Fourteen Broadband Forum companies that provide VDSL2 G.vector equipment came together June 17-21 for a plugfest at the University of New Hampshire InterOperability Lab (UNH-IOL).
Plugfests are gatherings where communications equipment makers get a chance to see how well their chips and systems communicate with those from other companies. This plugfest gave engineers a chance to test their products at speeds of up to 100Mbit/s over twisted-pair copper wires.
The G.vector standard, officially known as ITU-T G.993.5, is a recommendation for canceling far-end crosstalk (FEXT) in twisted-pair copper wires that carry DSL signals. At high data rates (VDSL2 can operate at 100Mibt/s), the frequencies contained in the signals can couple into adjacent twisted pairs, increasing the noise that reduces the signal-to-noise ratio. A lower ratio forces the DSL access multiplexer (DSLAM) and customer premises equipment (CPE) to rate-adapt to a lower data rate. EDN, a sister site, discusses the G.Vector standard in more detail.
What kind of testing took place at the plugfest? Surely, engineers don't just show up and randomly connect their products. Lincoln Lavoie, a senior engineer at the UNH-IOL, told me that these sessions run over an entire week. The first four days are devoted to giving participants a chance to test one another's products. Each DSLAM gets tested with each CPE. Thus, participants have scheduled times with one another.
Each test setup consists of a spool of cable made of four twisted pairs, with one pair used to connect the equipment under test. The remaining pairs are connected to a Telebyte line simulator that generates the noise caused by FEXT.
A Telebyte VxT-48 xTalk Emulator for vectored VDSL2 testing.
"As the G.Vector implementations have progressed, we've developed new test cases," Lavoie said. The UNH-IOL now has about a dozen for G.Vector. A basic test case consists of simply being able to get a DSLAM and CPE to synchronize and establish a link. The first test is to see the speed with no FEXT and then add noise to one line at a time and check connection speed. A DSLAM with G.Vector can measure the FEXT on all the lines and calculate how to compensate for noise in each line. If you're familiar with the pre-emphasis used on high-speed serial data streams, that's essentially what the DSLAM does -- it adds noise that cancels out the FEXT at the other end.
On the final day, with all test pairings complete, the participants are invited to connect all their equipment to a 100-pair bundle -- DSLAMs on one end, CPE on the other. This gives everyone a chance to see how their products might function once deployed. The unused twisted pairs are not connected.
After the plugfest, UNH-IOL engineers compile the test results and write a "state of the industry" report, Lavoie said. Individual test results aren't included, but the report, which goes to all participants, gives participants and the Broadband Forum an opportunity to see how the technology has progressed. Of course, each participant gets its own test results.
Have you ever participated in a plugfest? I've attended a few at the UNH-IOL. Tell us about your experiences.
Additional VDSL2 G.Vector resources: