Advertisement
Articles
Advertisement

Implementing DWDM-based networks

Mon, 08/31/1998 - 8:00pm
Esteban Sandino, Manager of Advanced Network Technologies; and James W. Neil, Project Engineer, TCI Communications Inc.
i_9809a.gif
Figure 1: Generic DWDM forward architecture.

Dense wavelength division multiplexing (DWDM) has emerged as a practical alternative for implementation of HFC networks. The basic principle involves the transmission of multiple wavelengths over a single fiber strand to individual fiber hubs. These wavelengths are selected from the 1550 nm optical transmission grid as defined in the ITU wavelength standard. At each fiber hub, individual narrowcast wavelengths are de-multiplexed, combined with broadcast services, and transmitted to the fiber nodes. Each wavelength can be used to transport, or narrowcast, a set of unique services, thereby targeting their delivery to specific subscriber areas or node groups.

Implementation of DWDM technology allows for the relocation of active electronics from the fiber hub to the system headend. The end result is a flexible architecture allowing service segmentation at the node level, while providing a high degree of transparency at the hub level. Furthermore, the system is easily scalable, allowing quick bandwidth expansion by simply adding incremental wavelengths to already active fibers. The basic architecture is illustrated in Figure 1.

Application of DWDM technology in HFC networks is not limited to transport of forward services. Return signals arriving at the hub from individual fiber nodes can also be transported to the headend using DWDM return lasers and a significantly lower number of return fibers. This approach is illustrated in Figure 2.

i_9809a1.gif
Figure 2: Generic DWDM reverse architecture.
DWDM deployment guidelines

DWDM technology deployment requires consideration of system and market issues as well as technical issues. No two markets are alike. Each has unique requirements that must be addressed prior to deciding on a particular implementation.

The fundamental decision to be made concerns the number of wavelengths that will be initially deployed in a system. The first factor affecting this decision is the size of the service areas that will be assigned to individual narrowcast wavelengths. This is in turn affected by the expected coverage of narrowcast services. For TCI implementations, the starting point is one dedicated RF QAM carrier for every 10,000-home service area. Because a single narrowcast wavelength is able to transport up to 10 QAM RF carriers in the forward direction, and thus a mix of targeted services, there is plenty of bandwidth available for service expansion within each 10,000-home area.

After activation of the first QAM carrier per service area, and before targeted service penetration and subscription rates increase, the remaining forward bandwidth within each narrowcast wavelength remains available for other uses. In fact, this remaining bandwidth can be assigned to additional QAM carriers to serve different 10,000-home areas within a system.

The assignment of the available bandwidth within a narrowcast wavelength across multiple service areas must be in balance with the expected penetration of targeted services. In the current TCI architecture, a single wavelength is deployed to serve no more than 30,000 homes. This accommodates initial activation of targeted services. It also accommodates future bandwidth expansion to meet the requirements of each of up to three 10,000-home service areas.

Allowing this expanded coverage per wavelength permits a reduction in the number of wavelengths that must be activated initially in a system. This in turn reduces initial technology deployment costs. However, there is a second factor to consider before deciding on the number of wavelengths that will be deployed. This concerns the availability of forward RF bandwidth in a system.

i_9809a2.gif
Figure 3: DWDM forward design bandwidth limited system. (Optical switches omitted for clarity.)

Typically, for systems with less than 550 MHz of forward bandwidth, the bandwidth available for transport of targeted narrowcast services is enough for one or two 6-MHz QAM carriers. This bandwidth must be reused across all service areas. To accomplish this, a single wavelength is assigned to no more than 10,000 homes. Therefore, the number of active narrowcast wavelengths must be scaled up accordingly.

On the other hand, if DWDM technology is deployed in systems where forward bandwidth is plentiful, i.e. 750 MHz systems, multiple forward QAM carriers within a single wavelength can be assigned to multiple service areas and an aggregate of up to 30,000 homes passed.

These fundamental wavelength deployment guidelines are illustrated in Figures 3 and 4. Figure 3 illustrates a system with less than 550 MHz of forward bandwidth available per service area. Here, even though a single wavelength can transport multiple RF QAM carriers, only one of them can be activated per area. Therefore, the same 6-MHz channel carrying different content is re-used across multiple 10,000-home areas. Each area is served by an individual narrowcast wavelength.

i_9809a2.gif
Figure 4: DWDM forward design — no bandwidth limitation. (Optical switches omitted for clarity.)

Figure 4 illustrates a typical 750-MHz system. In this scenario, multiple RF QAM carriers are activated per 10,000-home area. These multiple carriers, which have been allocated contiguous frequencies in the forward spectrum, are transported over a single narrowcast wavelength. In Figure 4, all three QAM carriers transported over a single wavelength appear at all three service areas. The terminal devices within each service area are able to select the appropriate forward frequency for proper operation. In this case, increased service penetration can be accommodated in two stages: First, by activating one additional QAM carrier per service area; second, as all RF bandwidth within a narrowcast wavelength is exhausted, additional wavelengths are activated with minimal disruption to the system.

DWDM design parameters

The desired performance specifications for broadcast and narrowcast services at the output of the fiber node, after transmission from the headend through the hub, play a critical role during the design process for DWDM-based HFC networks. The specified signal performance at the output of the fiber node for TCI networks is outlined in Table 1.

To meet the minimum performance in Table 1, DWDM designs must take into account minimum operating requirements for all equipment involved. The hardware involved in a typical new DWDM deployment includes:

  • Externally-modulated 1550-nm transmitters for the transport of broadcast services;
  • Directly-modulated 1550-nm transmitters for the transport of narrowcast services;
  • Erbium doped fiber amplifiers (EDFAs) for optical amplification of both broadcast and narrowcast signals;
  • DWDM couplers for multiplexing of broadcast and narrowcast signals; and
  • Optical switches.

Operating specifications for DWDM equipment define a basic set of DWDM design guidelines:

Externally-modulated broadcast transmitters. These will transmit at output powers of either +6 dBm or +9 dBm. The transmission distance from the headend to the hub site will determine if an additional EDFA is required to boost optical signal power. A +9-dBm transmitter for short optical links is a less expensive solution than a +6-dBm transmitter coupled with an EDFA. However, if a headend broadcasts signals to multiple hubs, the latter option allows for the split of the EDFA output power and spreads out the higher hardware costs across multiple receiver locations.

EDFAs. As indicated above, EDFAs can be used either at the headend to boost available optical power and feed multiple hubs, or at the hub itself to amplify the incoming optical signal and continue transmission to the fiber node. The available output power of an EDFA can range from +14 dBm to +22 dBm. Different vendors may offer even higher output powers. However, non-linear stimulated Brillouin scattering (SBS) fiber effects limit the maximum optical power that can be launched into a fiber to no more than +17 dBm. Therefore, higher-power EDFAs can only be used with optical splitters to feed multiple locations.

At the headend, a good DWDM design must ensure an optimal optical input power level of +5 dBm into the EDFA. Assuming the maximum allowed launch power of +17 dBm, optical signal reach will be determined by the type of receive equipment at the hub. For standard optical receivers, an optimal input level ranges from 0 dBm to +2 dBm. This translates into a link loss range of 15 dB to 17 dB at 1550 nm. If a second EDFA is in the path, as in the case of most hubs feeding multiple nodes, the optimal input level ranges from +3 dBm to +5 dBm or a maximum 12-dB link loss range at 1550 nm.

The type of EDFA selected for a particular design will vary by application. EDFAs with optical input isolators should be used at all receive locations to minimize potential signal distortion caused by scattering of light over long optical links. No input isolation is required for transmitter and EDFA combinations typically used in headend installations.

Directly modulated narrowcast transmitters. These lasers can transport up to 10 digital QAM carriers for delivery of narrowcast services and are commercially available in the eight ITU wavelengths listed in Table 2. The output of these transmitters is typically in the +7 dBm to +10 dBm range, and eight wavelengths can be multiplexed on a single fiber. The total optical power launched into the narrowcast fiber is determined by the logarithmic addition of the power of each individual wavelength.

A typical DWDM design involving narrowcast lasers will not require an EDFA at the headend. The available optical power per wavelength, after accounting for insertion losses of DWDM couplers, is enough to ensure a minimum of -5 dBm per wavelength at the receiver hub site. This is after a worst-case transmission loss of 8 dB at 1550 nm from headend to hub, which is typical of TCI implementations.

At the secondary hub, an EDFA is required to amplify all narrowcast wavelengths prior to de-multiplexing and transmitting them to the fiber nodes. For optimal performance, the total optical power at the input to this EDFA should not exceed +7 dBm after accounting for all wavelengths present.

i_9809a4.gif
Figure 5: DWDM design example — Baton Rouge. (Optical switches omitted.)

DWDM couplers. These are used in a typical TCI design to multiplex a maximum of four wavelengths on a single fiber, although eight-port couplers are also available. In some cases, DWDM couplers at the headend can be replaced with standard optical couplers to achieve the same objective. This may be an attractive alternative, given the premium that must be paid for DWDM couplers. The choice of one over the other will be determined by the fiber-loss budget. Four- and eight-port optical couplers can be used to combine several wavelengths but exhibit much higher insertion losses: 6.9 dB and 10.5 dB, respectively, vs. 2.5 dB and 4.5 dB for four- and eight-port DWDM couplers. This option is not available at the receiver sites, as DWDM couplers must always be used to separate narrowcast wavelengths prior to transmitting them to specific nodes.

Optical switches. These are designed into TCI secondary hub sites, and influence DWDM design. Optical switches select automatically between primary and backup signal paths, and their associated insertion losses will lower available optical power budgets. Addition of optical switches may force the DWDM designer to increase optical output levels at the headend, and thus increase EDFA hardware costs.

i_9809a5.gif
Figure 6: DWDM design example — Grand Prairie, Dallas. (Optical switches omitted for clarity.)

Optical node receivers. The input to the fiber node will be 1550-nm or 1310-nm broadcast signals multiplexed with a 1550-nm narrowcast wavelength. The desired optical level into the node for broadcast services is between +1 dBm and +2 dBm. To achieve the performance specified in Table 1, page 36, the DWDM designer must minimize the narrowcast wavelength noise contribution to broadcast services. This is achieved by keeping the optical level of the narrowcast wavelength reaching the receiver at the node at a level that is 8 dB to 10 dB below that of broadcast optical levels. To compensate for the lower RF carrier levels this would generate for targeted services, narrowcast lasers at the headend operate with higher RF input carrier levels, i.e., higher modulation index per carrier, to achieve proper RF levels at the output of the node.

DWDM field implementations

Final hardware and design choices leading to field deployment of DWDM technology will be influenced by additional factors such as:

Existing broadcast fiber architectures. A system may already have a 1310 nm-based distribution system in place. The DWDM designer must be particularly careful in this situation when deploying additional wavelengths for narrowcast services. DWDM signal transport from the headend to the hub is not an issue, as broadcast and narrowcast fibers remain separate. At the hub, however, narrowcast wavelengths must be optically multiplexed with already existing broadcast services. To minimize the additional insertion losses associated with the installation of an additional coupler in the path to the node, standard 1310/1550 WDM couplers should be used. This ensures that existing 1310-nm broadcast levels at the nodes are only reduced marginally after the introduction of a narrowcast wavelength. Care must be taken to ensure narrowcast optical levels at the node are kept 8 dB to 10 dB lower than broadcast optical levels.

i_9809a7.gif
Figure 7: DWDM design example — Grand Prairie, Dallas.

Average number of homes passed per fiber node. If this average is low, i.e. 300 to 600 homes, the increased number of nodes forces a high number of optical splits for broadcast and narrowcast wavelengths. In general, if a high-power, +22-dBm EDFA is used, the number of times a signal can be split is limited to 24 to ensure appropriate optical levels at the node. Thus, higher node counts force not only the activation of additional broadcast lasers, but also initial activation of a higher number of narrowcast lasers.

Available fiber counts. This is a factor when implementing DWDM designs for return signal transport. High fiber counts between the headend and secondary fiber hubs allow for the allocation of dedicated fibers per return narrowcast laser rather than paying a premium for the purchase of DWDM couplers.

i_9809a9.gif
Figure 8: PEG insertion design example.

Expected initial penetration of targeted services. For markets with high growth potential, consideration must be given to limiting the area served by individual narrowcast wavelengths to 10,000 homes. This may be the case even for 750-MHz systems. In such cases, the large number of forward narrowcast wavelengths may force the use of eight-port DWDM couplers to maximize use of available fibers.

Channel lineup customization. Broad-cast and DWDM signal transport using 1550-nm technology results in smaller hubs. Most signal processing is performed at the optical level because only EDFAs and couplers are required to feed fiber nodes out of the hub site. However, this may present a challenge when customized lineups are required for each hub location.

Some systems may have requirements to provide PEG and other programming that is targeted to specific communities or franchise areas. In these situations, having individual 1550-nm broadcast feeds to multiple hubs greatly increases implementation costs. A better solution is to use lower-cost 1310-nm transmitters for the transport of PEG and other area-specific content.

These services are transported to their associated secondary hubs on dedicated fibers. At the secondary hub, the 1310-nm signal is repeated as required and optically multiplexed with 1550-nm broadcast and narrowcast feeds before transmitting them to the node. This minimizes the number of 1550-nm broadcast feeds, which reduces implementation costs, yet allows for lineup customization at the secondary hub level.

i_9809a8.gif
Table 1: TCI DWDM performance specs. (Measured at the fiber node output.)

It also means that the fiber node now receives up to three different wavelengths: 1550 nm for broadcast, 1550 nm for narrowcast and 1310 nm for local content. This will introduce only marginal degradation in the performance of the broadcast signals, provided the levels of narrowcast and PEG wavelengths are kept 8 dB to 10 dB lower than the broadcast levels at the input to the node.

Figures 5, 6 and 7 illustrate examples of DWDM implementations. Figure 5 shows the design implemented in the TCI system in Baton Rouge, La. Broadcast signal distribution from the secondary hub to the fiber nodes was already in place using 1310 nm technology. Narrowcast lasers operating at 1550 nm were added at the headend and multiplexed over a single fiber for transmission to the hub. At the hub, narrowcast wavelengths are amplified, de-multiplexed, and individually combined with the 1310 nm broadcast signal before transport to the node. Standard 1310/1550 WDM couplers are used to combine broadcast and narrowcast signals. This approach resulted in less than 0.5 dB of additional insertion loss for the 1310-nm broadcast signals, therefore minimizing the impact on their performance at the node.

Figures 6 and 7 show a design implemented in the TCI-Dallas system. Broadcast and narrowcast signal distribution is implemented using 1550 nm technology. Broadcast transport from the Arlington headend to two field OTNs (optical transport nodes) uses no more than two EDFAs in cascade to reach all fiber nodes, while maintaining the performance in Table 1.

Narrowcast services are transported using eight different wavelengths multiplexed over a single fiber. All narrowcast wavelengths are amplified and de-multiplexed at OTN 1. Six of them are individually combined with the broadcast 1550-nm signal and split to feed a number of fiber nodes. The other two are transported over two dedicated fibers from OTN 1 into OTN 2, where they are finally combined with the broadcast wavelength and split to feed the remaining nodes in the system.

i_9809a10.gif
Table 2: Available DWDM wavelengths.

Figure 8 is an example of a design for optical insertion of PEG channels at the hub. In this case, the 1310 nm lasers carrying PEG programming are located in the hub. Standard 1310/1550 WDM couplers are used for insertion of PEG programming. Each 1310 nm laser targets a group of fiber nodes as required. Downstream from the 1310 nm insertion point, 1550 nm narrowcast signals are inserted using standard couplers prior to final transport into the nodes.

DWDM is a practical alternative now available to HFC network designers. It has been deployed successfully in a number of TCI markets. The technology allows for the narrowcasting of high-speed data and other services, flexible network segmentation and a reduction in the number of active electronics that must be deployed outside the system headend. Bandwidth expansion is easily achieved through the activation of additional narrowcast wavelengths per service area with minimal interruption to existing services.

e-mail: Sandino.esteban@tci.com


Acknowledgements
The authors wish to thank personnel from Harmonic Lightwaves and Antec for some of the data used in the preparation of this article.

Topics

Advertisement

Share This Story

X
You may login with either your assigned username or your e-mail address.
The password field is case sensitive.
Loading