Cell Relay Archive

["XuHonggang" <xuhg@cet.st.com.sg>]

Yes, AAL1 SDT requires only one pointer in a group of 8 cells.
Imlications of this? I see, every cell is precious, don't
waste it.
-----Original Message-----
From: skasera@dl.in.dsqsoft.com <skasera@dl.in.dsqsoft.com>
To: XuHonggang <xuhg@cet.st.com.sg>
Cc: cell-relay@cell.onecall.net <cell-relay@cell.onecall.net>
Date: Friday, July 28, 2000 2:33 PM
Subject: Re: CR: Structured Data Transfer in AAL Type 1 !


>
>To add to Mr. Raj's doubt's, I have the following to add: the spec also
>states somewhere that only one pointer field can be sent in a group of 8
>cells. Is this true? If yes, what are its implications?
>
>regards,
>sumit kasera
>
>
>
>
>                    "XuHonggang"
>                    <xuhg@cet.st.com.sg>           To:     "Rajgopalan
Sampath"
>                    Sent by:
<rajagopalan.sampath@wipro.com>, "Cell Relay"
>                    owner-cell-relay@cell.o
<cell-relay@cell.onecall.net>
>                    necall.net                     cc:
>                                                   Subject:     Re: CR:
Structured Data Transfer
>                                                   in AAL Type 1 !
>                    07/28/2000 07:28 AM
>                    Please respond to
>                    "XuHonggang"
>
>
>
>
>
>
>
>-----Original Message-----
>From: Rajgopalan Sampath <rajagopalan.sampath@wipro.com>
>To: XuHonggang <xuhg@cet.st.com.sg>; Cell Relay
><cell-relay@cell.onecall.net>
>Date: Thursday, July 27, 2000 7:21 PM
>Subject: Re: CR: Structured Data Transfer in AAL Type 1 !
>
>
>>Hi,
>>I still have some confusion.
>>Spec says that the DATA passed to the AAL1 is constant and the time
>difference
>>between two consecutive Requests is also constant.
>>
>>For example, in a Circuit Emulation Service ( N x 64kbps )  it has to pass
>'1'
>>byte of data from each channel to the AAL1 once in every 125/N micro
>seconds.
>>According to the spec, Circuit emulation used SDT. So, always a Structure
>>information will be passed to AAL1.
>
>Not exactly, it has to pass '1' byte for all chanels in Nx6k service in
>every 0.125 ms, 1/8k.
>
>>My question is:
>>1. Does the AAL1 user request the AAL1 once in every 125/N micro seconds
>with 1
>>byte of data? In that case do AAL1 need to do any buffering until it gets
>47
>>octets from the user?
>
>How many times AAL1 user request the AAL1 is up to AAL1 user, but the AAL1
>user
>have to keep its buffer from overflow and underflow. Just image its input
>is
>continuously
>in rates of one octet for each channel in every 0.125ms.
>
>>2. My assumption is, the pointer ( P-format ) is always send in the PDU
>with Seq
>>Count '0'? Is this right? If the pointer is sent in the first PDU of a
>cycle (
>>i.e. the PDU with Sequence count 0 ), what'll be sent in the remaining
>PDUs
>of
>>the same cycle ( i.e in the PDUs with Seq Count 2, 4, 6 )? Can we send a
>Non-P
>>format cells in place of P-format cells? i.e. the whole PDU ( 47 bytes)
>>will  be filled only with user information irrespective of the sequence
>number?
>>Is there any hard and fast rule like the Non-P format cells should be sent
>only
>>with Seq Count 1, 3, 5 and 7?
>>
>
>Use or not use P-format has to be decided before a AAL1 VCC connection
>being
>setup,
>it is out of I.363.1's scope. You have to use a signaling system or
>predefined to make sure two ends of a AAL1 connections know that before
>they
>start to transfer/receive data
>
>>If I'm wrong, what the mechanism, in which the AAL1 user sends data to the
>AAL1.
>>
>>Also, in a Voiceband signal service, only one 64kbps voice channel is
>used,
>in
>>which the data doesn't use any structure information. In this case, how
>the
>DATA
>>is passed to the AAL1?
>
>In our case, we fully use 47 octets for one channel in one vcc, but that
>cause 47 * 0.125
>ms delay. To reduce delay, you can use partial payload. Also you need to
>notice two sides
>your decision.
>
>>The data always sent in Non-P format in a cycle, except the 7th PDU with
>Seq
>>Count '6'. In which the pointer always takes the value "127" ( the
>default )
>>value.
>>
>>Can I assume that, at transmit side, the AAL1 user guarantees CBR? So,
>partial
>>filling of PDUs will not be needed at CS layer to compensate the "cell
>assembly
>>delay".
>>
>>Plse tell, whether my understanding is correct.
>>
>>Thanks in advance.
>>
>>-Rajagopalan.
>>
>>XuHonggang wrote:
>>
>>> Hi,
>>> Pointer field is 7 bits, so max offset which it can idicate is 127
>Octets
>>> which < 3 ATM cells p and > 2 ATM cells, so standard maker decided to
>use
>96
>>> as max number wich align with 2 ATM cells.
>>> You put a pointer in no.0 cell, it can cover all possible offset in no.0
>and
>>> no.1, so second pointer only need to appear in no.2, the same thing
>happen
>>> to no.4 and no.6 cell.
>>> Yes, a octet means 8 bits, means one sample of a voice channel. How many
>>> octets
>>> in your structure means how many voice channels in your structure.
>>> wish it help
>>> XuHonggang
>>> SwitchingSystemDepartment
>>> CET Techologies of Singapore Technologies
>>> 24 AngMoKio Str. 65
>>> Singapore 569061
>>> -----Original Message-----
>>> From: Nuzhath Shereen <shereen@chequemail.com>
>>> To: sasos@nortelnetworks.com <sasos@nortelnetworks.com>
>>> Cc: rajagopalan.sampath@wipro.com <rajagopalan.sampath@wipro.com>;
>>> cell-relay@cell.onecall.net <cell-relay@cell.onecall.net>
>>> Date: Thursday, July 27, 2000 5:26 PM
>>> Subject: Re: CR: Structured Data Transfer in AAL Type 1 !
>>>
>>> >
>>> >
>>> >Hello Saso,
>>> >
>>> >Can you explain more about this SDT format. The ITU-T I.363.1  says
>that
>>> the pointer format may be used only if the sequence count value is 0,2,4
>or
>>> 6. This is because the 1,3,5 and 7 values are reserved for SRTS.
>>> >
>>> > I didn't understand the rest of the explanation in the standard. What
>is
>>> this 93-octet payload it refers to. By octet, Is it referring to one
>64kps
>>> channel (I had read that somewhere) or is the digital 8 bits.
>>> >
>>> >Please give links or references , if you have any
>>> >
>>> >Regards,
>>> >Shereen@chequemail.com
>>> >
>>> >On Wed, 26 July 2000, "Saso Stojanovski" wrote:
>>> >
>>> >>
>>> >>
>>> >> hi rajagopalan,
>>> >>
>>> >> when using AAL-1 for circuit emulation, you either care to specify to
>the
>>> >> receiver where the TDM frame boundary is or you don't.
>>> >>
>>> >> in the former case you would use SDT, and in the latter case you
>would
>>> >> rely on the sync mechanism which is built in the TDM frame itself.
>>> >>
>>> >> for instance, if you emulate a whole E-1 circuit, then you could use
>UDT,
>>> >> since the first timeslot (out of 32) in the E-1 frame contains a sync
>>> >> pattern.
>>> >>
>>> >> on the other hand, if you are to emulate a fractional E-1 circuit
>(say
>>> >> only k 64 kbit/s channels), then you must use SDT in order to tell
>the
>>> >> receiver where exactly in the ATM payload the TDM frame (whose length
>is
>>> >> only k octets) starts. that is exactly what the pointer points at.
>>> >>
>>> >> cheers,
>>> >>     v
>>> >> --saso
>>> >>
>>> >>
>>> >>
>>> >>
>>> >> rajagopalan.sampath@wipro.com wrote:
>>> >>
>>> >> > Hi,
>>> >> >
>>> >> > Can somebody help me in implementing AAL type 1  ( ITU-T I.363.1 )?
>>> >> >
>>> >> > What's SDT ( Structured Data Transfer ) and why it's used?
>>> >> > How SDT is performed at transmitting and receiving end? It's
>basically
>>> >> > about when to select P-format and when to select Non-P format PDUs?
>>> >> > What does the pointer value represent in the P-format PDUs?
>>> >> >
>>> >> > Thanks.
>>> >> >
>>> >> > Regds,
>>> >> > Rajagopalan Sampath.
>>> >
>>> >
>>> >Chequemail.com - a free web based e-mail service that also pays!!!
>>> >http://www.chequemail.com
>>> >
>>> >
>>
>>--
>>Rajagopalan Sampath,
>>Wipro Technologies,
>>Bangalore.
>>Ph: 5722293/96. Extn: 5134.
>>
>>
>>
>>
>
>
>
>
>
>
>