Interleaving

[joe]

My current stats show a DS interleaf figure of 19 and US of 1. The former I believe shows interleaving to be operative, the latter not.




If interleaving is operating on DS - why is it? My figures are very stable and errors are low. Is this a hangover from pre FTTC days that can be corrected?

What, if any, detriment does interleaving cause?

[.Griff.]

After seeing numerous line stats from the Huawei I'm not convinced that "D" is the depth of interleaving and if it is then it's not being reported correctly.

As for what detriment interleaving causes it results in higher latency. The higher the level (depth) of interleaving applied to the line the higher the latency.

[Bald_Eagle1]

Hi .Griff,

After seeing numerous line stats from the Huawei I'm not convinced that "D" is the depth of interleaving and if it is then it's not being reported correctly.

What makes you think that "D" is not interleaving depth and/or if it is, that it is not being reported correctly?

There is a bit of an explanation & some document links here:-
http://forum.kitz.co.uk/index.php/topic,10289.0.html


These are my most recent 40 day stats:-







Paul.

[Steve]

 

[Ray]

 

[.Griff.]

Hi .Griff,

What makes you think that "D" is not interleaving depth and/or if it is, that it is not being reported correctly?

Hi Paul,

I've seen VDSL2 line stats from various people and on each occasion the "Interleaving" was reported as 19 (Down) and 1 (Up). For example I'm synced at 40/10 (95/35 Attainable) while a friend on a 1.4KM line syncs at 22/2 and if the modem is to be believed also shares the same interleaving depth as me.

[Bald_Eagle1]

Just to add a few more snippets of info:-

From a different user's stats we can see that DS & US Interleaving are OFF (values of D: are both 1):-

Interleaving levels appear to be connected to INP (Impulse Noise Protection).
For this user, the INP values are both 0.00 (shown lower down the extracted stats).

Code Select Expand

VDSL2 framing
Path 0
B: 239 223
M: 1 1
T: 64 35
R: 0 16
S: 0.1910 0.7805
L: 10054 2460
D: 1 1
I: 240 120
N: 240 240
Counters
Path 0
OHF: 26881481 1124041
OHFErr: 1492 140
RS: 0 1695887
RSCorr: 0 259
RSUnCorr: 0 0

Path 0
HEC: 4143 0
OCD: 0 0
LCD: 0 0
Total Cells: 2042409149 0
Data Cells: 114539597 0
Drop Cells: 0
Bit Errors: 0 0

ES: 409 130
SES: 4 0
UAS: 16 16
AS: 82459
Path 0
INP: 0.00 0.00
PER: 3.05 13.65
delay: 0.00 0.00
OR: 62.83 35.14
Bitswap: 3486 2599



Now for my connection, my Interleaving levels (D:) have been over 1700 on DS, currently it is at 457.
A depth of 457 is classed as ON - LOW by my ISP.
My US Interleaving level (D: i.e. depth of 1) is currently OFF:-

Code Select Expand

VDSL2 framing
Path 0
B: 63 175
M: 1 1
T: 64 39
R: 16 12
S: 0.0706 0.9447
L: 9065 1592
D: 457 1
I: 80 94
N: 80 188
Counters
Path 0
OHF: 140050172 728159
OHFErr: 4614 434
RS: 1119877676 3030422
RSCorr: 64282096 7803
RSUnCorr: 190815 0

Path 0
HEC: 55568 0
OCD: 0 0
LCD: 0 0
Total Cells: 625204143 0
Data Cells: 39608955 0
Drop Cells: 0
Bit Errors: 0 0

ES: 1091 411
SES: 71 0
UAS: 32 32
AS: 476469

Path 0
INP: 3.00 0.00
PER: 3.38 9.21
delay: 8.00 0.00
OR: 56.65 27.79

Bitswap: 146788 7864

My INP levels are currently 3.00 DS & 0.00 US.

When my Interleaving depth has been around 1700 or so, I have had INP values of 8.50 DS, but US has remained at 0.00

A little more discussion on this topic can be viewed via this link:-

http://forum.kitz.co.uk/index.php/topic,10526.0.html

Paul.

[Simon]

  Paul.

[Bald_Eagle1]

Hi .Griff,

I've seen VDSL2 line stats from various people and on each occasion the "Interleaving" was reported as 19 (Down) and 1 (Up). For example I'm synced at 40/10 (95/35 Attainable) while a friend on a 1.4KM line syncs at 22/2 and if the modem is to be believed also shares the same interleaving depth as me.

I have to admit I'm only guessing now, but perhaps your friend has a highly attenuated connection thus low sync speeds (due to distance from the cabinet), but does not suffer with much noise, or error levels.

Some graphs of yours & his stats, or even the raw stats may give us an insight as to what goes on regarding INP, Interleaving & even delay that I haven't mentioned previously.

I have harvested & plotted the stats in my graphs from some Windows only batch files that I have developed from the original Linux scripts.

You will see that the original poster in this thread (joe) has also used the same batch files as his example is an identical format to mine.
The only real difference being that he has a much faster & relatively error-free connection than mine 

My own connection was quicker, & as far as I am aware fairly error-free for the first month (8c Profile) than it is now.
I was actually able to download at 33Mb, suggesting a sync speed of 35Mb or so.
(My DS Interleaving was reported as OFF by my ISP at that time, but that was before I could access my own stats).
I have experienced many physical D-side copper cable problems since then (still not fully resolved).

Unfortunately I wasn't able to see my connection's stats back then (Late June/early July) to compare with my current stats.

If anything, the switch to 17a profile SHOULD have increased my sync speeds, possibly a lot closer the full 40Mb by now?

I THINK there is still a HR (High Resistance) issue still present somewhere in my D-side copper, that has only allowed me to sync at below 30Mb for a few months.
Currently my sync speed is only 28839 k.


Paul.


EDIT: FYI, Pauline is no longer with us (R.I.P.)

That was just my pseudonym when I was being a bit more discrete about having access to an unlocked modem (I was expecting a few BT Engineer visits etc. at that time).

[Bald_Eagle1]

Just one last comment before I let joe have his thread back that I appear to have hijacked:-

If anyone would like copies of the Windows only batch files, along with some setting up instructions & links to the other utilities that need to be downloaded, just send me a PM with your email address (for now).

I keep tweaking the batch files, so they are only "beta" versions at the moment (although they are fully working).

At some stage (quite soon-ish), I will upload them to the same place as the Linux scripts to keep everything together.
I believe the original author of the Linux scripts is planning to upload an update in the not too distant future.

Paul.

[.Griff.]

Cheers for the subsequent replies Paul.

Just in case you hadn't noticed I'm the same Griff that spoke to you on a few occasions on Kitz. You kindly plotted my line stats for me.

[Bald_Eagle1]

Hi .Griff,

Just in case you hadn't noticed I'm the same Griff that spoke to you on a few occasions on Kitz. You kindly plotted my line stats for me.

I did indeed notice.

I wasn't going to mention it, but as you have mentioned it, these were your stats from 12th November (Interleaving 19 & 1) :-

Code Select Expand

VDSL2 framing
Path 0
B: 240 239
M: 1 1
T: 64 35
R: 14 0
S: 0.1917 0.7622
L: 10644 2519
D: 19 1
I: 255 120
N: 255 240
Counters
Path 0
OHF: 68228540 653906
OHFErr: 312 9
RS: 71675433 2605243
RSCorr: 19186 0
RSUnCorr: 3610 0

Path 0
HEC: 2604 0
OCD: 0 0
LCD: 0 0
Total Cells: 3271607083 0
Data Cells: 140660543 0
Drop Cells: 0
Bit Errors: 0 0

ES: 10040 3324
SES: 374 10
UAS: 536 536
AS: 210048

Path 0
INP: 0.00 0.00
PER: 3.06 13.33
delay: 1.00 0.00
OR: 83.48 37.18

Bitswap: 136 976




I suspect, but I haven't found confirmation anywhere, that an interleaving level between 1 & "some other value" may be classed as OFF.
I see you had INP values of 0.00 & 0.00, with a delay of 1.00 & 0.00 at that time.

My ping times are usually 15ms now (at quiet times). I'm almost sure they used to be 5ms when my connection was performing better.



What sort of ping times do you usually see?


Paul.

[joe]

Thanks to .griff and Paul for their learned contributions to my understanding.

[Bald_Eagle1]

One final thing:-

As I understand things, when on Fastpath (i.e. Interleaving is definitely OFF), you will not see ANY FEC errors.

FEC (Forward Error Correction) is only in use when Interleaving is applied.

So, if you do see FEC errors, Interleaving must be ON. It may be at a very low level (say 19), but it is ON.

The HG612 modem does appear to get some of its error reporting a little mixed up in its GUI.
From its xdslcmd info --stats though, RSCorr (Reed Solomon Correction) is apparently another way of reporting FEC errors.

That is why I plot both FEC Errors (FROM GUI) AND RSCorr (FROM --stats) in my graphs.

You may have noted that the OHFErr (FROM --stats) graph (OverHead Frame Errors) is almost identical to the FEC Errors (FROM GUI) graph.

There are some very subtle differences in my plots that you may not see for your connections as you get nowhere near the quantity of errors that I experience, so they may not stand out as much.


Paul.