Talk:G.992.1

Does anyone know where the attenuation graphs came from or how they were calculated? They are not sourced. VDSL even seems a little high since most VDSL2 implementations I read about max out around 50 Mbps in a pair bonded configuration. Maybe the raw data and generating functions can be placed in the article body? (They are beautiful graphs ) Swintgs (talk) 19:59, 4 November 2008 (UTC)

Echo cancellation section:

Your calculation of the rates is not correct - dividing by 2 is not necessary (why would you do that anyway?). The theoretical max rate of the downstream direction is about 13 Mbps (224 * 15 * 4kHz), not 7. --69.111.160.185

BTW - the standard is called ITU-T G.992.1, T for telecommunications, IIRC.

When I initially wrote that section, I took into account various sources and knowledge from my university comms course.
 * From various graphs and other information, ADSL runs up to a maximum of 8Mb. It is ADSL2 which extends this to 12Mb and ADSL2+ which extends this again to 24Mb.  This suggests that the calculation of rates is correct.
 * I have modified the echo cancellation section with a link to Amplitude Modulation (AM) which hopefully explains the division by 2. Since G.DMT uses a form of AM called Quadrature Amplitude Modulation, it would suggest that the same properties of AM hold.  Also, in your above statement, it should be 256*15*4.3125 since on a good quality line, ALL 256 bins are used for downstream, with 32 being used for upstream (with echo cancellation be used on these 32 so that these are jointly used for downstream and upstream data simultaneously).  If any of this is misinformation, I am sure someone more knowledgeable than myself would have corrected that section by now (such is the reason for WikiPedia being in existence).

Hope that explains things! Larkymarky 10:12, 26 March 2006 (UTC)

No it is as User:69.111.160.185 writes above (maybe 222, if we start at carrier 33 and remove the pilot). If nothing else, should your 12 Mb/s for ADSL2 (which uses the same frequencies) indicate that your 8 Mb/s is incorrect. The data frame rate of 4kHz comes from the fact that one includes the overhead of the cyclic extension and the synchronization frame. Also note that carrier 0-6 is typically not used, even for upstream. The difference from 13.3 Mb/s down to 8 Mb/s can either be seen to come from framing overhead, FEC, and control channel overhead. Or, if you look up some early books on ADSL (or the ANSI standard), from the structure of bearer channels. In ANSI T1.413 issue 1 the maximum downstream bit-rate was when all four bearer channels where used and each bearer channel could bear 1.536 Mb/s, that is,the maximum was 6.144 Mb/s! Later on, some clever way of circumvent this was implemented, and the limit ended up being 8Mb/s instead. For ADSL2 these restrictions and the overhead for framing has been reduced and ADSL2 comes (theoretically) much closer to the 13.3 Mb/s than ADSL1. 128.130.90.53 14:28, 1 April 2006 (UTC)

Feel free to make appropriate modifications to that section as you see fit then. What you have described above is outside my scope of knowledge so if you feel this particular section needs a re-write to cover further (correct) technical details/explanations, feel free to update it! Larkymarky 18:06, 7 April 2006 (UTC)

I have deleted some of the stuff that is wrong, and made a few modifications to correct things. It would be better to rewrite the article, or just delete it and redirect to the already existing ADSL article, which would be my vote. --69.233.166.166

conflicting information - dB/bit
In the "BIN quality and bit rate" section, it states that there needs to be 6dB above the noise floor per bit in a bin, but in the summary at the end it says 3 dB.
 * The summary looks dubious to me, so I've added a 'citation needed' tag. Harumphy 13:36, 18 November 2006 (UTC)

It is 3 dB/bit. Or to be precise, around 10dB for the first bit ("SNR gap"), and then 3dB for each additional one. See G.992.1, or any textbook that explains QAM modulation. ===

I've updated this section - it is 3 dB/bit but the standard uses a minimum of 2 bits per BIN - meaning, the lowest SNR would have to be 6 dB. Larkymarky 10:45, 20 March 2007 (UTC)

other conflicts
This documents says the standard expands the used frequency range allowing faster communication. However, the same standard is also mentioned as ADSL Annex B, beind ADSL over ISDN, where the ADSL takes lesser frequency range, because of the ISDN. No document on wikipedia explains the difference between Annex A and Annex B, nor it is mentioned anywhere. However, ISP's and modem producers use this technology and relevant information is not available. If somebody could comment on this, it would be very much welcomed. 87.244.183.56 18:36, 16 December 2006 (UTC)

8 Mbit/s limit
With a rate of 4000 data frames per second and a maximum of 255 bytes (maximum RS codeword size) per data frame, the ADSL downstream line rate is limited to approximately 8 Mbit/s per latency path. The line rate limit can be increased to about 16 Mbit/s for the interleaved path by mapping two RS codewords into one FEC data frame (i.e. by using S = 1/2 in the interleaved path). S = 1/2 shall be used in the downstream direction only over bearer channel AS0. Support of S = 1/2 is optional. —Preceding unsigned comment added by 89.164.130.67 (talk) 20:01, 13 October 2007 (UTC)

Deleted the section on the 8Mbps limit because it is nonsense and pointless anyway. === —Preceding unsigned comment added by 75.18.210.14 (talk) 15:07, 3 November 2007 (UTC)

I'm concerned about the ceiling claim of 8Mbit/s in the intro paragraph. My own ADSL modem here in Japan, an Aterm WD701CV (link contains Japanese text), uses G.dmt Annex C to achieve downstream speeds up to 50 Mbit/s. The article is obsolete at best or downright incorrect at worst. You decide. Trane Francks (talk) 07:06, 16 July 2013 (UTC)

Quick-failed GA nomination
According to the quick-fail criteria of the GA nominations process, any article that has "a complete lack of reliable sources" as required by WP:V, must be failed immediately and does not require an in-depth review. All GA-class articles must fully comply with Wikipedia's verifiability policy. This means that reliable sources must be set aside in a special References section and inline citations to these sources must be made. Please remedy the lack of verification in the article before choosing to renominate. If you feel this review was in error, you may request a reassessment. Thank you for your work so far, Van Tucky 01:53, 10 January 2008 (UTC)

Overlap of Neighboring Bins
It was stated that neighboring bins overlap, and there is no interference due to orthogonality in OFDM. But, if I understand OFDM correctly, each "bin" represents a QAM constellation made of two orthogonal sinusoids ... but they are not (by my calculations) orthogonal to the sinusoids in the neighboring bins. For example, bin 1 (f = 4.3125kHz) has a cos(2πft) component and a sin(2πft) component which are orthogonal. But bin 2 has a cos(4πft) component and a sin(4πft) component, which are not orthogonal to bin 1's components.

Can someone who knows more than me about this verify?

Keithnazworth (talk) 20:24, 7 March 2008 (UTC)

Feel free to add ADSL/G.992 columns to the OFDM system comparison table
Mange01 (talk) 08:05, 5 May 2008 (UTC)

first graph incorrect?
"Line rate obtainable (Mbit/s) against corresponding line length (km) for ADSL, ADSL2+ and VDSL" graph. It has ~12 Mbaud data rate peak, contrary to 8 Mbaud in second graph (line rate vs attenuation). Also, does not reflect the extra throughput at -80dB available in ADSL2 compared to ADSL, as evident in second graph. Is the first graph erroneous, or am I mistaken?

Incidentally, what is the max upload rate for G.DMT? The article says 1.3 Mbaud, but Cisco says 864 kbaud.

64.180.126.80 (talk) 06:16, 4 May 2009 (UTC)