Forewarned is forearmed: Stuff you ought to know before you install.

Over the last two years, RF return over optical fiber has gone from its infancy to a widely deployed technology. Today, this technology, better know as RF over glass (RFoG), is a viable option in many circumstances for updating traditional HFC architecture.

In a traditional HFC architecture, a single optical node combines all of the RF signals from individual homes and then sends the aggregated signals over fiber via a single upstream continuously transmitting laser. In contrast, in an RFoG deployment, each optical network unit (ONU) has its own upstream laser. These optical signals are "combined" by an optical splitter (typically 1 by 32). The lasers do not transmit continuously, but instead turn on only when RF power above a certain threshold is sensed at the laser driver RF input. This is what is known as a "burst mode laser."

Through the cooperation of vendors and operators, an RFoG standard has now been completed by the SCTE's IPS SP 910 Working Group. In the meantime, thousands of residential customers have had optical fiber extended to their homes and are served by RFoG devices. Let's explore the lessons learned from these deployments for the benefit of service providers considering their own deployments.

Lesson #1:  Watch for what wavelengths are supported on WDMs
One of the first issues addressed by the SCTE RFoG standards committee was the optical wavelength(s) supported. Initially, vendors were using a variety of wavelengths, but the standards committee settled on 1310 nm and 1610 nm.

The 1610 nm wavelength is used by operators that offer, or intend to offer, GPON or EPON overlays to their system. The most frequently used "other" wavelength was 1590 nm.

Unfortunately, some operators learned that some WDMs they had deployed supported only the 1590 nm wavelength and had to switch out the WDMs to support 1610 nm. Service providers need to make sure that the optical components they source are compliant with the RFoG standard and meet the performance specifications in the standard.

Lesson #2: RFoG may require more fiber in the feeder plant
Many cable operators that were interested in deploying RFoG quickly learned that they had insufficient fibers in the feeder plant.

Since these operators had designed their optical feeders assuming RF return traffic would be aggregated at the optical node, they suddenly found themselves starved for fibers to support all of the upstream RF traffic. This caused many cable operators to delay, or even decide against, deploying RFoG.

Luckily, the vendor community is stepping forward to address the unique needs of the RFoG market. Several vendors have developed optical nodes that are modular enough to convert from receiving RF signals to receiving optical signals and aggregating the return path signals before sending upstream from the node.

Other vendors have combined WDMs and two-way EDFAs in temperature-hardened products that enable GPON and EPON OLTs to be placed out in remote cabinets, thereby eliminating the need to reinforce the number of fibers in the feeder plant.

Lesson #3: Return path receivers may not be specified for RFoG
Some return path receivers used in conjunction with HFC nodes may not have the necessary gain and CNR performance for RFoG. However, new receivers are typically designed "RFoG-ready" with the necessary improved performance.

There are also receivers currently deployed that will work with RFoG, but because they were designed for a return path signal that is always on and is an aggregate of many homes, the bursty nature of RFoG causes them to alarm when the signals are perceived to be too weak.

Operators have learned that since the receivers are not specified for RFoG but receive the signals just fine, they have had to go into the management systems and reset the default values for the alarm thresholds.

RFoG Reference Architecture

Lesson #4: RFoG devices should be intelligent and capable of being managed
The cable industry has invested millions of dollars in developing alarm and monitoring systems for all of the equipment used in the HFC plant. Unfortunately, many of the early-to-market RFoG devices are unmanaged.

So while extending fiber-to-the-home creates a more robust and trouble-free network, the inability to collect alarms and monitor and manage these RFoG devices can take cable operators a step backward and potentially make the network less reliable.

One example of this would be an RFoG device at a home that is funneling noise into the system that the operator cannot turn off remotely. A second example is that because critical services are being run over RFoG devices, they may need battery backup.

Therefore, collecting statistics and alarm information of the status of batteries becomes an important requirement. Several vendors have now introduced managed RFoG devices that allow operators more information and control of RFoG devices.

Lesson #5: RFoG impacts on the forward path 
The most immediate impact of RFoG is on forward path transmitters and EDFAs. RFoG requires higher-rated SBS threshold ratings - typically above 20 dB. A second forward path issue to consider is that the RF signal will generally be hotter when they reach the home in RFoG systems.

While this higher-level signal is usually helpful in handling more RF splitters within the home, it can also lead to more RF signal leakage on open splitter ports and non-terminated connectors in the home.

Lesson #6: Optical beat interference is a factor to consider
Many cable operators considering RFoG have been concerned about a phenomenon called optical beat interference (OBI). Since the RFoG architecture is based on multiple optical transmissions on the same optical fiber upstream from the optical splitter, OBI will occur if the transmissions are simultaneous and laser wavelengths are "close enough" together.

To date, field deployments have not seen any significant impacts from OBI. In a presentation to the SCTE RFoG standards committee, Calix and AOI (Applied Optoelectronics Inc.) analyzed the conditions under which OBI occurs and conducted a statistical analysis of the impact of OBI on upstream RFoG communications.

The study found that the probability of errors from OBI is low (0.1%). This probability is an average over time, so the error rate can be much higher for short periods of time. The study indicates that the best solution to OBI is to reduce or eliminate the chance of simultaneous traffic.

This is unlikely to be a problem when the upstream traffic is set-top box return traffic, but the probability goes up with higher levels of QAM and channel bonding. As a result, the RFoG standards committee is including recommendations and approaches to OBI, particularly when it comes to DOCSIS 3.0 and upstream bonded channels.

RFoG deployments are rapidly approaching 100,000 subscribers today and growing, and this figure is expected to grow significantly in the coming years. The migration of cable operators from HFC to fiber-to-the-home is underway, and RFoG will play a critical role as cable operators transition and migrate to this more robust and operationally efficient network.