Private optical networks have been used by large enterprises in metro area applications for several years. Now private wide-area optical networks (WANs) are considered by large commercial enterprises. The proposed contribution describes the field trial of a private optical network for a large financial institution using a single-fiber, shared-amplifier DWDM system offering low first cost, highly automated operation. Initial cost is a crucial decision factor for an enterprise in whether to deploy an extended-length private network with dedicated WDM transmission equipment or to satisfy their WAN requirements with the traditional model of leased capacity from carriers. We demonstrate the field deployment of a 14-span, 1250-km ULH DWDM transmission system with an optical architecture that significantly reduces the initial equipment costs by requiring only a single transmission fiber and a single optical amplifier and dispersion compensator per hut, as in Fig. 1(a). Figure 1(b) shows the structure of the co-directional amplifier for bi-directional traffic. Band splitters are used to direct the incoming traffic in separate sub-bands from the fiber through a uni-directional amplifier, and from the amplifier output back to the fiber spans. Both bands of traffic co-propagate through the shared amplifier and the dispersion compensator (DCM), and at some sites through a per-channel dynamic gain equalizer.In the contribution we will discuss the specialized design considerations for the dispersion map and the power equalization that are required for optimum performance in both traffic directions for a system architecture with shared (Fig. 1(a), bottom), rather than traditional separate (Fig. 1(a), top), amplification and dispersion compensation. This introduces particular issues for a field-deployed installation, where fiber insertion loss and dispersion can vary significantly even between neighboring spans.The field system connected terminal sites in Columbus, Ohio and Wilmington, Delaware, using thirteen intermediate amplifier sites. The majority of the fiber spans were Corning E-LEAF fiber, with two spans of OFS TrueWave-Classic (TWC), and spans of mixed E-LEAF/SSMF or TWC/SSMF at the system terminal locations. In the field trial configuration, four of the ten available full-duplex channels of the DWDM system were equipped, with one client-side interface each of OC-192, 4 x OC-48, native 10-GbE LAN PHY, and 4 x 2G-FibreChannel, for an active capacity of 38 Gbps out of the 100-Gbps maximum available capacity. After physical installation of all field equipment, the optical transport system was evaluated by monitoring raw (pre-FEC) error rates over a period of 7 weeks. In the contribution, we will report on initial troubleshooting of the dispersion map and show the OSNR and raw Q-factor derived from pre-FEC BER of all eight wavelengths (for four full-duplex channels) over the initial testing period. Figure 2(a) shows, as an example, the measured OSNR and Q-fcators of all 8 active wavelengths after system turn-up and Figure 2(b) shows the evolution of the system Q-factor over a 3-week period. The minimum observed Q-factor was 13.9 dB, which leaves a system margin of more than 6 dB over the minimum Q-factor of 7.8 dB (BER 7e-3) that is required by the FEC for a corrected error rate of 10-15.
© 2006 Optical Society of America