[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
REPOST from Emmanuel.Duros: Processing new Layer 2 encapsulatio
- To: <ip-dvb@erg.abdn.ac.uk>
- Subject: REPOST from Emmanuel.Duros: Processing new Layer 2 encapsulatio
- From: Gorry Fairhurst <gorry@erg.abdn.ac.uk>
- Date: Thu, 13 Dec 2001 15:21:47 +0000
- User-agent: Microsoft-Outlook-Express-Macintosh-Edition/5.02.2022
The following message has been manually sent by the list
admininstrator:
> Subject: Processing new Layer 2 encapsulation
> Date: Thu, 13 Dec 2001 04:05:45 +0100
> From: Emmanuel Duros <Emmanuel.Duros@udcast.com>
> To: ip-dvb@erg.abdn.ac.uk
>
> Following the discussions of the (non-officiel) IPoDVB BoF, I have a
> comment on how to support new Layer 2 encapsulation within an MPEG II
> transport stream.
>
> We have been working and implementing drivers from various DVB card
> vendors. Many of these cards deliver the complete MPEG II transport to
> the driver.
>
> The driver must then, perform the following functions:
> - demultiplex the different PIDs from the MPEG II transport stream
> - filter PIDs if necessary
> - extract the MPE frame from each PID
> - filter each MPE frame according to MAC destination address
> - depending the LLC_SNAP deliver the payload of the MPE packet to the
> right input queue (IPv4, IPv6, arp, etc)
>
> As we may notice, most of the Layer 2 processing is performed in
> software whereas it would be sensible to do it in hardware. For
> instance, an Ethernetcard is capable to filter and discard undesired
> ehternet frames without the help of its driver.
>
> This has an obvious disadvantage, it is CPU time consuming since
> filtering is performed in software.
>
> However, it has one major advantage: any data type can be encapsulated
> within the MPEG II transport stream. The reception hardware delivers the
> MPEG II transport stream to the driver regardless to its content.
>
> As a result, supporting a new MPE-like frame format would only require
> to update the driver without changing the hardware/firmware. This might
> be considered if new encapsulation schemes have to be defined...
>
> Emmanuel