RZ modulation can boost link budgets

March/April 2009

 

Historically, the optical networking industry has used non-return-to-zero (NRZ) modulation to encode optical signals, but it is now starting to favour return-to-zero (RZ) as the modulation format, because the shorter pulse length mitigates the effect of optical dispersion.

In NRZ modulation, the optical signal is turned on for the duration of a “1” and then turned off for the duration of a “0” – hence it is referred to as “on-off keying” (OOK). “OOK has existed for such a long time because it is so simple,” said Loi Nguyen, vice-president of technology at Inphi.

Unfortunately, OOK signals are not very tolerant of chromatic dispersion – an effect that causes pulses to spread out as they travel down the fibre. At 10 Gbit/s transmission speeds, the bit error rate in the signal becomes unacceptable after about 80 km. Dispersion effects get worse over longer distances and at increased transmission speeds. In fact, 40 Gbit/s transmission using OOK coding would be restricted to a distance of 2 km, which would render it unsuitable for metro and long-haul transport applications.

The solution, which has already been adopted in the ultra-long-haul and submarine cable markets, is RZ modulation. Unlike NRZ modulation, where the optical signal stays at “1” for the full duration of the bit, an RZ signal is pulled back down to “0” much sooner, although the bit period – the time between one bit and the next – stays the same. Typically, the duration of the pulse representing the “1” is 30–40% of the bit period. “As the bit travels down the fibre, you have room for the bit to spread out without colliding with its neighbours,” Nguyen explained.

 

Over the past few years, RZ coding has become the format of choice for manufacturers of 40 Gbit/s metro and longhaul modules and transmission systems. However, the industry has been missing a small but critical component: an electronic NRZ-RZ converter. Instead vendors apply an optical solution using two modulators – one to create the signal and a second to shorten the pulse. “It’s an expensive solution, but it’s all that’s available today,” Nguyen commented.

The conversion function itself is quite simple to design in electronics using a logical AND gate (figure 1), according to Nguyen. The difficulty lies in designing circuits that operate fast enough for the next generation of transmission speeds. An RZ signal switches on and off roughly twice as fast as the equivalent NRZ signal, and so occupies roughly twice the bandwidth. Therefore, electronic components such as the RZ converter must operate at twice the bandwidth compared with an NRZ signal at the same data rate.

For example, an NRZ-RZ converter for 40 Gbit/s long-haul transmission using differential quadrature phase-shift keying (DQPSK) modulation would require a bandwidth of around 40 GHz, while one for 100 Gbit/s dual-polarization DQPSK – a modulation scheme proposed for future 100 Gbit/s long-haul transmission – would need a bandwidth of 50 GHz.

High-speed logic gates are available from a number of suppliers. Inphi offers a range of OR gates that can be used as NRZ-RZ converters with bandwidths of 13, 25 and 50 GHz.

The electronic approach to NRZ-RZ conversion can have a positive impact on the cost and complexity of high-speed components and subsystems, says Nguyen. What’s more, as awareness of the advantages of RZ modulation grows, the technology could become more widely used at lower speeds, such as 10 Gbit/s. The increased system margin provided by RZ signalling gives the network operator more options: the transmission system can be designed to achieve longer distances without regeneration, or cheaper optics can be used inside the transceivers. This will open up a range of new applications for NRZ-RZ converters, he predicts.

Return to top