btdornMar 17
Dan York

An Architecture for IP in Deep Space - https://datatracker.ietf.org/doc/draft-many-tiptop-ip-architecture/

----
The IP protocol stacks used on Earth's Internet are typically configured based on assumptions of short delays and mostly uninterrupted communications. This document describes an architecture of the IP protocol stack tailored for its use in deep space. <snip> This architecture applies to the Moon, Mars, and general interplanetary networking.
----

#IETF #IETF125 #Internet #InternetAccess #space

An Architecture for IP in Deep Space

The IP protocol stacks used on Earth's Internet are typically configured based on assumptions of short delays and mostly uninterrupted communications. This document describes an architecture of the IP protocol stack tailored for its use in deep space. It involves buffering IP packets in IP forwarders facing intermittent links and adjusting transport protocol configurations and application protocol timers. This architecture applies to the Moon, Mars, and general interplanetary networking.

IETF Datatracker
Mar 17 at 1:59amWeb
Show threadKarl AuerbachMar 17

@danyork I build tools to help people exercise protocols under various conditions, including end-to-end latency measured in hours or days.

I remember when I was working with IP muliticast data streams (no clear end point) and muliticast file transfers (clarity about the end of the data). In that world we had basically two tools to assist accurate arrival of the data - to have the various receivers turn around and become sources to fill gaps (which is not possible for "deep space") or to use a lot of redundancy and forward error correction.

I don't know enough about "deep space" paths to know how often they are afflicted with errors - such as when the transmitter's location crosses the face of the sun. I do hope that those who design these protocols are aware of these kinds of "blanking intervals" (often entirely predictable) or other ways that the data flow can be damaged. (I have seen a lot of earth-satellite systems that forgot those things.)

Show threadDan YorkMar 17

@karlauerbach Oh, very interesting! Yes, so much of the "deep space" work is around incredibly long latency times.

For MANY years, the focus has been around "delay-tolerant networking" (DTN) and the BUNDLE protocol - more of a store-and-forward architecture.

A lot of the new work is looking at using the QUIC protocol and some of its properties to extend the #Internet into #space.

#IETF #IETF125

Show threadKarl AuerbachMar 17

@danyork Some decades ago someone suggested what seemed to me to be a rather useful paradigm shift when looking at protocols.

We tend to think in terms of streams - the main property being "sequence". (With the main difference between streams and file transfer being whether what we we doing was a potentially never ending data channel versus something with beginning/end demarcations, like files.)

The alternative way was to think of a protocol as a means to what amounts to memory paging - the main interest being blocks that have position relative to others but that the transfers need not be ordered - whatever order things arrived was OK.

Didn't old Multics sometimes use this latter approach?

Show threadThierry DumontMar 17
I remember there is an IP over Pigeon (something like that). At the origin, this was a gag. But, may be it could help, šŸ˜œ @danyork
Show threadDan YorkMar 17

@ThierryDumont Yes! Thatā€™s the April Fools RFC ā€œIP over Avian Carriersā€ originally defined in RFC 1149 - https://en.wikipedia.org/wiki/IP_over_Avian_Carriers

The only challenge is that weā€™d have to figure out how to give all the pigeons space suits. And I donā€™t know how well they would really be able to fly in space. šŸ˜ƒ

IP over Avian Carriers - Wikipedia