Power from the people: Reverse power feeding
Ultra-broadband over copper requires the installation of compact distribution point units (DPU) deep in the network. This article explains how reverse power feeding (RPF) will make G.fast ultra-broadband a reality.
To provide the ultra-broadband connections subscribers demand and remain competitive, service providers are calling upon the last drop of their copper infrastructure to provide ever-greater levels of performance. But thanks to new digital subscriber line (xDSL) technologies like VDSL2 vectoring and G.fast, the all-important range from 100 Mb/s up to 1 Gb/s is now within reach.
However, such extreme speeds require short copper. G.fast, for example, works best on loops of less than 250 m (0.15 miles).
As service providers move the fiber closer to customer premises, the traditional digital subscriber line access multiplexers (DSLAM) that were designed to handle many endusers and act as bridge between the network fiber and the last copper drop are no longer practical. Instead, the industry is adopting distribution point units (DPUs), which are much smaller, consume much less power, and are easily deployed in greater numbers close to subscribers. This deployment model is generally referred to as fiber to the distribution point (FTTdp) or fiber to the building (FTTB).
Traditional DSLAMs were designed for installation in the central office or in service provider owned cabinets that have access to power. DPUs don’t have the same luxury. Because they need to be in close proximity to customer premises, DPUs are installed in a variety of non-traditional locations, including:
- Attached to external walls of homes
- In the basement of apartment buildings or at the level of the apartment floor
- On telephone poles
- Under manhole covers
- In pedestals
In many of these locations, access to power is difficult and/or expensive. Reverse power feeding (RPF) addresses this difficulty. RPF draws power from the customer premises over the same copper pair used for data service. Reverse powering is included in the FTTdp reference model in the G.fast standard (G.9701) but RPF is also applicable in combination with VDSL2.
Figure 1 shows alternative means to power a DPU that is installed close to the end-user. RPF is expected to become a common deployment model for FTTdp and FTTB.
The challenge of RPF: Technical complexity
While the concept of RPF is fairly simple and straightforward, making it work requires careful definition and engineering and industry support.
The basic components of the RPF system as defined in ETSI are depicted in Figure 2:
- At the end-user the power source equipment (PSE) injects the power into the telephone line via a power splitter (PS). The PS separates the power and the xDSL broadband signals. An optional battery (BAT) provides backup power in case of a power failure condition. A power management transceiver (PMT) supports the exchange of a power management protocol with the DPU for start-up and monitoring. The PSE can be a standalone device or integrated in the CPE or home-gateway.
- In the DPU the power is extracted by the power extractor (PE) that is coupled to the line via a PS. The power supply unit (PSU) combines the power from the different PEs and converts it into the proper voltage levels. The reverse power control entity (RPCE) coordinates the RPF functions over all active lines. A PMT per line performs the power management protocol with the corresponding PSEs.
The RPF system is designed to power a DPU with a set number of lines, typically 1, 4, 8, 12 or 16. In addition, the RPF system must:
- Allow the DPU to work with any number of active users at a point in time, even a single one.
- Provide efficient power transfer from the end-user to the DPU in order to maximize the distance over which the DPU can be reversely powered for a maximum PSE input power as defined by the RPF Class.
- Allow error-free data transmission on active lines when other DPU lines disconnect or new lines activate.
- Survive short interruptions (in the order of tens of milliseconds) in power supply, e.g., due to a micro-interruption or bad contact on one of the active DPU lines.
- Ensure a fair distribution of power supply over the active users, even in case of loop length differences, and taking traffic and actual power state into account, by tuning the DPU load on each line.
- Allow interoperability of the DPU with a PSE device compliant to one of the RPF Classes defined in, with DPU and PSE possibly supplied by different vendors.
- Allow end-user self-installation and activation.
- Not create a safety hazard in case telephone sets or other communication devices (e.g., fax machine or alarm system) are accidentally connected to the in-house telephone pair that also carries RPF.
- Have minimal impact on the performance of G.fast or VDSL2 carried over the same twisted pair.
Some DPU applications impose additional requirements on the RPF system and the DPU:
Zero-touch operation of the DPU. In this model the operator does not need to revisit the DPU after installation, e.g., to connect new customers. Lines are put in bypass-mode at installation to pass transparently legacy narrow- or broadband services. When the end-user activates RPF, the line automatically switches from bypass mode to the new broadband service from the DPU.
Support of battery backup at the end-user. This is typically only required in combination with voice service (VoIP to the CPE or residential gateway). In case the end user falls back on battery operation the power drawn from the battery by the RPF system should be minimized.
Implications of RPF: low consumption and persistent management
RPF also imposes requirements on the DPU and its management: Low power consumption Power consumption of DPUs is critical, especially for compact DPUs with RPF. Figure 4 shows two extreme but realistic working points.
- If only a single line is active, that line must be able to power the DPU. Therefore, the DPU power consumption should scale with the number of active lines.
- If all DPU lines are active, the total power dissipation must remain within the thermal limits of the passive cooling, imposed by the housing and environmental conditions.
Offline management Even if no lines are active, the unpowered DPU should not fall off the operator’s radar. At a minimum, operators want the capability to pre-provision lines of unpowered DPUs. This requires a persistent management agent (PMA) that resides outside the DPU. Other possible functions of the PMA are translation of the management protocol (e.g., SNMP to the EMS or OSS, NETCONF/YANG to the DPU), and the storage of performance monitoring history.
Power splitter function RPF must work while VDSL2 or G.fast services are carried on the same line. This requires isolation between the low frequencies used for RPF and the high frequencies that carry the utra-broadband data. This separation comes from PS filters in both the DPU and the customer premises. The electrical requirements, combined with the minimal noise injection, minimal PCB footprint, and overvoltage protection make the design challenging.
The cost of RPF: Service challenges
RPF also creates challenges for delivering POTS service and metallic line testing (MELT). While these aren’t insurmountable, solving them often comes at a cost.
RPF and POTS: Compatibility with analog voice on the same twisted pair. Most operators are expected to use VoIP up to the end user in combination with FTTdp as this is also a common deployment model with FTTH. In case there is a need to continue legacy telephone service (POTS) from the CO there is a conflict between RPF and POTS DC feeding and low-frequency signaling. Two possible solutions to this problem are:
- Use of a separate twisted pair for the POTS.
- Conversion in the DPU of the POTS signaling into a format that is compatible with RPF. In this case POTS adapters (e.g., dongles) are needed in-house in front of the phone sets to reconstruct the POTS DC feeding and signaling. POTS adapters must either be fed locally by a battery or mains, or by the PSE. Using the PSE reduces the power budget for the DPU. See footnote for more details.
A similar situation arises if POTS is generated in the DPU (termination of VoIP from the network in the DPU), or if the voice signal that is regenerated in the CPE or home gateway (with integrated ATA function) is re-injected in the in-premises wiring (see bottom part of home network in Figure 3).
If lifeline support is required when the PSE is down, a battery is needed within the subscriber premises to keep the DPU and adapters powered.
RPF and MELT: Support of MELT from the DPU on the same twisted pair as used for RPF requires a controlled interruption of RPF for the duration of the test and the availability of an alternate power source for this period.
Emerging reverse power feeding
The telecom industry as a whole (providers, equipment manufacturers, standardization organizations) recognizes the importance of reverse power feeding and is spending significant effort to turn RPF into a reality.
The ETSI technical specification for RPF was published in September 2014. The Broadband Forum (BBF) is also working on RPF, with ongoing discussions and no technical reports yet released. BBF plans describing fiber to the FTTdp scenarios and related operations, administration and management requirements, taking RPF into account.
Alcatel-Lucent is contributing to these RPF efforts and to G.fast standardization.
-  ETSI Reverse Powering Specification TS 101 548 v1.1.1: European Requirements for Reverse Powering of Remote Access Equipment.
To contact the author or request additional information, please send an email to firstname.lastname@example.org.