ROY RUBENSTEIN
drive DWDM networks
two carriers, each using PM-16QAM. By doubling the symbol rate to 60 Gbaud, one PM-16QAM carrier delivers 400 gigabit. At ECOC 2015, last September, Oclaro, Teraxion and NeoPhotonics detailed optical components operating at 60Gbaud. But more work is needed, including developing a coherent DSP at 60Gbaud, before commercial systems become available. Equipment makers are already advancing the signalling rate beyond 30Gbaud to boost the data carried on a carrier. Cisco’s NCS 1002 is the first system that supports 250 gigabit on a wavelength using PM-16QAM and 40 Gbaud. This allows four CFP2-ACO modules to deliver a terabit of capacity instead of five, or 96 wavelengths instead of 120 overall to deliver 24 terabits over a fibre’s C-band. Telco BT says such “flexrate” transponders, configured to use different modulation formats and baud rates, is an important development. “The fundamental driver is about reducing cost but also giving us more flexibility in the network,” says Kevin Smith, research leader for BT’s transport networks. BT has been working with several vendors to increase the traffic carried over a fibre’s C-band. In a recent trial, BT and Huawei transmitted a 3 terabit super-channel made up of 15 carriers, each carrying 200 gigabit using PM-16QAM. Sent over a 360km link and using a flexible grid, each carrier occupied a 33.5GHz channel, increasing capacity by 1.5x compared to a 50GHz fixed-grid channel. “For 16QAM, it is pretty close to the limit,” says Smith.
multiplexed, quadrature phase-shift keying (PM- QPSK) modulation. Going to 16-quadrature amplitude modulation (PM-16QAM) doubles the bits on the wavelength but the received data has less tolerance to noise. The result is a 6-decibel loss in optical link budget compared to PM- QPSK, equating to roughly a quarter of the transmission distance. If PM-QPSK can span 2000-plus km, for example, for PM-16QAM it drops to 500km on the same link. The performance gap between QPSK and 16QAM is significant, which is why vendors have added PM- 8QAM as an intermediate scheme. PM-8QAM delivers 1.5x the capacity of PM- QPSK while doubling the reach of PM-16QAM. Vendors are not stopping there. For data-centre interconnect applications where large capacities may be required over distances up to 100km, 32QAM and 64QAM can be used, boosting the bits per wavelength by 2.5x and 3x compared to PM- QPSK, respectively. Clearly such schemes are unsuited for regional and long-haul links. Vendors are also upping the signalling rate. Doubling the symbol rate – “baud rate” – doubles the data on a transponder’s carrier. “Improving the baud rate has always proven cheaper than going into multiple carriers,” says Elbers. Current optical transport systems use a 30Gbaud symbol rate such that 400 gigabit of capacity can be sent using four carriers, each using PM-QPSK, or
With stacked datacentre platforms users can scale line-side capacity with traffic growth
This suggests the data centre interconnect vendors with their platforms are already approaching that limit. Adva Optical Networkings’s CloudConnect, Coriant’s Groove G30, Cisco’s NCS 1002 and Ciena’s Waveserver with flexgrid technology, deliver 24 to 25.6 terabit over the C-band. What next? There are two obvious directions going forward. One is to keep advancing the data centre interconnect platform’s density, adding more capacity per rack unit. The second approach is to use other fibre bands besides the C-band. Adva’s CloudConnect already supports the C and L bands, doubling line-side capacity to 51.2 terabit. Meanwhile, Nokia Bell Labs is investigating spatial-division multiplexing that promises to increased capacity by exploiting parallel fibre paths. The simplest way to create such parallel transmission
paths is to bundle several single-mode fibres in a cable. But speciality fibre can also be used, either multi-core or multi-mode. Nokia Bell Labs reported an industry first by using a real-time receiver to recover a dozen 2.5-gigabit signals sent over a coupled three-core fibre. Until now the signal processing for such spatial-division multiplexed transmissions has been done offline due to the computational complexity involved. BT is not researching spatial- division multiplexing. ”I’m very much more interested in how we use the fibre we have already got,” says Andrew Lord, BT’s head of optical research. Will web-scale players, with their rapid traffic growth and immense capacity needs, be the first to embrace spatial- division multiplexing in the coming years? If so, they will undisputedly move to the head of the DWDM table.
www.opticalconnectionsnews.com 27
Made with FlippingBook - professional solution for displaying marketing and sales documents online