home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!pipex!bnr.co.uk!uknet!pavo.csi.cam.ac.uk!dm
- From: dm@cl.cam.ac.uk (Derek McAuley)
- Newsgroups: comp.dcom.cell-relay
- Subject: Re: two more questions re: SONET and ATM
- Message-ID: <1993Jan5.115759.14560@infodev.cam.ac.uk>
- Date: 5 Jan 93 11:57:59 GMT
- References: <1993Jan4.173229.25170@sics.se> <1993Jan4.190553.29754@afterlife.ncsc.mil>
- Sender: news@infodev.cam.ac.uk (USENET news)
- Organization: U of Cambridge Computer Lab, UK
- Lines: 36
- Nntp-Posting-Host: malin.cl.cam.ac.uk
-
- > one could load the sync sequence (bytes A1 and A2) into the payload
- > of an ATM cell and screw up the SONET multiplexors.
-
- My understanding is that there are two scramblers, one at the SONET frame level
- and another for cell payloads. The frame scrambler is applied to the payload - I
- believe is is applied after the line overhead (which includes the sync sequence)
- but is reset to s defined state each frame - hence repeated idenitical payloads
- generate the same bits on the wire and so it is easy to arrange a repeated sync
- pattern in the payload. Note that so far this is nothing to do with ATM and is
- required even if SONET is used to carry a VC4 holding 2340 telephone calls.
-
- For ATM within a VC4, the H4 pointer is now no longer used; this is beacuse
- lower rate virtual containers do not have it and a generic solution is desired.
- (I guess the obvious one of aligning cells within each VC frame is considered to
- be a gross waste of space -- a VC4 has 2340 bytes hence 44 cells and 8 bytes.)
- So it is now suggested that cell alignment within the container is done by
- hunting for correct 5 octet headers within the SONET payload (after SONET frame
- descrambling) -- i.e. 4 bytes followed by correct HEC -- hence it is necessary
- to further scramble the cell payloads to stop (that is decrease to small
- probability) someone inserting a chain of 5 octet sequences that look like cell
- headers into cell payloads. In this case I believe the scrmabler is not reset
- per cell and hence repeated payloads do not generate the same pattern.
-
- Note that in both hunt modes, any sensible implementation should require a
- significant number of correct sync / HEC occurances *at the correct frequency*
- before declaring lock and then should not be thrown by a correct sync/HEC
- patterns that appear at random.
-
- A malicous person could easily defeat the SONET scrambler, the ATM level takes
- more work -- if one knew the state of the scrambler, one can clearly define a
- payload to defeat it. However, the trick is to find a pattern that can defeat
- the SONET scrambler but is also a legal VC4 containing legal ATM cells for legal
- VCI/VPI pairs so that you might have a defence when the carrier tries to sue you
- for screwing their network |-)
-
- Mac.
-