Re: Interplanetary communications protocols
From: Tom Kent (teeks99stuff_at_yahoo.com)
Date: 01/19/05
- Next message: Earl Colby Pottinger: "Re: Huygens probe terrain generation"
- Previous message: Cameron Dorrough: "Re: Dual-mode SCRAM/conventional jet possible?"
- In reply to: Ami Silberman: "Re: Interplanetary communications protocols"
- Next in thread: Mark Adler: "Re: Interplanetary communications protocols"
- Messages sorted by: [ date ] [ thread ]
To: sci-space-tech@moderators.isc.org Date: Wed, 19 Jan 2005 21:35:29 GMT
"Ami Silberman" <silber@mitre.org> wrote in
news:csmdul$os$1@newslocal.mitre.org:
>
> "Henry Spencer" <henry@spsystems.net> wrote in message
> news:IAIuML.C7s@spsystems.net...
>> In article <1105902023.240764.257410@z14g2000cwz.googlegroups.com>,
>> <dave.harper@gmail.com> wrote:
>> >I was wondering what communications protocols have been used in the
>> >past with probes traveling beyond the earth-moon system?
>>
>> The CCSDS protocols <http://www.ccsds.org> are essentially universal
>> in deep-space missions, although there is some interest (mostly
>> outside the traditional deep-space organizations) in TCP/IP as an
>> alternative. --
>> "Think outside the box -- the box isn't our friend." | Henry
>> Spencer
>> -- George Herbert |
> henry@spsystems.net
> Blech! I once assigned my Operating System class an assignment to look
> at changes that would need to be made to TCP/IP to be used over
> interplanetary distances. It makes a radio network look pretty awful.
> A cursory glance at "Next-Generation Space Internet: Prototype
> Implementation" from the ccsds site leads me to believe that the
> authors propose to use TCP/IP for communication among and to/from a
> constellation of NEO satellites. (Currently there are applications
> that do use a modified TCP/IP over satcom.) Fundamentally, it isn't
> much different in concept that running it over a radio net -- you have
> less bandwidth than in the wired world, and much longer end-to-end
> delays.
>
> I don't see the point, however, for deep space missions. TCP/IP adds a
> lot of overhead, mostly to manage routing and reliable delivery. Deep
> space missions are essentially fixed to point to one of a small set of
> fixed points, and the long time delay really prevents a sliding window
> acknowledgement-based protocol. Ideally, you just want the ability to,
> if you really have to get some data you missed in transmission, to
> request the probe resend it. The only acknowledgements per se you need
> are those from the probe in response to commands sent from earth.
>
> One place that TCP/IP might be useful in a deep space mission is if
> you have a constellation of sub-probes. Possibly a situation like
> Huygens where one probe is acting as a relay for another might benefit
> from a stream-based TCP/IP connection between the two probes, but for
> communication over interplanetary distances it is probably a very
> suboptimal protocol.
>
>
>
I agree that TCP is a crappy solution for extreme distances, but how
about UDP over IP? This was the DSN could be more like a router, some
controller at mission control would send and recieve messages to/from an
IP address over the internet, the routing tables would eventually end up
at the DSN which would use their (PPP type? or whatever) modem for the
last leg. I like the idea of using standard internet protocols, that way
controllers can use tools that have been developed for other purposes,
but people designing for earthly IP protocols have a very different set
of requirements, so its not surprising that things don't work great.
- Next message: Earl Colby Pottinger: "Re: Huygens probe terrain generation"
- Previous message: Cameron Dorrough: "Re: Dual-mode SCRAM/conventional jet possible?"
- In reply to: Ami Silberman: "Re: Interplanetary communications protocols"
- Next in thread: Mark Adler: "Re: Interplanetary communications protocols"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
|
