Demodulating DOCSIS Downstream: MPEG2 Transport Stream

Hi guys,

I am trying to demodulate QAM256 for DOCSIS.
DOCSIS is using a sublayer of the DVB-C system for multiplexing with other
content, so the sublayr is a constant stream of MPEG2-TS packets of 188
bytes, which also by themself are contained within RS and FEC frames of
their own.

This PDF helps understanding the layers:
http://eu.sabotage.org/www/ITU/J/J0083e.pdf

I want to do carrier recovery and symbol timing recovery.
There are some non-data-aided algorithms but I'd like to use something that
IS data-aided.
The problem is that I don't know what data to use.

I saw every FEC frame has a trailer of 5 known bytes. Is that good enough?

haxorcize wrote:

> Hi guys,
> 
> I am trying to demodulate QAM256 for DOCSIS.
> DOCSIS is using a sublayer of the DVB-C system for multiplexing with other
> content, so the sublayr is a constant stream of MPEG2-TS packets of 188
> bytes, which also by themself are contained within RS and FEC frames of
> their own.
> 
> This PDF helps understanding the layers:
> http://eu.sabotage.org/www/ITU/J/J0083e.pdf
> 
> I want to do carrier recovery and symbol timing recovery.
> There are some non-data-aided algorithms but I'd like to use something that
> IS data-aided.
> The problem is that I don't know what data to use.
> 
> I saw every FEC frame has a trailer of 5 known bytes. Is that good enough?

On Wed, 09 Nov 2011 02:06:18 -0600, "haxorcize"
<haxorcize@n_o_s_p_a_m.gmail.com> wrote:

>Hi guys,
>
>I am trying to demodulate QAM256 for DOCSIS.
>DOCSIS is using a sublayer of the DVB-C system for multiplexing with other
>content, so the sublayr is a constant stream of MPEG2-TS packets of 188
>bytes, which also by themself are contained within RS and FEC frames of
>their own.
>
>This PDF helps understanding the layers:
>http://eu.sabotage.org/www/ITU/J/J0083e.pdf
>
>I want to do carrier recovery and symbol timing recovery.
>There are some non-data-aided algorithms but I'd like to use something that
>IS data-aided.
>The problem is that I don't know what data to use.
>
>I saw every FEC frame has a trailer of 5 known bytes. Is that good enough?

Looks like it.   You need to sort out how the synch sequence maps to
the constellation points and go from there.

FWIW, since this is a continuous system that operates with a pretty
high SNR I suspect there's no real need to do data-aided
synchronization.  The usual blind methods should work just fine and
will be independent of the framing.






Eric Jacobsen
Anchor Hill Communications
www.anchorhill.com

Dear Vladimir,

F-U-C-K Y-O-U.

I've searched the word STUPIDENT throughout this news group, and all I've
seen are your stupid comments over posts.

Will you get a life?
How come you're always such a DICK with your answers?

I am not a stupid student having a hard time comprehending this stuff, I am
merely trying to study this as a hobby, and unfortunately I cannot dedicate
my whole life - unlike you - to sit here and criticize people, and try to 
emphasize my superiority.

HOORAY FOR YOU, for working in this field for many years and knowing most
of this stuff. I'm sure that if you'll come to my field of work I could
have made more fun of you than you can imagine, so don't be such a smart
ass.
This is not an easy subject because documentation is worth shit and
experience here can save a lot of time, like in many other fields.

When I come here to ask for help, I am openly looking for shortcuts, to
make what otherwise will take me many hours and days to look for and try to
access through some obscure IEEE documentation I can't.

But you know what, even if I were a stupid student,
WHO THE FUCK ARE YOU to criticize?

Go and get a life, stop rubbing your nose in other people's questions.

>
>
>haxorcize wrote:
>
>> Hi guys,
>> 
>> I am trying to demodulate QAM256 for DOCSIS.
>> DOCSIS is using a sublayer of the DVB-C system for multiplexing with
other
>> content, so the sublayr is a constant stream of MPEG2-TS packets of 188
>> bytes, which also by themself are contained within RS and FEC frames of
>> their own.
>> 
>> This PDF helps understanding the layers:
>> http://eu.sabotage.org/www/ITU/J/J0083e.pdf
>> 
>> I want to do carrier recovery and symbol timing recovery.
>> There are some non-data-aided algorithms but I'd like to use something
that
>> IS data-aided.
>> The problem is that I don't know what data to use.
>> 
>> I saw every FEC frame has a trailer of 5 known bytes. Is that good
enough?
>

And if it wasn't clear before, I will emphasize again:

I am not a student (at least not for EE or DSP).
I am not looking to cheat on some exam or task.
I don't know any university that asks their students to demodulate DOCSIS
traffic.

This was, and is, out of a personal interest.

>Dear Vladimir,
>
>F-U-C-K Y-O-U.
>
>I've searched the word STUPIDENT throughout this news group, and all I've
>seen are your stupid comments over posts.
>
>Will you get a life?
>How come you're always such a DICK with your answers?
>
>I am not a stupid student having a hard time comprehending this stuff, I
am
>merely trying to study this as a hobby, and unfortunately I cannot
dedicate
>my whole life - unlike you - to sit here and criticize people, and try to

>emphasize my superiority.
>
>HOORAY FOR YOU, for working in this field for many years and knowing most
>of this stuff. I'm sure that if you'll come to my field of work I could
>have made more fun of you than you can imagine, so don't be such a smart
>ass.
>This is not an easy subject because documentation is worth shit and
>experience here can save a lot of time, like in many other fields.
>
>When I come here to ask for help, I am openly looking for shortcuts, to
>make what otherwise will take me many hours and days to look for and try
to
>access through some obscure IEEE documentation I can't.
>
>But you know what, even if I were a stupid student,
>WHO THE FUCK ARE YOU to criticize?
>
>Go and get a life, stop rubbing your nose in other people's questions.
>
>>
>>
>>haxorcize wrote:
>>
>>> Hi guys,
>>> 
>>> I am trying to demodulate QAM256 for DOCSIS.
>>> DOCSIS is using a sublayer of the DVB-C system for multiplexing with
>other
>>> content, so the sublayr is a constant stream of MPEG2-TS packets of
188
>>> bytes, which also by themself are contained within RS and FEC frames
of
>>> their own.
>>> 
>>> This PDF helps understanding the layers:
>>> http://eu.sabotage.org/www/ITU/J/J0083e.pdf
>>> 
>>> I want to do carrier recovery and symbol timing recovery.
>>> There are some non-data-aided algorithms but I'd like to use something
>that
>>> IS data-aided.
>>> The problem is that I don't know what data to use.
>>> 
>>> I saw every FEC frame has a trailer of 5 known bytes. Is that good
>enough?
>>
>

haxorcize wrote:

> And if it wasn't clear before, I will emphasize again:
> 
> I am not a student (at least not for EE or DSP).
> I am not looking to cheat on some exam or task.
> I don't know any university that asks their students to demodulate DOCSIS
> traffic.
> 
> This was, and is, out of a personal interest.
> 
> 
>>Dear Vladimir,
>>
>>F-U-C-K Y-O-U.
>>
>>I've searched the word STUPIDENT throughout this news group, and all I've
>>seen are your stupid comments over posts.
>>
>>Will you get a life?
>>How come you're always such a DICK with your answers?
>>
>>I am not a stupid student having a hard time comprehending this stuff, I
> 
> am
> 
>>merely trying to study this as a hobby, and unfortunately I cannot
> 
> dedicate
> 
>>my whole life - unlike you - to sit here and criticize people, and try to
> 
> 
>>emphasize my superiority.
>>
>>HOORAY FOR YOU, for working in this field for many years and knowing most
>>of this stuff. I'm sure that if you'll come to my field of work I could
>>have made more fun of you than you can imagine, so don't be such a smart
>>ass.
>>This is not an easy subject because documentation is worth shit and
>>experience here can save a lot of time, like in many other fields.
>>
>>When I come here to ask for help, I am openly looking for shortcuts, to
>>make what otherwise will take me many hours and days to look for and try
> 
> to
> 
>>access through some obscure IEEE documentation I can't.
>>
>>But you know what, even if I were a stupid student,
>>WHO THE FUCK ARE YOU to criticize?
>>
>>Go and get a life, stop rubbing your nose in other people's questions.
>>
>>
>>>
>>>haxorcize wrote:
>>>
>>>
>>>>Hi guys,
>>>>
>>>>I am trying to demodulate QAM256 for DOCSIS.
>>>>DOCSIS is using a sublayer of the DVB-C system for multiplexing with
>>
>>other
>>
>>>>content, so the sublayr is a constant stream of MPEG2-TS packets of
> 
> 188
> 
>>>>bytes, which also by themself are contained within RS and FEC frames
> 
> of
> 
>>>>their own.
>>>>
>>>>This PDF helps understanding the layers:
>>>>http://eu.sabotage.org/www/ITU/J/J0083e.pdf
>>>>
>>>>I want to do carrier recovery and symbol timing recovery.
>>>>There are some non-data-aided algorithms but I'd like to use something
>>
>>that
>>
>>>>IS data-aided.
>>>>The problem is that I don't know what data to use.
>>>>
>>>>I saw every FEC frame has a trailer of 5 known bytes. Is that good
>>
>>enough?
>>

Cheers,
Dave

On Wed, 09 Nov 2011 18:24:14 -0600, haxorcize wrote:

> Dear Vladimir,

Please don't feed the troll.  He actually has some positive input, 
sometimes, when he's not trying to drive people away from the group.

This is USENET.  There is no material that you can pour onto a flame war 
other than gasoline -- your best bet is to not say anything (or limit 
your comments to the minimum, as I'm going to do right about -- now)

-- 
www.wescottdesign.com

On Wed, 09 Nov 2011 02:06:18 -0600, haxorcize wrote:

> Hi guys,
> 
> I am trying to demodulate QAM256 for DOCSIS. DOCSIS is using a sublayer
> of the DVB-C system for multiplexing with other content, so the sublayr
> is a constant stream of MPEG2-TS packets of 188 bytes, which also by
> themself are contained within RS and FEC frames of their own.
> 
> This PDF helps understanding the layers:
> http://eu.sabotage.org/www/ITU/J/J0083e.pdf
> 
> I want to do carrier recovery and symbol timing recovery. There are some
> non-data-aided algorithms but I'd like to use something that IS
> data-aided.
> The problem is that I don't know what data to use.
> 
> I saw every FEC frame has a trailer of 5 known bytes. Is that good
> enough?

AFAIK most data aided algorithms work by first decoding the data, then 
saying to themselves "OK, given that the transmitted data was X, what are 
the channel's various gains & offsets that I must correct for?"

So they're really not much good until you actually _are_ decoding the 
data, or if the modulation is simple enough (like BPSK) that the data 
ambiguity is not great.

-- 
www.wescottdesign.com

On 11/10/11 12:09 PM, Tim Wescott wrote:
>
> This is USENET.  There is no material that you can pour onto a flame war
> other than gasoline --

weren't you the least bit firebug when you were a kid?

i remember several times, in the 60s, having some fun in the middle of 
Nowhere North Dakota with less than 10 cents worth of Regular.  even 
more so when mixed with fireworks and cherry bombs and M-80s.

this thread might be fun to watch, not that i support Vlad in abusing 
the newbs, or anyone else he wants to pick on, but it can be interesting 
to observe.

-- 

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."