Deployment of a new technology across a large network can take years, usually with a strong economic imperative to load traffic onto the new layer as quickly as possible. Routing such traffic through the early-stage network can result in nonoptimized usage of resources in the fully built network, impacting the network’s cost-effectiveness in ways not directly quantified by conventional design studies. Using network simulations of quasi-static traffic, we show that the order in which links are built affects the efficiency of the final network. If the optical layer can be hitlessly reconfigured, significant benefit is achieved by network re-optimization as the technology’s footprint grows over time.

We explore challenges faced when deploying a new technology to serve a large network. Our simulations demonstrate the benefits of having an optical layer that can be reconfigured as the technology’s footprint grows over time.

2 ; 1 Bell Laboratories, Alcatel-Lucent, USA; 2 Micram Microelectronic GmbH, Ger- many. We present a novel 2-channel arbitrary waveform generator (AWG) prototype, generating complex signal constellations with 6-bit digital-to-ana- log conversion (DAC) of up to 72 GS/s at a 6-dB bandwidth of 23 GHz. We generate a 55-GBd root-raised cosine shaped single-carrier 64-QAM signal. 1 ; 1 TE Sub- Com, USA. We experimentally study the performance of coded-modulation formats based on Nyquist-spectrally- shaped mQAM constellations with spectral efficiencies from 2.4 to 8.0 bits/s/Hz, and demonstrate that the relative performance in uncompen- sated links depends only on their respective OSNR sensitivity. TeraXion, Canada. We experimentally investigate the impact of sinusoidal laser phase on parallel and pipelined decision-directed phase recovery in a 5 Gbaud 64-QAM system, including the effects of frequency offset com- pensation and equalization. M2B.4 • 14:15

We have studied the resource requirements and capacity of a continental-scale backbone network supporting shared mesh restoration (SMR) and service velocity (a rapid provisioning method). Network simulations of wavelength routing by colorless, nondirectional reconfigurable optical add/drop multiplexers (CN-ROADMs) show that performance depends on the routing scheme (minimum distance or minimum regenerator) and on whether or not regeneration is limited to a selected subset of sites. Comparing SMR with dedicated protection, we find that up to 27% fewer regenerators are required for the shared mesh case, while capacity is increased by approximately 40%. Regenerator site concentration and minimum-regenerator routing provide the best results.

Simulations of shared mesh restoration in a backbone network supporting rapid provisioning show up to 27% fewer regenerators than dedicated protection, plus ~40% increase in capacity. Regenerator site concentration and minimum-regenerator routing provide best results.

The advent of reconfigurable optical add/drop multiplexers with colorless and non-directional add/drop ports enables transponders and regenerators to be pre-deployed, without a priori knowledge of which wavelength or direction they will eventually serve. We study pre-deployment of optical regenerators as a means to drastically reduce the provisioning time, using Monte Carlo simulation of an optical backbone network. With appropriate placement strategies, regenerators can be efficiently pre-deployed so that new connections can be established rapidly, without the delays caused by service visits to intermediate network nodes.

We study the impact of partitioning client-side fiber cross-connects on the blocking performance of colorless, non-directional ROADM based dynamic optical networks. Simulations show that smaller fiber cross-connects (16×16, 48×48) can achieve low blocking.

We propose a coordinated shared mesh restoration scheme at sub-wavelength connection level using integrated OTN switch and CDC-ROADM nodes. The optical layer flexibility of CDC-ROADMs in the integrated node enables additional OTN line card sharing and simulation results show up to 18% savings in OTN line cards compared to classic ROADMs.

Dynamic photonic networks rely on colorless, non-directional Reconfigurable Optical Add/Drop Multiplexer (ROADM) nodes to enable rapid re-routing of wavelength channels without optoelectronic conversion. We report numerical simulations of the wavelength contention that can occur in such multi-degree ROADM nodes. Intra-node blocking rates and transponder utilization are computed for node designs with and without a client-side fiber cross-connect, and both are compared to results predicted for an ideally contention-free ROADM.

We show that blocking in dynamic networks of colorless, non-directional ROADMs is tolerant to intra-node contention when contention-aware routing/wavelength assignment (RWA) algorithms are used. An optional client-side cross-connect enables low blocking with simpler RWA variants.

Using numerical simulation of a dynamic network, we quantify the intra-node blocking occurring in colorless, non-directional Reconfigurable Optical Add/Drop Multiplexers (ROADMs). Designs including a client-side fiber cross connect are shown to offer lower blocking probability.