In 1881, Alexander Graham Bell’s company was granted a patent for what was, essentially, the local telephone loop. This system connected every telephone to the switchboard and, later, the switch via two wires that completed a circuit using DC voltage. For the better part of the twentieth century, this system used 48 V that powered analog telephones using a power source independent from the home or office where the telephone resided, such as a battery. This created the expectation-born of experience-that if the power went out, the telephone service still would be operational. That level of reliability later came to be associated with 911 lifeline service, where telephones remained operational even in the event of an emergency.
But operating telephone service via battery power originally had nothing to do with emergency response; it was simply a means to provide DC voltage to operate the telephone circuit.
Power over Ethernet (PoE) began as an attempt to provide the same experience to consumers who use network-based telephone service or voice over Internet Protocol (VoIP). The standard for PoE-IEEE 802.3af-provides 48 V of DC power over two of four available pairs on essentially any Category UTP cable. After line losses, about 13 W are available. This standard has become so ubiquitous, it has had a “tail wagging the dog” effect on the design of LAN infrastructure. However, not all the information about how to implement VoIP is true.
The Fiber Optic LAN Section (FOLS) of the Telecommunications Industry Association (TIA) has been an advocate of fiber LAN technologies since its inception and has developed interactive models that help users compare the architecture and potential cost differences between traditional hybrid copper-fiber networks and those that are mostly fiber. Recently, we’ve found that even when the cost model shows fiber networks to be competitive or even less expensive than copper-fiber networks, many users still do not install fiber because they believe that fiber-based networks can’t support PoE. Some even believe that VoIP and PoE are codependent, and therefore VoIP doesn’t work on fiber either. Unfortunately, the message that comes through loud and clear from many-especially in the enterprise-is that VoIP is not a fiber technology and if you want to implement PoE, don’t use fiber.
In reality, VoIP works with fiber. It works with copper, and it even works with wireless because it is media independent. In fact, it’s about as different from traditional telephone service as you can get. The table provides a comparison between VoIP and traditional PSTN technology.The fact that VoIP is portable means it can be used on more than just a VoIP telephone. It will work on a desktop computer connected to a fiber network with a fiber network interface card (NIC), a laptop computer using a wireless card, or even a PDA.
Many people may not realize that when a telephone service provider takes fiber to the home or a cable company provides telephone service, they are implementing VoIP. In fact, they are doing it on different media, neither of which supports PoE, and therefore they have to deploy other methods of providing 911 emergency service, usually with the use of onsite batteries.
Implementing VoIP along with PoE takes some of the independence out of VoIP. It makes it media dependent (copper), and its portability is reduced by forcing it to remain within range of dedicated power circuits, universal power supplies, PoE switches, and patch panels. Some would argue this is the price you pay for providing emergency phone service-and they may be right if no alternatives are available.
Generally, PoE does not work with fiber, but it can work in a complementary with fiber. One fiber architecture that can be combined with PoE is fiber to the telecom enclosure (FTTE). Modeling by FOLS demonstrates that FTTE brings both price and performance advantages; it is one of the least expensive architectures that users can deploy, and it offers extreme flexibility and high throughput. TIA FOLS is in the process of revising its downloadable cost model to include PoE in FTTE and other architectures. The figure illustrates the FTTE and PoE model that FOLS will incorporate.
In an FTTE architecture, fiber cable is deployed from a main equipment room (ER) to a zone in the work area and into a telecom enclosure (TE) mounted on a wall or ceiling. The enclosure houses patch panels and switches that convert the fiber to copper, which is then deployed the short distance to outlets within the work area. FTTE makes optimum use of the capabilities of copper and fiber by extending the use of fiber beyond the riser and into the horizontal, using copper only for the short distances within the work area.
The FTTE architecture is covered by the new TIA/EIA-569-B Pathways and Spaces standard, which defines the TE “space,” and TIA/EIA-568-B.1 Addendum 5, which defines TE cabling.
FTTE easily supports PoE through the use of midspan patch panels or PoE-enabled switches. The beauty of this architecture is fourfold:
• It allows for fiber to be deployed almost all the way to the end user, taking advantage of the extended distance that fiber offers as well as the much higher bandwidth and application speeds it can support.
• It can offer a totally nonblocking network architecture, so throughput is much higher than traditional architectures. (FTTE networks tend to have a higher ratio of uplinks to end-user ports, thereby increasing the possibility of nonblocking performance.)
• The flexibility of this architecture allows for the deployment of PoE, wireless, and/or fiber to the desk (FTTD), if needed, while moves, adds, and changes become straightforward and simple.
• Depending on the density of the telecom enclosure, the installed first cost can be substantially lower than any other standards-based architecture.
Most importantly, FTTE is a win-win proposition. It increases the use of fiber, doesn’t change the desktop environment that people are used to (i.e., copper outlets and patch cords), deploys PoE, and is less expensive to boot.
However, we have encountered two issues with FTTE. First, it does require more dedicated electrical outlets for the enclosures; second, there appears to be a dearth of enclosure cabinets to choose from as manufacturers have seemed slow to pick up on the benefits of the architecture.
Independent 911 services seem to be one of the barriers to deploying fiber-based networks; however, there are several alternatives to PoE for providing independent power.
Dedicated power. Dedicated electrical outlets have been used for many years to provide clean or emergency power for various types of equipment. They do require more complex planning and construction of the electrical plant, but the critical function is to provide a certain percentage of telephones with power from dedicated circuits in an emergency. In fact, for enterprises such as hospitals, the entire building has emergency power, thus eliminating the need to use PoE as a source of emergency power.
Battery power. We have become very dependent on batteries for the sake of our mobility. Cell phones, iPods, laptops, and cordless telephones are commonplace. Networks and many desktop computers are backed up with battery power. FTTH relies on batteries for 911 emergency powered telephones. The beauty of battery technology is that it preserves our portability, especially when it resides on the device itself. And battery technology has not been pushed to its limits; devices such as hydrogen power cells are lurking on the horizon. In addition, core electronics get smaller and consume less power to accomplish the same job over time.
Dual copper and fiber networks. It is possible to install FTTD alongside a traditional copper network using PoE that supports only the VoIP telephones. At first, this approach seems time-consuming and expensive, but in reality it is very similar to deploying a data network and a voice network (except that the voice network is VoIP). The two networks can perform computer-telephone integration through the main ER, and the VoIP network can rival a traditional PBX network in cost and function. VoIP also consumes low bandwidth, allowing the VoIP network to be designed to its own standards and neither competes with nor affects the bandwidth of the FTTD network.
Power over fiber. If VoIP phones were able to consume or recharge on a few watts, there is a budding technology that someday could provide emergency power directly over fiber. JDS Uniphase last year acquired Photonic Power Systems, developer of a laser-photovoltaic power converter system that produces electrical power over fiber. This system can currently deliver up to 1 W of power over 500 m of multimode fiber and promises more power in the future.
PoE originally was intended as a method for providing emergency power for VoIP telephones but has since grown to become a generic low-voltage power system. The original standard was designed for 13 W, and there is a current effort-IEEE 802.3at-to increase the standard to 30 W. This flies in the face of the way technology in general has changed: Over the years, CPUs and other microelectronics have consistently consumed less power, not more. Back in 1995, for example, Intel processors consumed 3.3 V and featured 300,000 gates per device, while the company’s latest dual-core processor consumes just 1.5 V and has upwards of 200 million transistors. The need for more power is not attributed to the needs of VoIP but to the desire to power everything from wireless access points to desktop fans to LED-based Christmas lights.
One of the original intentions of PoE was to reduce costs and complexity by combining Ethernet cable and low-voltage power. It was meant to reduce the need for electrical outlets. However, the National Electric Code (NEC) and local building officials already dictate the number of outlets required in a building. Moreover, it’s not just the cabling that’s affected but the electronics as well. Switches and patch panels must be replaced to support first one and then the second version of PoE. In addition, the cooling requirements of cabinets and closets suddenly increase. Providing the power isn’t enough, either; network managers must also provide the backup generators and universal power supplies to support this power.
While these issues represent a tradeoff, the sacrifice that we shouldn’t be willing to make is the one that includes fiber optics. Fiber represents the long term, long haul, long stride of progression in providing significantly more bandwidth to more people in a more efficient way. Those of us who have dedicated years of labor to advancing fiber-apologizing the whole time for its inherent advantages-should look at PoE as one of several solutions to a specific VoIP-based problem and not get sidetracked.Daniel L. Harman is the chair of the Fiber Optics LAN Section of the Telecommunications Industry Association and an applications engineer for 3M’s Communications Markets Division. Visit FOLS on the Web at www.fols.org.