Technical papers

Here is a small sample of the technical paper highlights being presented at the conference.

Doubling core network capacity 

Microsoft has conducted a study measuring the performance of its North American core backbone network to determine how the use of bandwidth-variable transceivers (BVTs) could boost capacity.

The highest capacity modulation scheme suited for each link from the choice of polarisation-multiplexed, quadrature phase-shift keying (PM-QPSK), polarisation-multiplexed, 8 quadrature amplitude modulation (PM-8QAM) and PM-16QAM can then be used.

By measuring the signal (Q-factor) for all its PM-QPSK based 100 gigabit links, Microsoft's study found that network capacity could be increased by 70 percent using BVTs. Equally, having the ability to increase capacity in 25-gigabit increments would increase capacity by a further 16 percent while a finer resolution of 1-gigabit would add an extra 13 percent.

Microsoft says such tuning of links need not be done in real time but rather when a link is commissioned or undergoing maintenance.

[paper M2J.2]

Architecting a new metro

How can operators redesign their metro network to enable rapid service innovation? This is the subject of a joint paper from AT&T, the Open Networking Lab and Stanford University. The work is part of a programme dubbed CORD to redesign the central office as a data centre using commodity hardware and open software to enable the rapid scaling of services. In particular, OpenFlow-enabled white boxes, the Open Network Operating System (ONOS) - a software-defined networking (SDN) operating system, and network control and management applications are used.

As part of CORD, three legacy telecom devices - optical line termination (OLT), customer premises equipment (CPE), and broadband network gateways (BNG) - have been virtualised and implemented on servers.

The paper details how a single SDN control plane based on ONOS is used to create a converged packet-optical metro network and how its support for bandwidth on-demand and automatic restoration at the optical level is used for enterprise connectivity and video distribution services.

The paper also discusses how the metro architecture supports 'disaggregated' reconfigurable optical add/ drop multiplexers (ROADMs). By disaggregating a chassis-based ROADM into commodity components, an operator can size its infrastructure as required and grow it with demand, the paper says.

[paper Th1A.7]

400 gigabit single-carrier transmission

Nokia Bell Labs reports work on 400 gigabit-per-second (Gbps) single-carrier optical transmission over submarine distances. The attraction of adopting 400 gigabit single-carrier transmission as that it is the most efficient way to reduce the cost-per-bit of optical transmission systems.

The Bell Labs' paper reviews state-of-the-art 400 gigabit single-channel transmissions over 6,000km and greater distances, and discusses the tradeoffs between such variables as symbol rate, modulation and forward error correction (FEC) schemes.

400Gbps single-carrier submarine transmission is likely in the next few years

PM-16QAM is proposed as a promising modulation scheme to achieve beyond 6,000km distances and a spectral efficiency exceeding 5b/s/Hz. But this requires a symbol rate of greater than 60 gigabaud to accommodate the 20 percent overhead FEC. Pulse-shaping at the transmitter is also used.

Exploring the receiver performance with the varying symbol rate/ FEC overhead, Bell Labs reports that the best tradeoff between coding gain and implementation penalties is 64 gigabaud and 27.3% overhead. It concludes that single-carrier 400Gbps submarine transmission is likely in the next few years.

[paper Th1B.4]

Silicon modulator for CFP2-ACOs

Cisco has developed a compact flip-chip assembly that combines a silicon photonics modulator and a silicon germanium BiCMOS Mach-Zehnder modulator driver. Such an assembly forms the basis for low-cost advanced coherent optical transceivers such as the CFP2-ACO.

Cisco has demonstrated the assembly operating at 128.7Gbps using PM-QPSK and 257.3Gbps using PM-16QAM. Cisco believes this is the first demonstration of transmission at 257.3Gbps using PM-16QAM over 1,200km of standard single-mode fibre using a silicon photonics-based device.

The modulator has also been demonstrated operating at 321.4Gbps using PM-16QAM transmission and a 20 percent FEC overhead, the highest bit rate yet achieved using a silicon-photonics based transmitter, claims Cisco.

Cisco is already using CFP2-ACO modules as part of its NCS 1002 data centre interconnect platform that implement PM-16QAM and deliver 250 gigabit due to the use of a higher baud rate than the 32 gigabaud used for existing 100-gigabit coherent systems.

[paper Th1F.2]

Flexible Ethernet to exploit line-side efficiencies

Given how the optical network network is starting to use adaptive-rate interfaces, a paper from Google asks how the client side can benefit from such line-side flexibility.

The paper points out that traditional DWDM transport equipment typically multiplexes lower-rate client ports but that this doesn't apply to network operators that manage their own data centres. Here, traffic is exclusively packet-based from IP routers and typically matches the line rate. This is why data centre interconnect platforms have become popular as they require limited grooming capability.

Google highlights how Flexible Ethernet (FlexE), for which the Optical Internetworking Forum has just defined an Implementation Agreement for, combined with data centre interconnect equipment is an extremely effective combination.

FlexE supports Ethernet MAC rates independent of the Ethernet physical layer rate being used. Google shows examples of how using FlexE, sub client rates can match the line-side rate as well as how multiple client ports can support a higher speed router logical port.

The paper concludes that combining FlexE with data centre interconnect results in a low cost, low power, compact design that will enable Internet content providers to scale their networks.

[paper W4G.4]