Copper to the Home

Copper to the Home

By: Nathan Stratton nathan@robotics.net
This is a draft document, please email me with changes or suggestion nathan@robotics.net

Interdiction

Today many CLECs are working to build high-speed broadband networks to provide a wide range of services. This document will take a look at current broadband technologies weaknesses, look at some interim broadband solutions, and finally long term solutions. I will admit that I may not be an unbiased source of information, but I hope you will find it at least interesting.

Current Technologies

DSL

DSL is one of the fastest growing high-speed technologies being deployed today. All RBOCs are deploying some form of DSL in their markets and many ILECs and CLECs have been very aggressive in such deployment.

Normal voice grade communications uses the frequency of 300 Hz to 3.4 kHz this band covers most of the human speech spectrum. Using a large frequency band (up to 30 MHz with VDSL) and with advances in DSP technology combined with coding methods such as Carrierless Amplitude Phase Modulation (CAP) and Discrete Multi-Tone (DMT) large amounts of data can be send down a copper pair.

Current offerings

DSL

The term DSL was originally referred to basic-rate Integrated Services Digital Network (ISDN-BA). ISDN-BA allows full duplex data transmission at 160 kbps on a single twisted pair. It uses frequency band of about 10 kHz to 100 kHz with the peak power spectral density at around 40 kHz. Normally ISDN-BA is formed with two 64 kbps bearer channels known as B channels and 16 kbps signaling channel known as a D channel. A user could use 128kbps for data and if a call came in using the D channel the data can be knocked down to 64K and the second B channel can be used for voice.

HDSL

HDSL has been widely used by ILECs to transport T1s or E1s. It supports 1.544 Mbps or 2048 Mbps using two pairs and a frequency band of 100 Hz – 292 kHz.

ADSL

Asymmetric DSL (ADSL) is the DSL form that most people have heard about. It was originally designed to transport Video on Demand (VoD) and other highly asymmetrical services. ADSL is normally provided over a single pair with a frequency band of 25.875 kHz to 1.104 MHz. This provides a max big rate of 8 Mbps downstream and 640 kbps upstream. In order to achieve such bit rates over a loop length of up to 18,000 feet Forward Error Correction (FEC) needed to be used. The downside of such FEC techniques added a lot of delay to the system. ADSL has a delay of around 15ms compared to HDSL delay of no more then 1.25ms. ADSL is also a best effort services and should not be used where reliable full duplex bandwidth is needed.

SDSL

Symmetric DSL (SDSL) provided symmetric DSL speed from 192K to 1.536 Mbps. It is normally used for business grade services, or where symmetric services is needed.

VDSL

Very high-speed DSL is currently the highest speed DSL services available. Using one pair VDSL supports 24 Mbps downstream and 4 Mbps upstream using a frequency band of 300 kHz to 30 MHz. One of the downsides of VDSL is it requires very short loop lengths of less then 2,000 feet.

DSL-Light

There is a new push among PC and modem manufactures to apply the same principles as ADSL using DMT (ADSL is CAP). This should allow for plug-and-play DSL modems and provides the user up to 1.5 Mbps downstream and 512 kbps upstream.

Problems

Load Coils

As the cable plant in the US was constructed there was a need for longer and longer loops. Typically 66 mH coils would be spaced 4,500 feet apart in the network. The coils dramatically cut off frequencies above 4,000 Hz. This removed a lot of lines from the line and aloud ILECs to deploy longer loops. What is good for a voice network is not good for a data network. To make matters worse many ILECs got carried away with load coils and loaded many loops that did not require it. Load coils are a big problem for DSL because all forms of DSL start over 4,000 Hz. If DSL is to be deployed on such loops all load coils need to be removed.

Bridge Tap

A bridge tap is when a cable is tapped in one or more locations to extend the loop to other locations. Lets say we have 3 buildings that are fairly close to each other A, B and C. One year building A may need 50 pair, building B may need 20 pair, and building C may need 80 pair. The telco could run the required copper to each building, but a few years later as the business in the building change the requirement may change. Building A may now need 20 pair, B 100 pair, and C 60 pair. One way to sold this problem is to run a 200 pair cable into building A, the outside that building the same cable would be taped and sent to building B and taped again for building C. Now all 3 buildings have access to any of the two hundred pair and growth is easier to handle.

This may work great for voice, but can play havoc with DSL. This is because when the DSL signal hits the unconnected tap it causes a reflection that cancels the signal. The amount of bridge tap and the distance from the CO has different impacts on DSL. In most instances bridge tap must be removed if DSL is to be provisioned on a loop, especially if the bridge tap is closer to the CO. In some cases limited bridge tap farther from the CO can be left on the loop.

Spectral interference

Cable plant is comprised of a pair of wire twisted together, 25 pairs are then twisted together to form a binder group. The binder groups are wrapped with a colored tape and grouped with as many binders needed for a cable. Typically 2,200 trunk cables would be divided into smaller 100 – 400 pair cables that enter buildings. In residential areas the cables would be spliced in a splice case to a 2 – 6 pair service cable that runs to a house.

Plant Issues

The phone plant in the US has aged quite a bit since it was put in the ground in the decades ago. As this plant degrades it can cause many problems for higher speed services. Over time small amounts of water seep into the cable destroying it’s electrical properties. Copper wires breakdown if they are not protected properly, they may work fine for normal POTS, but may pass little signal above 4 KHz. In the past to protect the cable plant from lightning and other surges carbon insulators and other cable protectors were installed. Many times they must be modified or replaced if DSL services are to be provisioned on the pairs.

Cable

Cable has been around for a long time and are available in most metro areas. Cable systems were originally designed to deliver broadcast television signals efficiently to subscriber’s homes. Cables companies have made great strides providing high speed data services and in some areas even are providing phone service.

Current Offerings

One-Way

Typical cable systems operate with a bandwidth of 330 MHz or 450 MHz of capacity. Each standard television channels occupies 6 MHz of RF spectrum. Thus a traditional sable system can carry the equivalent of 60 analog TV channels. Video would be fed into a head end unit. As signals degraded they would be amplified along the route allowing a head-end to feed 500 – 2,000 homes.

Fiber optics entered the cable world and allowed signals to be transmitted longer distances without distortion. Hybrid fiber/coax (HFC) systems expanded the bandwidth to 750 MHz. This allows a HFC system to support 110 analog TV channels.

Two-Way

It did not take long for cable operators to see that their system could be expanded to two-way. On most cable systems downstream video programming signals began at 50 MHz so 5 MHz to 42 MHz. A single downstream 6 MHz television channel may support up to 27 Mbps of downstream data using 64 Quadrature Amplitude Modulation (QAM) technology. If the operator uses 256 QAM the downstream data rate can be expanded to 36 Mbps. Upstream channels may deliver 500 kbps to 10 Mbps from the subscriber using 16 QAM or Quadrature Phase Shift Key (QPSK) depending on the spectrum the operator allocates for the service. This bandwidth is shared by 500 – 2,000 homes served by each head end.

Problems

Sharing

The biggest problems with cable networks are that the available bandwidth is shared by typically 500 – 2,000 users connected to the head-end. Many cable operators have lower the number of users connected to a head-end when they upgraded the systems to newer two-way HFC systems, doing so also greatly reduces the number of amplifiers needed. Even if the number of subscribers per head-end is lowered to 500 users there still is not a lot of bandwidth per user.

Reliability

How many times have you picked up the phone and not found dial tone? My guess is not that many times. However cable systems are not known for the same reliability. This is mostly due to the shard system and the high number of individual components used; amplifiers going bad has been a major problem. Newer HFC networks and smaller subscriber count per head end have lowered the number of amplifiers, but most networks still have a ways to go.

Limited Upstream

As cable operators struggle to offer telephony and data services, the limited amount of upstream data is becoming more of a problem. Telephony for example is a full duplex data stream. There may be plenty of downstream data for the call, but upstream congestion can cause packets to be lost. One way to help with the problem is to chop the voice into smaller packets and limit the number of users on each head end. This does help, but in no way solves the problem.

Individual Content

Cable systems take a video source and multiplex it onto a 6 MHz channel. Subscribers simple tune to the channel to view the source. This is fine if you have 100 channels you want to send to 2,000 homes. This is fine for many cable operators; in fact many subscribers feel that 100 channels is a lot. The cable operator may provide something close to video on demand by having multiple pay per view channels with movies starting at different times. The cable operator does not have the ability to offer 500 channels let alone a unique video source for each subscriber.

Wireless

Ok let me start off by saying that I don’t like wireless technology. I think companies have wasted billions on the technology and many are trying to do things that don’t make sense. Have said that there are a lot of good uses for the technology, I just don’t think you will see wireless replace the local loop.

Current Offerings

Satellite

Sputnik was the first satellite put into space. It was launched by the Soviets on October 4^th, 1957; the US launched its first satellite Explorer 1 on January 29^th, 1958. Since that time, we have made quite a bit of progress with satellite technology. Geostationary (GEO) satellite orbits the earth along the equator 23,300 miles above the earth and appear to be stationary in space. Such satellites broadband uses are primarily used to provide one-way video and data services. Two-way communications is not as popular because of the signal delay. It takes about 120 ms for the signal to reach the satellite and another 120 ms for the signal to be bounced back to earth. Three GEOs are needed to cover the entire planet.

Middle Earth Orbit (MEO) orbit between 1,000 and 22,300 miles above the earth and mainly have been used for GPS type services, but there are a few new data services being developed. It takes 10 to 12 MEOs to cover the entire planet.

Low Earth Orbit (LEO) satellites normally orbit between 100 and 1,000 miles above the earth. They are relatively new, but many companies are scrambling to provide data and voice services using LEOs. You can cover the earth with 46 – 66 LEOs.

Cellular/PCS

Cellular has primarily been used for voice services. Many networks expanded to offer limited data services such as text messaging, recently low speed data services have also been tested. The down side of such services is spotty service and limited data rates.

Microwave

Microwave has been used been used for many years, many people have heard of the phone company MCI, but how many know that it stands for Microwave Communications, Inc. :-) In 1963 the company was seeking permission to build a microwave transmission network for use by truckers traveling between Chicago and St. Louis. It was the first challenge to AT&T government-sanctioned phone monopoly. Few new microwave networks have been built in recent years in the US. The technology is point-to-point and not cost effective with new fiber based networks.

LMDS

Local Multipoint Distribution Services (LMDS) technology operates like a cellular network, transmitting signals at around 1 to 1.5 GHz bandwidth. The technology is very appealing because of the high data rates of 27 – 30 GHz of spectrum. Using QPSK allows a typical spectrum to carry around 1.5 Gbps.

Problems

Line of sight

The biggest problem with wireless technology is that in most cases you need to be line of site for it to work. So if you have your cool $1,000 Iridium cell phone and were in a building or even outside next to one you would not be able to use it. Of course line of site is less of a problem in flatter areas, but even in such areas vegetation or water can mess with the signal.

Interference

Interference is also a major problem with wireless technologies, other wireless hardware, heavy machinery, and even the earth itself can interfere with signals. In public 2.4 GHz and 5.7 GHz spectrum operators have even been found jamming others.

Bandwidth

The amount of bandwidth to individual users has been a problem with wireless technologies. Even with spread spectrum technologies you can’t provide a high concentration of users broadband services. LMDS is working to change that, but many doubt they will ever get there.

Interconnection

Wireless technologies generally work over short distances and require may require dozens of hops to complete a link. Each hop adds complexity to the network and increases delay. With MEO and LEO system this problem becomes very apparent, as the satellite spin around the earth the signal must be handed off every few minutes.

Long-term solution

We have looked at the strengths and weaknesses of many different broadband technologies. I am sure it is possible to find experts who think their technology will one day provide broadband voice, data, and video services to residential customers. I would like to take some time to share what I believe will be a long-term solution.

Copper has been here a long time and I think it will continue to be uses to provide residential services. I believe copper offers the most attractive solution available today to provide broadband services as long as it is deployed correctly.

Many CLECs are scrambling around to provide broadband services over copper. The current copper plant was just not designed for it. Sure it is possible with DSP and coding technology to squeeze a lot of data down a normal copper loop. With line conditioning we can squeeze even more. If we are agree that customers require voice, high-speed data, and unique video content to the home we need more bandwidth then the current technologies will provide.

CLECs need to be building new copper plant to the home. Loops length should be kept short (2,000 feet or less) and minimum data bandwidth to every home should be at least 4 meg with the ability to expand to ~50 meg. Loops would be fed from distribution nodes that feed 100 – 500 homes depending on home density. The distribution nodes would be connected back to the main central office with a radius services area of 2 – 10 miles.

Local telephone traffic would be exchanged with the ILEC at each central office. Cache services also would be implemented in each office to help manage the Internet traffic. The remaining voice, data, and video traffic would be transported to the main distribution centers located in each region.

By using a combination of copper and fiber the CLEC is able to meet all lifeline (E911) requirements and still have the ability to provide voice, date, and video. The network is actually very cost effective to build compared to what the ILECs is charging for copper loops that do not meet the requirements. Most ILECs for example are charging $150 – $300 a year for a copper loop. If you compare this to the cost of ~$500 to build out a plant capable of support high speed services the decision is obvious.

Interim Solutions

If a CLEC decides to overbuild in a residential network time to market can be a critical factor. It is not possible to build a network that passes 30,000 homes in 90 days. One interim solution is to interconnect with the ILECs copper plant and also start building a new cable plant. As the CLECs plan grows they can start offering advanced services over that plant.

Conclusion

We have taken a quick look at broadband technologies strengths and weaknesses. Technology has allows us to greatly expand the users of the existing copper plant, but if we are to provide broadband access to entire communities new networks will need to be constructed.