The Promise of 100 Mbps and Vectored Services Raises Bar on Network Quality
The next generation of VDSL2 technology call Bonding & Vectoring promises performances upwards of 100Mbs on your existing copper network (FTTN). This allows you to economically deliver significant performance boost to your customers without costly fiber investment. But can your copper plant handle the demands of bonding & vectoring ?
As a provider of VDSL2 based residential services (IPTV, High Speed Internet, etc.) you have already made a significant investment in your copper network. But will it be fast enough to support customer demands over the next several years? Given that cable companies are already offering network speeds upwards of 100 Mbps, what are the challenges facing your network to provide a similar performance and beyond? How will you meet those challenges?
The good news is that you don’t have to make that huge investment in a fiber-to-the home network. Once again, next generation copper based technologies are coming to the rescue. By using a combination of new VDSL2 based copper bonding and vectoring technologies, you will be able to deliver “fiber-like” speeds (upwards of 100 Mbps) that rival those of the competition.
These are serious speed increases that will require a network free of impairments to reliably support these ultra-fast speeds. Now, more than ever is the time to truly condition your network to deliver this performance. You can’t ignore impairments or marginal network conditions. Simply put, you can’t “fake” fast.
What is the best way to prepare the copper network to deliver on the promise of 100Mbs services to customers? Read on to get answers to these questions and more as we take a deep dive into what it takes to get the copper network ready for 100 Mbps speeds.
VDSL Bonding Technology VDSL bonding technologies address the need for higher speeds by dealing with the DSL signal itself — two or more DSL signals are better than one. Pair bonding is the process of combining the capacity of one or more pairs in such a way that the “bonded” pairs act like one BIG capacity pair (see Figure 1 — VDSL2 2 Pair Bonding).
The data carrying ability of each pair remains the same, but bonding two equally capable pairs together effectively doubles the total data carrying capacity. It is kind of like a super highway — more lanes mean a BIGGER capacity to carry traffic.
VDSL Vectoring Technology DSL vectoring technologies address the need for higher speeds by dealing with speed killing noise. DSL receivers deliver high speeds when noise levels are low. Therefore, it makes sense to mitigate noise at the receiver to keep speeds high. The total noise present at a DSL receiver is best thought of as a group of individual noise components that add together to create a “total” noise problem. Among those noise components that contribute significantly to “total” noise would be Near End Crosstalk noise (NEXT) and Far End Crosstalk (FEXT) noise. DSL Vectoring is a complex microprocessor intensive method of mitigating a specific kind of noise — FEXT noise . Today, several DSLAM vendors are providing VDSL2 vectoring solutions.
NEXT Noise In the early days of DSL planning, it became clear that the best way to mitigate NEXT noise was to keep the Downstream and Upstream transmission bands separate from one another. Such an approach would not add significantly to the cost of silicon.
FEXT Noise Dealing with FEXT noise is more challenging. In those early days of DSL planning the computing power required to cancel FEXT noise at the silicon level would not have been cost effective. However, today vendors are shipping DSLAMs with cost effective VDSL2 vectoring FNEXT noise mitigation capabilities.
Speed Possibilities DSL speeds are measured in bits-per-second (bps). The 100 Mbps speed objective bandied about today came from the ITU-T G.993.2 VDSL2 standard which defines a VDSL2 17a profile with a “minimum bi-directional net data rate capability” of 100 Mbps (the Shannon-Hartley theoretical limit of a 17 MHz bandwidth is closer to 246 Mbps). The standard also defines a VDSL2 30a profile with a minimum capability of 217 Mbps (the Shannon-Hartley theoretical limit of a 30 MHz bandwidth is closer to 417 Mbps). When pair bonding is used in conjunction with DSL technology, there is a significant jump in total speed. Two bonded pairs as used in VDSL2 deployments also introduce the possibility of a third “phantom” channel to further increase speed — 300 Mbps or higher may be possible.
Bandwidth Anytime there is an increase in bandwidth, there will be a need to re-visit existing copper network qualification methods. Bandwidth is the available spectrum for transmission. POTS service only uses 4 KHz of bandwidth. Qualifying copper pairs for POTS is relatively easy. But then along came analog modems and fax machines that used frequencies within that 4 KHz bandwidth that had not been qualified. Do you remember all those new trouble tickets? Now imagine what happens when you upgrade from an 8 MHz bandwidth network (VDSL2 Profile 8d) to a 17 MHz network (VDSL2 Profile 17a). Sure, you have qualified the network for 8 MHz, but what about that part of the bandwidth from 8 to 17 MHz that wasn’t qualified? How do you know it will be free of frequency dependent trouble?
The 17Mhz and 30Mhz are the new VDSl2 profiles being adopted to support bonding & vectoring to achieve 100Mbs speeds. The 17Mhz profile is expected to be the most widely adopted profile for bonding & vectoring due to an attractive combination of rate (supports up to 100Mbs with bonding& vectoring) and reach (loop length up to 2kft or 600m).
Preparing the Copper Cable for 100 Mbps or Higher Speed Services Now that you have an understanding of the technologies, standards and speed possibilities, let’s focus on preparing the copper network for 100 Mbps or even higher speed services. The addition of bonding and vectoring technologies to existing VDSL2 deployments will require accurate records and additional qualification tests.
Records Accuracy Records verification is a process unto itself. Telephone companies maintain database records (cable, pair, binding post, status, etc.) about copper pairs. These records are changing to keep pace with advancing DSL technologies. They need to be periodically “refreshed” to reflect the current state of pairs in the field. When a customer calls in to request high speed DSL service it is important that customer agents quickly respond with accurate information as to whether or not that service can be provided. Here are some of the new challenges in a VDSL2 bonded and vectored network related to records accuracy: There will be an increase of bonded pairs in the new network that will require a better understanding of total pair capacity. How many pairs do you have to work with? Pair selection for bonding will require a better understanding of the utilization of available pairs. How many pairs are going to each customer? How long are the pairs? Are there spare pairs in the serving terminal? VDSL2 vectoring can be done at the board or system level. Since vectored noise mitigation only works for common VDSL2 based services, how will you manage these vector groups? The DSLAM vectoring engine can only mitigate so much FNEXT noise. How will you manage its resources?
Pair Qualification Testing A pair that is free of trouble will deliver the highest speeds. Here are some of the new qualification testing challenges for a VDSL2 bonded and vectored network that makes use of bandwidths out to 17 MHz or 30 MHz: The total speed of bonded pairs will be limited by pair quality. Pairs that are free of trouble and with the best longitudinal balance will deliver the highest speeds. Noise is a speed killer. Is there noise in the new bandwidth? Noise on one pair can crosstalk onto adjacent pairs. Is the total crosstalk noise at a DSLAM limiting the speeds for all? Will the high level of FNEXT noise overwhelm the DSLAM vectoring engine that is attempting to mitigate this noise?
Signal attenuation is a speed killer. Is there frequency dependent trouble in the new bandwidth that will attenuate the signal? Bridged-Tap can reduce speeds. The new network will have to deal with the effects of even shorter bridge-tap lengths. Clearly it is still important to perform the same core physical layer testing required for VDSL2 qualification — tests that identify foreign voltage, insulation resistance breakdown, opens (capacitance), noise and bridged-tap. But the new network will need some additional tests that identify any frequency dependent trouble in the new untested bandwidth. The following describes the comprehensive set of qualification tests we recommend for VDSL2 bonding & vectoring to support 100Mbs service.
First, the following core tests help find some common line faults :
1. Foreign AC and DC Voltage The only DC voltages allowed on a copper pair are those that have been designed to be there. Unfortunately, DC voltages from other pairs may appear if the cable has been compromised — water, broken insulation, etc. The only AC voltages allowed on a copper pair are those that have been designed to be there. Unfortunately, AC voltages couple onto copper pairs from all those power lines that run parallel with telephone cables.
2. Insulation Resistance There should be high insulation resistance between copper pairs (Tip-to-Ring, Tip-to-Ground, Ring-to-Ground). Unfortunately, over time cables begin to deteriorate resulting in a lower insulation resistance.
3. Capacitance Copper cable is designed to have the same mutual capacitance between the wires that make up a pair (Tip and Ring). However, it is possible for this capacitance to change over time due to bad splices, water, bridged-tap, etc. Also, the Tip-to-Ground capacitance should be very close to the Ring-to-Ground capacitance.
4. Bridge Tap Detection The International Engineering Consortium describes a bridge tap as “any unterminated portion of a loop not in the direct talking path between the CO or DLC and the customer’s premise equipment (telephone or modem).” When a signal is reflected from the unterminated portion, the reflection can interfere with the original signal. Bridge taps, particularly short taps, are much more of a problem in VDSL than in ADSL due to the higher frequency of VDSL. Bridge taps cause what is referred to as a “notch” effect, which can adversely affect the carriers at these frequencies. In general, the length of bridge tap will determine the severity of the “notch”. Finding and removing bridge taps is more critical than ever before when it comes to providing reliable 100Mbs performance at 17Mhz and 30Mhz profiles.
But as we now know, bonding and vectoring performance requires additional tests that you may not have been performing. These additional tests are important especially out to 17Mhz and 30Mhz to find faults that could impact performance at these higher frequencies.
5. Bandwidth Noise (out to 30 MHz) Noise is a speed killer. Noise should be analyzed across the entire bandwidth being used. The noise on a tested pair must be broken down into its component parts such that proper mitigation techniques can be applied. VDSL2 Vectoring technology deals specifically with the cancelation of FEXT component noise — it will not be able to deal with RF or impulse noise. Along with loss, this test provides key information related to bit rate estimations. The noise level at a receiver will limit transmission.
6. Wideband Longitudinal Balance (out to 30 MHz) When you hear the term “longitudinal”, think Tip-to-Ground or Ring-to-Ground. When you hear the term “balance”, think equality between Tip-to-Ground and Ring-to-Ground. When you hear the term “impedance”, think frequency dependent resistance. A pair is longitudinally balanced IF the Tip-to-Ground impedance and Ring-to-Ground impedance is equal for each frequency in the transmission bandwidth. Why is it important for pairs to have good longitudinal balance? Today’s telephony transmission methods use differential mode transmitters to send signals across twisted pairs to differential mode receivers capable of rejecting common mode noise. Since most noise is common mode, the receiver can effectively reject it. However, when a pair is “unbalanced” at a particular frequency, any noise at that frequency will appear as differential mode noise to the receiver, and the receiver will not be able to reject it.
Since reducing noise is a critical aspect for vectored services, it is important that the line is longitudinally balance. In addition, since VDSL2 utilizes distinct sub-carriers are 4.3125Khz spacing (or “bins”) it is important that the longitudinal balance test be performed at each of this frequency across the entire 17Mhz or 30Mhz operating spectrum.
Eliminating a longitudinal balance problem has a double benefit — a properly balanced pair will generate less disturber noise and will have greater immunity to disturber noise generated by adjacent pairs. Of course the ability to perform this Wideband Longitudinal test single-end (without requiring a far end device) would be best to reduce test and troubleshooting time.
7. Bandwidth Loss Measurement (out to 30 MHz) Having the ability to measure the loss of each frequency within a useable bandwidth (the attenuation curve) reveals important information about the overall transmission ability across a pair. Utilizing a single-ended technique to make this measurement eliminates the requirement to have a test set on the opposite end of the pair. Along with noise, this test provides key information related to bit rate estimations. The attenuation curve accurately represents the signal level at the receiver along with any loss due to physical faults (bridged-tap, bad splices, etc.).
8. Bandwidth Bit Rate Estimations Accurate DownStream and UpStream bit rate estimations provide key insight into the ability of a pair to deliver on the 100 Mbps performance target. A properly designed bit rate estimation algorithm knows how to utilize all sources of information (physical tests AND database inquiries) pertaining to the tested pair to provide an accurate estimate of expected DownStream / UpStream rates. Physical tests reveal important information related to attenuation and noise on the pair. Database inquiries reveal important information about the effects of other pairs in the binder on the tested pair. Bonding and vectoring introduce new complexities that must be accounted for by a good rate estimate algorithm: Impact of noise and attenuation across two pairs Changeable bandwidths (e.g. 17a versus 30a) Noise contributions of adjacent pairs Expected improvements when vectoring is turned on Expected improvements when bonded pairs are used instead of a single pair Lost performance due to bridged-tap Lost performance due to foreign noise Performance impact on neighboring pairs When done correctly, this estimation can serve as the sole pass or fail indicator of a tested pair.
In conclusion, you can see that your network’s ability to support the latest VDSL2 bonding & vectoring technology to deliver up to 100Mbs requires that it be free of impairments. The need to identify problems such as 17Mhz noise and longitudinal balance are now more important than ever before. You are pushing your network more than ever, delivering more bandwidth than ever before. This requires the highest quality network cable testing, you cannot mask these impairments if you want to achieve upwards of 100Mbs.