JOHN WILLIAMSON TRANSCEIVERS
HOWTRANSCEIVER DEVELOPERS ARE RISING TO THE CHALLENGE
As network data rates push up from 100G to 400G (and who knows where next!), transceiver developers are rapidly responding to network operators’ demands. By John Williamson.
optical impairments and further improve performance. In similar vein, Carter reckons that enhanced DSP will be necessary for the line side of 400G, and some signal processing will be needed to support the desired laser bandwidths at the higher speeds. “Really what the industry will require is a 7 nm type CMOS DSP to run the power consumption that is needed by very small form factors,” he suggests. “In time the physics of lasers is also going to dictate that you need some electronic signal processing to enable the required bandwidths to be achieved.” 100G/400G toolbox. Lipscomb notes that, with the proper DSP, the same optical elements can support 100G for up to 2,000 km using QPSK modulation, and 400G for metro distances of 500 to 800 km, and even 600G for datacentre dynamically switched between these dierent applications after deployment in a software defined network”, he says. “In a meshed long-haul or metro optical network, the fact that 100G coherent is essentially independent of optical fibre dispersion and distances makes re-routing wavelengths conveniently done via software control without worrying about the physical layer limitations.” AI is also making its mark, enabling transmission systems to optimise the capacity per wavelength for a given path and amount of desired margin at a specific point in time. “This capability is essential for handling the increasingly unpredictable nature of today’s trac requirements” concludes Roberts. “Additionally, machine learning algorithms are useful in deriving design equations, in optimising configurations.” interconnect distances of 80 km. “The same transponder can be SOFTLY DOES IT SDN is another useful tool in the
G lobal 100G and 400G optical transceiver markets are rapidly expanding. The drivers of demand for the high- speed connectivity enabled by these transceiver classes include: the growing video component of telecommunications trac; the rise of mega-datacentres and the interconnection of such centres; the popularity of cloud computing; rapidly increasing consumption of FTTX/FTTP/ FTTH services; the massive expansion of Internet usage and the dawning prospect of the Internet of Things/Everything. According to WinterGreen Research, the overall worldwide optical transceiver market could grow to $41.1 billion by 2022, driven by the availability and cost eectiveness of 100G, and 400G devices. HOT AND BOTHERED Although there are obvious dierences between new requirements on the line side and the client side of very high- speed networks, and between dierent applications run on those networks, heat and component size are collective concerns. “In general, the challenge that we face is maximising the reach of a signal in the face of optical noise and nonlinearity, given the constraints of heat and size that arise from each particular application - from metro Data Centre Interconnect to trans-Pacific submarine,” states Kim Roberts, Vice President WaveLogic Science, Ciena. Related to this, there is an across-the- board need to reduce power consumption and improve thermal performance and management in the 100G/400G landscape. “It’s all about shrinking the size of the pluggable modules and reducing power consumption,” says Adam Carter, Oclaro’s Chief Commercial Ocer. He observes that, for example, reducing module footprints compared to that of the CFP8 form factor will drive lower power consumption, as well as increase density. However, while very small form factor
optical transceivers will play well on the DCI client side, Carter thinks that longer haul at 400G is probably going to need transponder-type solutions. “When you shrink down integrated photonic circuits, you are making decisions about specifications versus the size of the chip,” he judges. President of Marketing Ferris Lipscomb, 100G and 400G inside the datacentre and between closely spaced data centres can be accomplished with direct detect signalling technologies such as NRZ and PAM4. “Longer distances can only be done with coherent technology,” he observes. “Coherent technology corrects digitally for fibre impairments which become limiting at the longer distances, and permits up to 100 separate channels per fibre, increasing spectral density.” Lipscomb contends that the current coherence cut-over point is around 80 km, but as coherent technology comes down in cost with higher volume, the cross over point may push down to 40 km or even 10 km. DSP has a starring role in 100G/400G. In coherent systems, DSP is used to remove impairments from the optical fibre, or from other “non-ideal” components. Roberts believes that the scope of DSP will be expanded to address more COHERENTWHOLE According to NeoPhotonics’ Vice
OSFP AGREEMENT In March 2017 the Octal Small Form Factor Pluggable (OSFP) Multi Source Agreement (MSA) group released the electrical and mechanical specifications for the new pluggable OSFP form factor that is capable of supporting 400G optics technologies for datacentre and metro applications, and is designed to support the next generation of 800G optics modules. The same month the Quad Small Form Factor Pluggable Double Density (QSFP-DD) MSA released a specification for the QSFP-DD form factor to address solutions up to 400 Gbits/s aggregate per port. Is one superior to the other? ”Dierent system houses have dierent views,” states Carter. “I’m not going to get religious about the issue,” adding that Claro can accommodate either.
ISSUE 10 | Q3 2017
Made with FlippingBook - Online magazine maker