Ask businesses if they would like more bandwidth for their network, and if they would rather get it sooner than later, and close to 90% of them would likely answer an enthusiastic "yes" to both questions.
That's because only 10% of U.S. business locations are now served with optical-fiber cable, and most of those are in concentrated business districts of larger cities. The vast majority of enterprises get their bandwidth the old-fashioned way—via copper.
No matter what the current state of the economy, businesses still want to connect faster, move their data more efficiently, and not have to take out loans just to pay the monthly bill from their telecommunications provider. The good news is that the providers are aware of this demand, and are keeping a close eye on technology hitting the market this year that will help them address the needs of their customers.
The technology involves aggregating the bandwidth of several copper pairs in the local loop. It opens the door to services that let enterprises connect to the public network at speeds far greater than the 1.5 Mbits/sec limit of their present T-1 or DSL service. These services could range from simple provision of a "bigger pipe," with bandwidth at 4 or 5 Mbits/sec, up to an Ethernet service that lets enterprises use native-rate 10 Mbits/sec Ethernet networking over the wide area.
Ethernet services such as these are expected to be especially attractive to remote or satellite offices of larger companies. These facilities are often "islands" of inefficient networking because they are in an area where fiber is unavailable. Until now, without the fiber, they couldn't get the necessary bandwidth to fully participate in the corporate network.
In most cases, enterprises will see the fruits of this multi-loop copper technology in the form of services delivered from their telecom provider. But some organizations (such as universities or corporate campuses where an organization runs its own telephony and data service operation) will be in a position to use the technology themselves.
Inverse multiplexing
Architecturally, this technology bundles multiple pairs of copper and aggregates their bandwidth through inverse multiplexing to move well beyond a single pair's capacity limitations. The preferred underpinning of this technology is the G.shdsl standard, because of its superior spectrum compatibility and high rates for symmetric transmission over copper. Solutions are achieved through a line card at the central office (or its equivalent), with a CPE (customer-provided equipment) device at the customer location for aggregation and transmission.
The three main approaches are Inverse Multiplexing over ATM, or IMA; Multi-Link Point-to-Point Protocol (MLPPP) and Multi-Link Frame Relay (MLFR); and Multi-Loop DSL (MLDSL). IMA and the multi-link approaches are typically applied to DS-1 lines. MLFR is used to provide higher bandwidth, fractional DS-3 frame relay services. MLPPP is often applied to DS-1s and ISDN B channels to provide higher-speed Internet access and transparent LAN services. MLDSL, on the other hand, is applied directly to copper loops .
High speed but with drawbacks
IMA is used primarily to bond DS-1s together to provide ATM service to subscribers not served by fiber. The DS-1s can be provided using HDSL over copper.
IMA can be used to deliver high-speed Internet access, but with a few drawbacks. First, IMA is based on ATM equipment, which is typically more expensive than Ethernet-based gear. Second, the IMA protocol and encapsulation for IP data is inefficient. As much as 20% of the available throughput is taken up in the form of ATM overhead. Third, the ATM IMA protocol requires all ports to operate at the same data rate. This is not a problem when bonding DS-1s together, but if IMA was applied directly at the modem, the lowest common data rate would be used.
For instance, if one pair out of six is degraded due to a bridge tap or other limitation, then all six pairs will transmit only at the maximum data rate achievable by the degraded pair. Specificlaly, instead of 9 Mbits/sec (6 pairs × 1.5 Mbits/sec), if all the pairs are operating at 1.5 Mbits/sec, the degraded pair could drop the total bandwidth to 4.5 Mbits/sec (6 pairs × 768 kbits/sec).
Protocols minus ATM
Multi-link protocols, including MLPPP and MLFR, offer an approach that does not involve ATM. Both multi-link protocols group together several DS-1s, which may be transported over copper with HDSL or any other transport technology. MLPPP is typically deployed in the ISDN market to bond B channels to provide 128K data transport, and in the Internet access market to bond DS-1s for higher-speed Internet access or transparent LAN services.
MLFR bonds DS-1s to provide fractional DS-3 frame relay service. Both have low transmission overhead, since ATM is not incorporated. These protocols use relatively large packet sizes, which introduce latency into DS-0, or DS-1 TDM transport applications. In addition, these methods enable automatic link recovery and bandwidth redundancy.
Loop failure protection
Multi-Loop DSL is the only one of the three technologies that doesn't base its per-pair transmission rate on DS-1. It has the ability to transmit at 2 Mbits/sec and above. MLDSL's inverse multiplexing technique (operating between Layer 1 and 2) reduces latency and overhead, providing a high throughput. It can bond copper pairs at different line rates, with the ability to transmit data on each pair according to its maximum capability. So, it isn't severely limited if one pair is degraded.
In addition, MLDSL can operate even if a pair in the bundle goes out of operation, and can, in many cases, maintain the total required bandwidth even without the defective pair. Conversely, if an additional copper pair is added to the bundle, it is immediately utilized to increase bandwidth. Using this technology, Multi-Loop DSL has the capability to offer loop failure protection.
Another advantage is that MLDSL does not rely solely on spare copper pairs. In instances where there may not be enough available spare pairs, Multi-Loop DSL can also share the frequency of a POTS line to use it for its own services, accomplished with a passive splitter.
Comparing the three technologies in the way they accomplish their multiplexing, IMA and MLPPP/MLFR are typically implemented at the edge switch, being fed by DS-1s. This results in the restriction of each port to a DS-1 rate. MLDSL does its multiplexing at the copper interface, which allows a wider range of data rates.
Using Multi-Loop DSL as an example, an enterprise could establish access to 12 to 18 Mbits/sec of bandwidth (depending on distance and line quality) by employing eight spare or POTS-utilized copper pairs. The solution would let either the telecom provider or the enterprise itself achieve 1.5 to 2 Mbits/sec on ‰ each of the pairs. Through inverse multiplexing, the bandwidths of each pair are aggregated into a single large pipe.
Customer premise equipment
Whether an enterprise is purchasing a multi-loop copper service from a provider or installing the equipment itself will determine what type of CPE will be found on the customer site. Presuming it is a provider-based service, the equipment will be minimal.
If an enterprise is running its own service, it will also have the central office equipment, a 19-inch shelf supporting 18 plug-in cards, and an integral SNMP management card. Each card supports up to eight copper pairs and has an 8-pin modular Ethernet port and a T-1/E-1 port for voice/data support. It would interface with the Ethernet ports on the telecom service switch, is -48 V DC-powered, and has a local craft interface for Web-based configuration and maintenance.
Unlike previous copper line technologies, the DSL system does not need manual adjustment. The DSL modem automatically analyzes the line and adapts to start up the link within seconds. This process continues once the link is started, as the modem compensates for ongoing changes (due to temperature, for instance). The modems contain advanced digital signal processing (DSP) algorithms that produce mathematical models of the distortions caused by the line, and produce automatic corrections.
There is the persistent performance trade-off of faster data and shorter distance, but Multi-Loop DSL carries higher data rates much further than traditional DSL. Other factors that affect range are the cable gauge and the amount of interfering noise.
In most cases, service providers can find eight spare pairs in a wiring bundle that go to a company's premises. But even in situations where there are not enough spare pairs available, the technology can still be applied. With a MLDSL implementation, you can use passive splitters, which can use a POTS line for simultaneous data and voice transmission.
The splitter is a three-node device that carries the telephony and DSL signals on the same copper pair without interference. The POTS signal is in the low-frequency band, while DSL signals operate at higher frequencies—25 kHz or above.
The splitter provides a low-pass filter between the copper line node and the telephone node, and a high-pass filter between the copper line node and the DSL device. The splitter also prevents the DSL signal from interfering with the analog phone, and prevents the analog phone service from interfering with the DSL signal.
Inexpensive equipment
The good news about multi-pair copper equipment coming onto the market is that it is fairly inexpensive. Its price point lets a provider deliver service to a customer for a fraction of the cost of deploying fiber. Even at an aggressive, customer-friendly price, a provider can see a return on investment in a matter of a few months.
With the current economic environment in the telecommunications industry, service providers are clearly in no hurry to invest in new installations of fiber in the access network. Such a sizable capital expenditure takes money they don't have and takes a long time to deliver a return on investment. But as demand for more bandwidth continues to bottle up from a large segment of business customers in the access network, service providers know they need to respond.
No one expects a multi-pair copper technology to displace fiber; it is instead intended as a method of bridging the gap between the bandwidth of a single pair and the nearly limitless bandwidth of fiber. With industry analysts calculating that it will be another 10 years before fiber is anywhere near ubiquitous, the multi-pair copper approach can fill a need for many years to come.
Randy Nash is vice president of business development for Spediant Systems, Red Bank, NJ. He has worked on access network technologies through most of his 20-year telecom career, and holds 13 U.S. patents. He can be contacted at [email protected]