Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3_ISAAC] Questions on Conrad presentation

Hi Ariel, all,

In general, everything on slides 3 to 8 are purely ASA ML background explanation. This includes the protocol layer.

The first slide talking about MLE is slide 9, which should actually be headlined "ASA Motion Link Ethernet".

2. &  3.
Thanks for the feedback. You are correct, the protocol layer with Ethernet is less efficient and the overhead higher.
However, it does not change my point about the required MAC rates for the volume use cases.

Imager
Mpxls
fps
bpp
emb data
Payload rate [Gbps]
MAC Rate
3
15
12
1.08
0.58
2.5G
30
12
1.08
1.17
60
12
1.08
2.33
5
30
12
1.08
1.94
60
12
1.08
3.89
5G
8
30
12
1.08
3.11
45
12
1.08
4.67

4.
Even with 8MPxls and high quality settings, 5G MAC rate is fulfilling the requirement.

Only when you go to highest settings on 8MPxls like 60fps and 16bpp, will you need 10G MAC rate.
My point was not, that there are zero devices, which will require 10G MAC rate. But rather, that the fraction of devices, which will need that is very very small. And thus, the focus of optimization should be on the lower MAC rates, because this is where the volume of devices is and will remain.


BR, Conrad


From: Ariel Lasry <00002b0585801bb1-dmarc-request@xxxxxxxxxxxxxxxxx>
Sent: Thursday, October 10, 2024 8:04 PM
To: STDS-802-3-ISAAC@xxxxxxxxxxxxxxxxx
Subject: [802.3_ISAAC] Questions on Conrad presentation

CAUTION: This email is from an external origin!

Hi Conrad,

Just to recap the questions on your presentation.

  1.  Pages 3-6 discuss ASA ML own ASEP, PTB, ASAsec
I wanted to clarify if ASEP, PTB and ASAsec are proposed to be used with 802.3dm.
Following the question from Mr. Zimmerman, my understanding is NO.
This is OK for me. Just wanted to point out ASEP etc. cannot be re-used 1:1 as there will be MIPI CCPAL/A, IEEE1722 ACF GISF, PTP etc which are different than ASEP/PTB.


  1.  Page 11 you mentionned only 5% overhead is considered
When going over ethernet, CSI-2 (as well as I2C) will have additional protocol overhead for MIPI CCPAL/A, 1722 ACF GISF, ACF I2C (or GBB), MAC framing etc.
Is 5% realistic here?
Encapsulating CSI-2 raw data in Ethernet Frames requires:

     *   Network header: 18B (DA (6) + SA (6) + TPID (2) + TCI (2) + EtherType (2) = 18)

     *   AVTP header (1722): min is with NTSCF V0 = 12B; max is with TSCF V1 = 40B;

     *   GISF header: 20B

     *   SEP Header: min is 8B (more if CSE FuSa and/or Security is used)

     *   FCS: 4B

  1.  To that Packet over head need to be added: Preamble (7) + SFD (1) + IPG (12) = 20B
Taking only the minimum size for each layer overhead it adds 82 Bytes. With a regular 1500B payload it is already > 5%.

Additional at minimum per each video frame (lowest overhead):

     *   FuSa elements such as Message/Frame Counter and CRC-64
Security elements such as MACsec header and ICV (or SEP's Message Authentication Code) will add to each video frame.

  1.
The 8MP imager in the provided example may have to use 10G MAC rate mainly depending on the frame rate. This is more dominant than the protocol overhead. Why is the example stopping at 45fps?
Kind Regards
Ariel





________________________________

This email has been scanned for spam and viruses by Proofpoint Essentials. Click here<https://us2.proofpointessentials.com/app/report_spam.php?mod_id=11&mod_option=logitem&report=1&type=easyspam&k=k1&payload=53616c7465645f5facaa6b127ece5a9b13bf017ecadf0e2dd863b80465a6fb9eeb51214d51fc6946c4ee2a220fb54a28c4be184fb90d1c3961baff7d47df0c71d74910caf57c106993febc265080902f9d930947734797d29f62ebf535770508b3e2cdb03b197b29638eebbc102590d5fceb0e9b31f11d38ef100415b7c1cf93353dab4dcc238aad8280950c4bcc8ad46216c1a164e70da27d33c0a5d438ac9d> to report this email as spam.

________________________________

To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1

________________________________________________________________________
To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1