Authored by the Category 6 Consortium
The demand for higher data rates and faster applications over cost-effective balanced cabling continues. Many members of IEEE recognized this potential and are thinking of ways to deliver 10 Gbits/sec over balanced cabling. At the IEEE 802 Plenary Meeting held November 10 through 15, 2002 in Hawaii, a tutorial was organized on "10GBase-T Challenges and Solutions" that considered a 10-Gbit/sec Ethernet standard for twisted-pair copper cabling. The follow-up "Call for Interest" meeting included many interesting presentations on technical feasibility, broad market potential, and economic feasibility. At the end of the "Call for Interest" meeting, a vote to establish a study group was unanimous with 110 participants expressing support. Later in the week, the IEEE 802.3 working group and the IEEE Standards Executive Committee (SEC) unanimously approved the formation of an IEEE 10GBase-T study group. This study group will evaluate the feasibility of standardizing 10-Gbit/sec data rate over horizontal balanced cabling up to distances of 100 meters (i.e., connecting work areas to a telecommunications room) specified in TIA 568-B and ISO/IEC 11801 building cabling standards.
Timeline for the project
The 10GBase-T study group is on an aggressive schedule to develop the Project Authorization Request (PAR) and five criteria necessary to get approval from the parent IEEE 802.3 working group, IEEE 802 plenary and the IEEE SEC. The study group has spawned several ad-hoc committees to study modeling issues, cabling issues, and coding issues leading to much discussion over the e-mail reflectors. Additionally, the study group has had two face-to-face meetings-in January 2003 in Vancouver, and in March 2003 in Dallas. The projected timeline for the project is as follows: Call For Interest was in November, 2002; Project Request Authorization in October 2003; Draft 1 in June 2004; Task Group Review in July 2004; Working Group Ballot in March 2005; Draft 2 in April 2005; LMSC Ballot in October 2005; Draft 3 in November 2005; and Final Standard in June 2006.
Cabling choices
There is much discussion and debate in the IEEE 10GBase-T study group regarding the Category of cabling that the application should use. This debate covers both technical feasibility and economic viability. Part of the economic viability argument is to design the application to run over the current installed base to reduce the cost of cabling.
There are others who believe that the choice should be made on the installed base when the application will be deployed in 2006. It would seem appropriate that the PAR should be written specifying Category 6/Class E cabling based upon the rapid growth of Category 6 cabling in the marketplace, and its significantly improved performance. Others argue that Category 7/Class F cabling should be used, despite the fact that very little is installed, or projected to be installed, over the next three years.
The cabling industry is supporting Category 6/Class E cabling or better as the optimal choice to develop the IEEE 10 Gigabit Ethernet standard. The position of cabling vendors became very clear at the recent (February 2003 in Miami) TIA TR-42.7 meeting, when 29 companies agreed to do additional work to support 10GBase-T over Category 6 cabling while no company indicated a desire to do additional work to support 10GBase-T over Category 5e cabling. The committee responsible for ISO/IEC 11801:2002 also reached a similar consensus position, and wrote an official liaison letter to IEEE 802.3 in February 2003, following their meeting in Wellington, NZ. The letter stated that they would be glad to support the development of the 10GBase-T application over Category 6/Class E or better cabling, but there was no interest in further characterizing Category 5e/Class D cabling. It will be up to the IEEE 10GBase-T Study Group to take into account these cabling industry positions, in addition to the digital signal processing (DSP) issues, when choosing the media to support this developing application. The cabling decision/options will be included in the Study Group report when it formally initiates a PAR after the study is competed in July 2003 (projected).
What is the cabling industry doing?
The cabling industry standards are well positioned to support the IEEE 10GBase-T effort, having just published TIA 568-B.2-1 and ISO/IEC 11801:2002. ISO/IEC and TIA cabling committees are expected to support IEEE requests for information with regard to measurement methods and test data that indicate the performance of installed balanced cabling. As indicated earlier, both cabling committees have no interest in doing further work on Category 5e/Class D cabling, which has a characterized bandwidth currently limited to 100 MHz. Measurements of installed Category 6/Class E channels up to 500 MHz have been provided to the Study Group in support of PAM-9 encoding schemes, and up to 650 MHz in support of PAM-5 encoding schemes. This cabling test data, together with the environmental data and FCC requirements for EMC, will be used to determine the capacity of the system to establish feasibility of 10GBase-T over Category 6/Class E or better cabling.
What about Category 7/Class F cabling?
The study group has discussed the Category 7/Class F cabling option, but realizes that it is only 0.4% of the worldwide installed base, making it unattractive from the broad economic potential criterion. However, Category 7/Class F will be covered by default if the study group chooses Category 6/Class E or better cabling. There is also the question of screening/shielding. More recent IEEE 802.3 applications for "copper" have been designed to work over balanced cabling, irrespective of whether it is screened or unscreened. The scope of IEEE 802.3 applications is that it defines the cabling channel performance-not the cabling construction or topology.
What frequency bandwidth will 10GBase-T use?
There are various signaling and encoding schemes that are being discussed in the IEEE Study Group. On group has been promoting the PAM-9 encoding scheme, which will transmit signals with most of the signal power below 417 MHz (Nyquist frequency is 417 MHz). To support this scheme, measurements out to 500 MHz have been used to develop cabling models and system simulation models. Another group within the Study Group wants to use the tried-and-tested PAM-5 coding scheme that was used in the IEEE 1000Base-T application. To achieve 10 Gbits/sec, a PAM-5 scheme will have to transmit signals up to 625 MHz. Measurements up to 650 MHz are being provided to IEEE for characterizing Category 6/Class E cabling. It should be noted that these measurements out to higher frequencies use existing Category 6 components built into a full horizontal channel, consisting of 100 meters of Category 6 cable and four connections. The crucial point is that existing Category 6/Class E cabling is being characterized up to higher frequencies so that digital signal processing (DSP) simulations can be run to determine the capacity of a 100-meter Category 6/Class E channel with extensive DSP.
What are the current limitations?
Simulations of 100 meters of balanced cabling with various degrees of DSP indicate that the objective of running 10GBase-T over Category 6/Class E cabling can be achieved with some effort by the silicon vendors and the cabling vendors. Using DSP, the silicon vendors can cancel, or compensate for, most of the internal cabling impairments including return loss, NEXT, ELFEXT, and insertion loss. The one area that they cannot cancel out effectively by DSP is alien NEXT and alien ELFEXT, which is the crosstalk coupling into a cabling pair from surrounding cabling and other external noises. The cabling industry is studying this issue and developing proposals for methods to reduce alien NEXT and alien ELFEXT.
Summary
The objective of IEEE 10GBase-T is to create an application that is capable of transporting data at 10 Gbits/sec over 100 meters of horizontal balanced twisted-pair cabling. State-of-the-art DSP will be used to cancel impairments in the cabling system and noise from the external environment to ensure adequate signal-to-noise ratio (SNR) to achieve a proposed Bit Error Rate Requirement of 10(-12) at data rates of 10 Gbits/sec. The cabling industry, represented in TIA and ISO/IEC, is promoting the use of Category 6/Class E cabling or better to support this demanding application. It is expected that this Category 6/Class E cabling will be characterized up to some higher frequency. That frequency will be determined by consensus within the IEEE 802.3 10GBase-T Study Group. This additional characterization is similar to the original characterization of Category 6/Class E to 250 MHz, when the original design of Category 6/Class E cabling based up 0-dB PSACR (a measurement of SNR) is limited to 200 MHz. This characterization of current Category 6/Class E cabling up to some higher frequency should not require new components, but simply characterization of existing Category 6/Class E components and channels to higher frequencies.
Developing 10 Gbits/sec over balanced cabling is going to be very challenging. To optimize costs, while ensuring robust and reliable performance over 100 meters and through four connections, Category 6/Class E cabling is the most likely choice to support the emerging IEEE 10GBase-T application.