to 25 ps of differential group delay (DGD) tolerance and 70 to 150 ps per nanometer of CD tolerance. However, coherent detection boasts much higher tolerance to both signal impairments (>100 ps for DGD and >35,000 ps/nm for CD). By using ADC and integrated equalizers, coherent detection leverages the latest electronic signal-processing technologies to enable much better performance.

This data is particularly significant to carriers deploying reconfigurable optical add/drop multiplexing (ROADM)-based networks. Coherent detection modulation is the only option for these configurations and no suitable DPSK implementations exist for supporting multiple cascaded ROADMs—a requirement for next-generation optical networks. In ROADM-based links, tolerance to cascaded optical filters is important to achieve low bit-error-rate (BER) performance. The signal transmission through a cascade of ROADMs results in passband narrowing of the optical signal. But due to the lower symbol rate and other capabilities of coherent detection, its narrower optical spectrum overcomes this issue.

Another key attribute of CP-QPSK is its low latency. Latency in data networks is an important issue, especially in critical data transmission links for financial institutions. The physical medium of the network, mainly the fiber, is the greatest contributor to optical latency and includes dispersion/slope compensation modules (DCMs) and ROADMs.

For 40-Gbps CP-QPSK with its inherent advanced modulation scheme and coherent detection, CD tolerance is better than most 10-Gbps technologies. The latency of the DSP engines is also negligible when compared to that of the fiber plant. From a pure network-level perspective, 40-Gbps CP-QPSK offers the lowest latency compared to other technologies. For example, compared to 40-Gbps DPSK and RZ-DQPSK, the coherent approach requires no tunable dispersion compensation on the line card and delivers better CD tolerance to enable a network designer to eliminate DCMs—both decreasing latency and lowering capital expenses.

Cost: Optical versus electronics

Coherent detection also offers significant capital expense (capex) savings. A network cost model compared the bill-of-materials cost of DPSK, RZ-DQPSK, and CP-QPSK. This generic transmission link model comprised terminal/regenerator hub nodes, bidirectional line amplifier nodes, and branch hub nodes (for optical add/drop multiplexing, or OADM).

Three link scenarios were studied to compare forward error correction (FEC)-enabled 43-Gbps transponder performance for the three transponder schemes. The first was based on using low-PMD fiber with a mean DGD between 0 and 3 ps. The second scenario used high-PMD fiber with a mean DGD between 7 and 30 ps (optical PMD devices were required for both DPSK and RZ-DQPSK cases). The third scenario focused on the same parameters as the second, with 50% of the hub nodes being OADM nodes. All scenarios assumed that linear, point-to-point links were used with in-line DCM compensation and 88 C-band channel propagation.

The network reference model was based on a typical Tier-1 U.S. carrier network consisting of 30 hubs and 40 point-to-point links with a total capacity of 150 Tbps (see Fig. 3). This reference model was devised by J.M. Simmons for his study, “On Determining the Optimal Optical Reach for a Long-Haul Network.” All traffic was consolidated to the 43-Gbps interfaces and the costs of traffic aggregation equipment and protection systems were considered negligible.

The results for the first scenario using the low PMD fiber with a mean DGD between 0 and 3 ps showed a 16% savings when using CP-QPSK over DPSK for all system reach/average hub-to-hub (SR/) ratios. CP-QPSK saved 36% in network costs across all ratios when compared to RZ-DQPSK.

In the second scenario based on using high PMD fiber with a mean DGD between 7 and 30 ps, the results were even more significant. The CP-QPSK network saved 39% on network costs versus the same DPSK network. The comparison between CP-QPSK and RZ-DQPSK showed an advantage of 50% savings with the coherent detection scheme.

In the last scenario based on OADM nodes over high-PMD fiber with a mean DGD between 7 and 30 ps, the cost analysis again favors CP-QPSK. In this scenario, all OADM nodes had 50% add/drop capacity and 50% of all nodes from hub-to-hub had OADM capability. It should be noted that OADM channels do not require regeneration. Also, the OADM filter costs were assumed to be half that of the terminal filters. The results yielded a 40% savings for CP-QPSK over DPSK and 51% savings over RZ-DQPSK for all ratios.

Table 2 shows the absolute cost savings for all three scenarios for the 30-hub, 40-link DPSK network model. In summary, capex savings using CP-QPSK range from 16% to 51% over direct-detection technologies.

Coherent detection delivers

All of the components in the core of today’s networks—ROADMs, switches, filters, amplifiers, etc.—were provisioned for 10-Gbps networks. The demand for higher bandwidth rates is driving a new requirement of at least 40 Gbps, and many carriers are already considering the right migration path for their networks. Many important considerations await these carriers, but the ease of migration and capex costs are high on the list.

Coherent detection and polarization-multiplexed DQPSK technology is superior to direct-detection technologies in cost savings and optical performance for migration to 40 Gbps. To get to 100 Gbps, coherent technology is the only choice. Using coherent electronic technologies will reduce complexity, cost, and support for 40-Gbps data rates on existing 10-Gbps networks—resulting in a viable migration path for 100 Gbps. With its flexibility and universal attributes enabled by enhanced DSP, coherent detection provides carriers with the building blocks for today’s and tomorrow’s bandwidth demands.

Links to more information

LIGHTWAVE:Is DP-QPSK the Endgame for 100 Gbits/sec?
LIGHTWAVE:Industry Plots PMD Compensation Paths at 40G, 100G
LIGHTWAVE: Vendors Align Technologies to 40G Strategies

Saeid Aramideh is senior vice president of global sales, marketing, and business development at CoreOptics (www.coreoptics.com).