Minutes of 2nd Open IP-DVB Meeting (IP over MPEG-2 Transport)
14:00 Monday, 24th June
Hosted by Alcatel Space Industries, Toulouse, France

Stephane Combes started the meeting with a welcome to ASPI Toulouse. Gorry Fairhurst and Marie-Jose Montpetit chaired the meeting.

There were no new agenda items, but Stephane Combes was asked to start with the first presentation to allow time to prepare the other presentations.

Presentation of Alcatel Space Industries (ASPI) activities

Stephane Combes provided a presentation of a range of Alcatel Space Industries (ASPI) activities relating to Internet service provision via satellite using DVB-based systems. Slides are available HERE.

This showed the perceived need for two work in two areas:

MPE enhancement (short term)

He noted the issue of no MAC-level source address in the basic MPE encapsulation. This raised issues for multicast encpasulation when there were several sources.

There was also interest from ISPs in MPEG-2 delivery of an Ethernet Like layer (in the manner of ADSL). MPLS and PPoE options were also being investigated.

MPE replacement (long term)

Longer-term he saw a need for a dynamic address resolution process, and better support for multicast. Security issues also needed to be addressed.

Some work had already been done by ASPI on MPE replacement, defining a protocol called IP-Dedicated. This was aimed at two-way DVB-RCS based satellite systems. One key feature of this work was the use of L3 filters (i.e., internet address filters in the receivers), but he concluded a simple level 2 filter was also desirable to reduce layer 3 processing. This would also allow virtual subnetworks to be configured and to lead to future L2 switching within a DVB multiplexor (or DVB-capable on-board processing satellite).

He described two current projects looking at IP over MPEG-2 transmission that have been funded by the European Commission (IST-Brahms, IST-SATIP6).

Longer-term he saw the need to improve IPv6 support. Some ASPI work had some associated IPR. It was thought that the ip-dvb activity would complement this work, and ASPI would like to support it.

REVIEW OF CHARTER

Gorry presented a status report for the ip-dvb activity. This covered, liasons with DVB, the desired work of the group, and the charter. Copies of all gorry's slides are availabe HERE.

Marie-Jose suggested that a short one page summary document should be prepared to position the work clearly. This could be used as a marketing document for briefing DVB equipment manufacturers on the proposed new protocol stack. When asked, she agreed to draft this document for the mailing list.

No new items were suggested for the charter.

REVIEW OF IDS

Gorry Fairhurst presented the status of the two IDs (no other authors were present).

REQUIREMENTS (draft-fair-ipdvb-req-01.txt)

The requirements draft had recently been updated to -01, to include rationale for signalling following the combined meeting with DVB-GBS. This DVB working group had produced an DVB draft for signalling (INT). This would be based on DVB SI signalling tables. A new version of the INT document was in preparation, status was not known. Diagrams had also been added, to the -01 draft. Gorry asked for corrections / comments.

One issue raised by the DVB INT draft was the need for 3platform ids2 to describe the scope of the addresses being advertised. A single MPEG-2 TS logical channel could be used to address several virtual private networks, each with overlapping address space, and uniquely identified by MAC address and platform ID. Was this a good idea? There was some discussion of whether
this approach was suited to more general Interent needs. No conclusion.

ENCAPSULATION (draft-unisal-ipdvb-enc-00.txt)

The encapsulation draft was presented and a discussion followed. Patrick Cipiere asked why there was no MAC address in the basic header format, only as an extension header?

Since MPE included a destination MAC address, some suggested that L2 addresses should at least be part of the basic encapsulation.

There was discussion about the possibility of using the most significant bits of the length field to hold two flag bits (as originally suggested by Patrick on the mailing list). One signalling bit could indicate presence ofa MAC source address, another a MAC destination address. This also seemed to have the advantage that a 01 setting would be backwards compatible with MPE, and 00 would be equivalent to the current proposal (-00).

Marie-Jose asked when a MAC address was actually required.... This should be documented.

If we proceede with this - we would need to find space for the signalling bits in the fixed header. Two places were possible, one in the length field and one in the type field. Do we need to support packets greater than 16 KB?

When this had previously been posted as a proposal to the list it had received little comment. It was agreed to take this to the list again, in time for the next rev of the draft.

Gorry said the current case for using the AFC bits in the MPEG-2 TS Packet header was not well stated at the moment. This should be updated in the next draft.

ADDRESS RESOLUTION

Gorry Fairhurst suggested there was now a need for an ID from the ip-dvb activity to clarify how address resolution could work.

IPv6

Marie-Jose presented a set of slides on IPv6 and looked at RFC2461 and RDISC. The meeting concluded that some attention should be paid to router and neighbourhood discovery when looking at addrress resolution for IPv6. A critical look at multi-homing may also be appropriate for home scenarios were DVB and other broadband services may all be present.

IP-CC

Pekka (Nokia) presented a set of slides on the proposed role of IP-CC for service discovery. This draft proposal had previously been presented at DVB-GBS working group, and was intended to provide service discovery. A copy of the draft document is available HERE.

The current proposal was aimed at DVB-T (terrestrial) TV networks supporting IP services. There were similarities to the way SAP is used in current IP multicast networks, but a broader scope was envisaged for the new protocol. One key issue was to provide mechanisms for location-based delivery, in which the service was associated with a geographical region.

The proposal seemed to be complimentary to the proposed work on IP over MPEG-2. It was discussed whether this work was within the charter of the ip-dvb activity - or whether it should be placed with another working group. Since it was not within the current charter, there was no decision to progress with this within the group.

CONCLUSIONS

There was discussion of the possibility of developing some reference implementations to allow experimentation with the new encapsulation. Frank Zeppenfeldt from the European Space Agency said they would be interested in receiving project ideas associated with implementation of the new
encapsulation (and possibly other protocol components).

The meeting agreed that more work was required to understand exactly what was being proposed before we were ready to bring this to the IETF, but that considerable progress had been made and that there was a clear enthusiasm to proceed. Would we be ready for October?

Gorry agreed to contact the appropriate IESG Area Directors to ask about the possibility of a BoF in Atlanta. He would also contact the DVB-GBS WG chair to find out progress with their signalling work.

The circulated attendance sheet showed 14 people were present. Thanks to Stephane Combes for arranging lunch and for the good local organisation of this meeting.