Speed for country folk

[#1]

Hi Alex
Has my router had its 10 days training yet? 

[#2]

Quote from: Alex

Personally i'd say that a target SNR of 6dB would be a little low for your line given the attenuation.  As a rule of thumb I use: Attenuation <30dB go for 6dB target SNR. Attenuation >30dB go for 9dB target SNR and above. (Scaling upwards +3dB target SNR for for each 10dB increase in attenuation.  i.e. Att. 30dB @ 9dB SNR, Att. 40dB @ 12dB...... etc)

Why do you think this, any factual evidence? dB is a ratio not absolute values so a 6dB margin on a line with an attenuation of 20dB is the same as a line at 40dB.

[#3]

My connection and absolutely stable although it may not be too happy if it wasn't interleaved (rebooted a week ago due to power cut before that it had been up for weeks) and you think I should have a 15dB attenuation - no way

[#4]

Quote from: itsme

Why do you think this, any factual evidence? dB is a ratio not absolute values so a 6dB margin on a line with an attenuation of 20dB is the same as a line at 40dB.

It's a rule of thumb.  There of course will be anomalies to this rule.  I have worked on thousands of customers broadband faults during my time at Plusnet and I was also one of the handful of agents involved with the BT 21cn DLM trial which involved manually fixing line profiles in place (this includes the SNR and Interleaving configuration).
We experiemented we all line profiles available to us and found that the afore mentioned rule tended to work for most customers.  Yes it's entirely possible to have a stable line with say 60dB loop loss and target SNR of 6dB.  It's just unlikely.  The rule was derived purely from my experience and experimentation with the systems directly over a pretty large sample size.
That being said, we're happy to try any configuration if a customer requests this within reason.

[#5]

Sorry but the maths will not support your assumption.
Lets say a modem will need to have a 70dB SNR for a stable connection plus a 6dB SNR margin. The modem will then sync at a speed with a 76dB SNR. Why will the attenuation change this? The SNR margin is for local conditions not a general factor based on attenuation.

[#6]

As per my previous comment, the rule was derived from real world testing on real lines.  This is not an assumption.  We have a large body of evidence to draw from as in the early stages of the trial every single line was documented.  At first we set every single line on the trial to target SNRM 6dB with Interleaving.  We found that as attenuation got higher we needed to make further adjustments to the tune of increasing the target SNRM to stabilise the lines.
Whilst it's true that if a modem needed an SNR of 70dB to operate at say 6Mbps stable and the target SNRM was set to 6dB this would mean a total SNR of 76dB.  However, as attenuation increases the variability in the noise level begins to increase which means it becomes more difficult to maintain a stable connection with a low SNR.  For this reason a lot of high attenuation lines begin to become unstable with low target SNRM values.  This is why as a rule of thumb we use this policy of <30dB = target SNRM 6dB, >30dB = target SNRM 9dB, etc.
Of course each line is an individual case and will operate perfectly well outside of this rule.  But this rule gives is a good starting point.  If your line was 60dB attenuation we'd probably start you on 12dB or 15dB target SNRM.  If the line was stable we'd gradually lower this until you see instability and then revert back to the last stable value.

[#7]

And if BT ever enable Seamless Rate Adaption, like the LLU suppliers, and which is a standard part of ADSL2+. We could all have 3dB SNR targets.
Then the speed of the line will seamlessly adapt to the noise conditions of the line at any given instance.

[#8]

Quote from: Alex

As per my previous comment, the rule was derived from real world testing on real lines.  This is not an assumption.  We have a large body of evidence to draw from as in the early stages of the trial every single line was documented.  At first we set every single line on the trial to target SNRM 6dB with Interleaving.  We found that as attenuation got higher we needed to make further adjustments to the tune of increasing the target SNRM to stabilise the lines.

Is this body of edvidence in the public domain as I would love to read it. As I keep saying dB is a ratio not an asolute value so a 76dB SNR on a line with an attenuation of 40dB is no different then on a 60dB line.The ratio of noise to signal is the same so your statement below does not come into the equation.

Quote

as attenuation increases the variability in the noise level begins to increase which means it becomes more difficult to maintain a stable connection with a low SNR

unless the system is incapable of reporting the true SNR which I have always suspected as you need to know the actual level of the signal at the modem and I can't see how this is done. 

[#9]

Does this explain this report on my connection

Quote

Estimated line speed:     1.5Mb (This may vary between 1Mb and 3.5Mb) - Checked on 2011-07-08 22:22:58 Current line speed:     4 Mb

as with a noise margin of 15dB my speed was about 3Mb and profile 2.5Mb and would the test mean that under the 3.5Mb rule I wouldn't have been eligible for ADSL2

[#10]

Quote from: itsme

... you need to know the actual level of the signal at the modem and I can't see how this is done.

Both modems tell each other the dBm power they each output, then each modem can calculate the attenuation.Whether all the tones are output at that power though is a good question.

[#11]

Bump
Has my router finished its training? see my speed below 

[#12]

@itsme,
I don't think we can publish the results that we collated as it contains personal details from customer accounts.  I understand where your coming from with regards to the ratio of signal to noise but during our trial the most useful rule we found was the one stated.  We don't stick rigidly to it so I don't feel it needs to be scrutinised further.  We are happy to try any configuration if we think it may be beneficial to the customer.
@Oldjim,
That just means your line is capable than higher than the estimated speed.
@godsell4
Attenuation is proportional to the signal frequency so as the ADSL signal frequency increases the reported line attenuation increases.  Most modems report the attenuation at 300kHz.
@PeterLoftus
Your line training period completed on 13/08/2011.

[#13]

Its just that i have been running routerstats and the dump I have attached shows uptime 6 days 9 hours 
How does it work? I havent reset my router is it part of the training 

[#14]

During the training period most of the line adjustments go through within the first 72 hours.  It's quite rare to see adjustments going on for the full ten days unless there is a fault of some kind.

[#15]

I had an event around 9pm this evening. I lost internet on Google Chrome and also could not get the router webpage with ethernet and Wifi. Eventually I had to power off the router and restore to get a connection back. I had RouterStats running during the event.
What happened? Is the line still unstable since the recent retrain? Is the router faulty?