In today's competitive environment, service providers must respond quickly and cost-effectively to new revenue opportunities, such as advanced video and IP services.
In order to capitalize on these opportunities, they need a transport solution which provides high capacity, flexibility and availability. However, many networks have reached a saturation point during the process of expanding and interconnecting their existing SONET/SDH networks. They are running out of both capacity and fiber needed to address new expansion. Through both network and business models analyses, service providers are coming to the realization that DWDM (dense wave division multiplexing) transport networks provide the most cost-effective means of addressing both of these requirements.
Figure 1: Sample four-node network of fixed or reconfigurable OADMs.
There are three major DWDM architectures that can be used to provide additional bandwidth: Point-to-Point Mux/Demux DWDM, Fixed-OADM, and Reconfigurable-OADM Architectures (PLC-based Wavelength Selective Switch-ROADM or Scalable-ROADM). The flexibility required for today's networks means that service providers do not consider point-to-point mux/demux architectures, except for single spans that have no churn. Therefore, with flexibility as an underlying requirement, mux/demux architectures will not be considered in this analysis.Fixed-OADM architecture
The fixed OADM DWDM architecture is much more cost-effective than a point-to-point Mux/Demux DWDM architecture. In a fixed OADM DWDM architecture, only the specific channels destined for particular sites are selected and dropped. All of the other channels are optically bypassed across the site. The optical bypass ability drastically reduces the capital investment requirement by eliminating the need to regenerate all of the channels.
While this architecture is cost-effective in terms of initial capital expense, it is operationally very complicated and expensive to manage. Because all of the optical paths are predetermined when the system is initially deployed (using fixed OADMs), traffic changes or churn require network technicians to physically visit the sites and redesign the network (changing filters, retuning the existing optical interfaces and/or adding new ones). This process normally disrupts the other working traffic on the line.Reconfigurable OADM (ROADM) architectures
In legacy DWDM networks without ROADM technology, frequent traffic disruptions and truck rolls result from capacity upgrades and traffic churn as referenced in the previous section. ROADM solutions alleviate these operational expenses (OPEX) through their software-configurable wavelength add-drop functionality. In the highly competitive metro environment, ROADM technology delivers the service provider a competitive edge by enabling swift response to customer requests for new circuit turn up.
Figure 2: Sample wavelengths being dropped at sites B, C and D.
Many believe that there is a trade-off between the flexibility attained from ROADMs and the initial first costs. Therefore, any ROADM deployment must have a good return-on-investment (ROI): low OPEX, low first cost, and low total cost of ownership.
The right time to deploy ROADM is when an optical metro or regional network is expected to transport converged voice, video and data services. Today, all service providers strive to offer "triple play" services for enhanced revenue and customer retention. The competitive nature of this market and the newness of these services imply that there will be tremendous growth and traffic churn in the metro/regional environment. Therefore, ROADMs should not be considered an option—they should be a requirement.
The following sections describe two different ROADM technologies.
PLC-based WSS ROADM (Wavelength Selective Switch ROADM) architecture
In a PLC- (Planar Lightguide Circuit) based WSS ROADM system, the input signal is split into two separate signals in order to accommodate both traffic expressing through the node and traffic being dropped at the node. The drop leg is then demultiplexed in order to drop those specific wavelengths dropping at the node. A multi-wavelength multiplexer is then used to combine any wavelength additions being added from the node with the express traffic. A significant disadvantage of the PLC WSS ROADM architecture occurs when the input signal is split off for the drop traffic—all of the wavelengths are hardwired to specific drop ports; the same hard-wiring occurs for add wavelengths; therefore, the PLC-based WSS ROADM does not support colorless add/drop. The PLC-based WSS ROADM also experiences high insertion loss due to the high number of wavelengths hardwired.
Figure 3: Hub site with and without ROADM interface card (RIC).
In the competitive metro/regional DWDM market, the ability to support multicast video conferencing and broadcast digital video applications on day one is becoming a requirement—therefore a Broadcast and Select architecture is required. In order to support a Broadcast and Select architecture, the WSS architecture requires all of the sites in the network to have the ROADM interfaces installed from the start. In a scenario where there is little or no traffic churn at a particular site, the service provider would still have to invest in the ROADM interface cards (RIC) instead of Fixed-OADM interface cards just to support Broadcast and Select functionality and compensate for ASE. Ideally, the service providers would prefer to have the flexibility to add the ROADM functionality to any node on their network on an as-needed basis.S-ROADM (Scalable-ROADM) architecture
The Scalable-ROADM architecture combines the benefits of both the Switch-Ring-Architecture (SRA) and Wavelength Blocker (WB) technologies. The SRA technology is based on utilizing Broadcast & Select and Optical-Unidirectional Line Switched Ring (O-ULSR) protection architectures. The add/drop configuration for all the nodes is essentially the same, with the exception of one node which contains a pair of 1x1 optical switches. The broadcast and select architecture gives visibility of all wavelengths at all the hubs across the network, which provides the required flexibility and also facilitates any-to-any connectivity. Wavelength Blockers to allow for wavelength reuse can be added as the network scales and wavelength reuse becomes necessary. The flexibility to use both the Wavelength Blocker and ROADM cards on an as-needed basis makes this architecture the most cost-effective. Thus, the low first cost and low total cost of an S-ROADM architecture makes it the ideal DWDM network choice.
Figure 4: Total cost of ownership with 25 percent card additions per year
with 25 percent discount in price year over year.
To examine the cost-effectiveness of the alternative DWDM architectures, a four-node sample network with a 2.5G wavelength originating at Site A, traversing through sites B and D, and terminating at Site C was built using Fixed, and Reconfigurable OADM DWDM (WSS and S-ROADM) architectures (Figure 1). Each of the architectures was evaluated on the flexibility, scalability, availability and economic viability of each of the solutions under the following scenarios.:
- Case Study 1: Changing the termination interface at Site C.
- Case Study 2: ROADM technology comparison: 12 wavelengths originate at Site A, of which the first 6 wavelengths are dropped at Site B and continue to Site D. The second 6 wavelengths are dropped at Site C. The traffic pattern at Site B is fixed and changing at Sites C and D.
- Case Study 3: The capital cost of adding ROADM Interface Cards to the network.
- Case Study 4: Upgrading the network infrastructure in order to support the addition of a 10G wavelength from Site A to Site C.
The following assumptions were made in performing the economic analysis:.
- The initial build-out of the network has 12 wavelengths.
- There is a 50 percent drop per site.
- Traffic churn (additions/deletions) requires two site visits per week.
- There is an average cost of $75/hour for each truck roll, not including equipment cost. Each truck roll requires an average of three hours.
- The average circuit distance is approximately 400 km.
- The average time to turn up a customer circuit in legacy DWDM networks is approximately four weeks.
- The price of new technology, such as ROADM interface cards or wave blockers, will decrease by 25 percent each subsequent year.
- The networks must support Broadcast and Select functionality from day one.
Case Study 1: Reconfigure/change the drop port at site C.
In this case study, a wavelength terminating at Site C is moved from one interface port to another. On the assumption that the new destination interface card is already present on the system C, no capital costs are incurred during this traffic management exercise. For the Fixed-OADM architecture, the network technicians would still be required to physically be present at Site C to de- provision the channel from one client port and retune it to another interface card. Both the WSS and S-ROADM architectures can address this request to change the termination interface port at Site C via software.
In the dynamic metro/regional network environment, both the customer turnover and demand for traffic churn is very high. The high maintenance requirement (OPEX) of Fixed OADM DWDM networks makes them an undesirable choice.
Case Study 2: Add 12 wavelengths at Site A and drop the first 6 at Site B and continue to Site D and second 6 at Site C (Figure 2).
In order to support Broadcast and Select functionality across the network, the WSS ROADM architecture requires all of the sites in the network to have an ROADM interface card installed on day one. In the scenario where the traffic pattern is fixed at a particular site, it does not make economic sense to invest in ROADM interfaces. The service provider would prefer to have the flexibility to use Fixed-OADM interface cards on day one and add the ROADM functionality to any site on their network as needed. For example, in our sample four-node network, assume that the first six wavelengths are being dropped at Site B and then continue on to Site D. The second six wavelengths are being dropped at Site C. The traffic pattern at Site B is fixed and not expected to change. In this scenario, ROADM interfaces in an S-ROADM architecture would only be required at Sites A, C, D. On the other hand in the WSS-ROADM architecture, all of the sites would require ROADM interfaces on day one. From this example, we can see that the flexibility to add ROADM interfaces on an as-needed basis makes S-ROADM architecture a much more cost-effective solution (25 percent initial savings assuming ROADM interface cards for both vendors cost the same).
Table 1: Summary of Case Study No. 4 results.
The flexibility of adding one wavelength at a time (S-ROADM) instead of paying for all 32 or 64 wavelengths from day one is another major economic differentiator between both architectures. The S-ROADM architecture has a 25 percent savings over the WSS architecture.
Even though service providers may not have reached full capacity in terms of wavelength utilization today, they ideally would like to have the assurance to have a scalable architecture implemented from day one in their networks. While most of the WSS architecture vendor solutions support 32 to 64 wavelengths, the S-ROADM architecture vendor solutions support up to 320 wavelengths. The ability to support 320 wavelengths today future-proofs a service provider's infrastructure to handle next-generation high bandwidth-intensive applications and traffic demands.
Similarly, the ability of the S-ROADM architecture to support from one to 320 add/drop granularity gives service providers great flexibility. The one represents the low initial first cost and 320 represents the maximum add/drop wavelength capability. Both the low granularity and high wavelength support make S-ROADM architecture the ideal choice.
Case Study 3: The capital cost of adding ROADM interface cards to the network (Figure 3).
The flexibility of the S-ROADM architecture and the ability to add ROADM interfaces on an as-needed basis provides a cost-effective approach to adding both capacity and flexibility into the network. In the sample four-site network, if only Site A required an ROADM interface on day one, it would cost a WSS-ROADM architecture four cost units as all the sites in the network need to be configured with the ROADM interfaces. As for the S-ROADM architecture, the service provider would only have to pay one cost unit on day one for the ROADM interface for Site A.
Table 2: Summary.
In telecommunications, technological advances are driving component costs for new technology down, on average, 25 percent year after year. This scenario takes this cost reduction into account and reduces the ROADM interface card cost by 25 percent over the previous year's price. After four years, the total S-ROADM architecture sample network cost is 32 percent lower than the WSS-ROADM architecture. Figure 4 shows the total cost of capital investment required as a new ROADM interface card is added to one of the sites in each of the following years.
Case Study 4: Upgrade the network infrastructure to accommodate 10G traffic addition at Site A and dropped at Site C.
In order to support a 10G wavelength, Mux/Demux DWDM, Fixed-OADM and WSS-ROADM architectures require network planners to completely redesign and build a 10G DWDM transport network infrastructure. At this higher bit rate, a higher amount of bandwidth is required due to modulation induced line-broadening. This sets a strict limit for the channel distance. Furthermore, a higher bit rate worsens the effects of disturbances such as Polarization Mode Dispersion (PMD) and non-linear effects. Amplifiers and dispersion compensation modules are required to compensate for all these limitations. Therefore, a significantly higher capital and operational investment would be incurred on day one for this capacity upgrade.
U-DWDM technology combined with the S-ROADM architecture allows service providers to upgrade their existing 2.5G DWDM transport network to carry a 10G wavelength with no network re-engineering. U-DWDM de-multiplexes the 10G client signal into four 2.5G subcarrier wavelengths, which traverse the same ITU grid wavelength window as the existing system. Because the wavelengths are transported as 2.5G, there are no effects of disturbances such as PMD or non-linear effects. Therefore, the U-DWDM 10G upgrade requires no amplifiers or dispersion compensation modules, which are needed with the other architectures.Conclusion
While multiple DWDM architectures [(Mux/Demux, Fixed-OADM and ROADM (PLC-based WSS and S-ROADM)] are commercially available in the market today, the intelligence of the S-ROADM architecture enables service providers to cost-effectively harness the flexibility of Broadcast and Select architecture, availability of Switch-Ring architecture and the scalability provided with U-DWDM technology. Table 2 highlights the finding of this study.