Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Dale Stanbrough]

Robert Munck writes:

"The satellites will be about 1000 Kg (2400 lb) each [...]
 This means a whole lot of software on each (perhaps 20% of
 the total weight)..."

Wow, that's really heavy code!

Dale

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Kaz Kylheku]

In article <3368134a.21286600@news.mindspring.com>,
Robert Munck <munck@acm.org> wrote:

>low-level control from ground stations!  This means a whole lot of
>software on each (perhaps 20% of the total weight) with tremendous
>reliability, security, and throughput requirements.

I wasn't aware software had that much mass! :)

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Dale Pontius]

In article <5kb77u$89@bcrkh13.bnr.ca>,
        kaz@vision.crest.nt.com (Kaz Kylheku) writes:
> In article <3368134a.21286600@news.mindspring.com>,
> Robert Munck <munck@acm.org> wrote:
>
>>low-level control from ground stations!  This means a whole lot of
>>software on each (perhaps 20% of the total weight) with tremendous
>>reliability, security, and throughput requirements.
>
> I wasn't aware software had that much mass! :)

Not only that, but software is really information. That means you
can assign an entropy value to it, which also means you can assign
a temperature to it, as well.

Hot now software takes on a new meaning.

Really, I suspect the 20% of weight they're talking of is, of
course storage media, be it RAM, ROM, or tape. But what may have
really happened here is that a threshold has been crossed. Perhaps
the storage technique that worked with less sophisticated (or
bloated?) software has become insufficient. Perhaps the issue is
no longer "more of the same" storage, but a new hierarchy or new
media.

Is it possible to use a hard drive on a small satellite, or will
the rotating platters foul things up. I'd expect that the shuttle
is so big that the effect is insignificant.

Dale Pontius
(NOT speaking for IBM)

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Nick Roberts]

Dale Pontius <pontius@btv.vnet.ibm.com> wrote in article
<5kcmkd$f3q$1@mdnews.btv.ibm.com>...
[...]
> Is it possible to use a hard drive on a small satellite, or will
> the rotating platters foul things up. I'd expect that the shuttle
> is so big that the effect is insignificant.

A single hard disk drive on a satellite would tend to cause the satellite
to rotate significantly, unless it was spun up once and never spun down
during operation. Two identical drives mounted back-to-back would solve
this problem. However, they would have to be rigged so that if one failed
and spun down, the other would be spun down also.

Conventional hard disk drives would be unsuitable for a satellite, because
they are not sufficiently immune to the vibrations caused during the launch
into orbit. They would also be prone to failure in space, if kept
continuously spun up. Hardened versions exist, but are expensive. I suspect
that FLASH RAM (or similar) would pose a better solution.

The software controlling the satellite's ability to contact its base
control station would have to be highly reliable. Ada would be less
suitable than an 'artificial intelligence' language, such as Prolog, LISP,
or other, for this section of the software, since it would be best to endow
the software with some rudimentary 'intelligence', so it could use
rules-based induction to solve certain problems for itself.

The remainder of the software would not necessarily have to be very
reliable (depending on what sort of communications it was being used for).
I would suggest that the reuse of existing software (such as all that
TCP/IP and related stuff) would be very important (most of which is written
in 'legacy' languages).

But, for some of the software, Ada is likely to be a good choice.

Nick.

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Robert Munck]

On 2 May 1997 23:59:55 GMT, "Nick Roberts"
<Nick.Roberts@dial.pipex.com> wrote:
>... Ada would be less
>suitable than an 'artificial intelligence' language, such as Prolog, LISP,
>or other, for this section of the software, since it would be best to endow
>the software with some rudimentary 'intelligence', so it could use
>rules-based induction to solve certain problems for itself.

See, I'm not the only one to put little jokes in their postings.
Of course, the idea of "rules-based induction" in a sattelite
comms system is a bit more outre' than my suggestion that
software has mass.

Unfortunately, the only people who will get this joke are 
old-timers who remember when people thought there was
really such a thing as AI.

Bob Munck

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Rennie Allen]

Robert Munck (munck@mindspring.com) wrote:

[...]

> The satellites will be about 1000 Kg (2400 lb) each, orbiting at about
> 720 Km (450 mi).  It appears that data uplinked to one satellite will
> be routed through several others before being downlinked with no
> low-level control from ground stations!  This means a whole lot of
> software on each (perhaps 20% of the total weight) with tremendous
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

I wasn't aware that software had weight; then again if it's gatesware....

--
Rennie Allen                                       <rgallen@qnx.com>
QNX Software Systems Ltd.
175 Terence Matthews Crescent                      (613) 591-0931    (voice)
Kanata, Ontario, Canada K2M 1W8                    (613) 591-3579      (fax)

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Dale Pontius]

In article <3368134a.21286600@news.mindspring.com>,
        munck@mindspring.com (Robert Munck) writes:
> (from Reuters)
>>SEATTLE - Teledesic, the communications startup mostly owned by
>>Seattle billionaires Bill Gates and Craig McCaw, has picked Boeing
>>to build and deploy its planned $9 billion satellite network.
>
>>Boeing also will invest up to $100 million in Teledesic, giving it a
>>10 percent stake in the privately-held company, which plans to launch
>>288 low-orbiting satellites by 2002 to form an "Internet in the sky."
>
> The satellites will be about 1000 Kg (2400 lb) each, orbiting at about
> 720 Km (450 mi).  It appears that data uplinked to one satellite will
> be routed through several others before being downlinked with no
> low-level control from ground stations!  This means a whole lot of
> software on each (perhaps 20% of the total weight) with tremendous
> reliability, security, and throughput requirements.
>
> Could it be written in anything but Ada?  Who will decide?  If it's
> Boeing, what would they choose?
>
Consider the customer... How about Visual Basic or Visual C++?
How about a variation of ActiveX called OrbitX?

Obviously the presence of 288 satellites in a relatively low orbit
won't constitute much of a "fence" between LEO and high orbits. But
if several more outfits decide to do the same thing on the same
scale, does it begin getting difficult to find a decent launch
window to get above LEO? Fortunately 450 mi is about the top range
for the Shuttle, or I'm sure this would give them some difficulties.

Dale Pontius
(NOT speaking for IBM)

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[John Woodruff]

>>>>> "Bob" == Robert Munck <munck@mindspring.com> writes:
In article <3368134a.21286600@news.mindspring.com> munck@mindspring.com (Robert Munck) writes:


    > (from Reuters)
    >> SEATTLE - Teledesic,  ....    lines omitted    ....

    >> Boeing also will invest up to $100 million in Teledesic  
       ....    more lines omitted    ....


    > The satellites will be about 1000 Kg (2400 lb) each, orbiting at
    > about 720 Km (450 mi).  It appears that data uplinked to one
    > satellite will be routed through several others before being
    > downlinked with no low-level control from ground stations!  This
    > means a whole lot of software on each (perhaps 20% of the total
    > weight) with tremendous reliability, security, and throughput
    > requirements.

Wow!   200 Kg of software;  just imagine.

It is a "good thing" (tm) that preferred Ada style no longer prescribes
UPPER CASE Identifiers, as it would be even harder to get the software
down to weight with all those HEAVY characters ..... ;-)

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Robert Munck]

On 01 May 1997 13:02:18 -0700, woodruff@tanana.llnl.gov (John
Woodruff) wrote:
>    > ...  This means a whole lot of software on each
>    >  (perhaps 20% of the total weight)
>
>Wow!   200 Kg of software;  just imagine.
>
>It is a "good thing" (tm) that preferred Ada style no longer prescribes
>UPPER CASE Identifiers, as it would be even harder to get the software
>down to weight with all those HEAVY characters ..... ;-)

Come on! Everybody knows that the source code is compiled into
binary object code, and that the important thing is to maximize the
number of 0's because they weight much less than 1's.

Jeez! I put in one little joke, and everybody makes comments on it
and ignores the main point of the posting.

Bob Munck@acm.org

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Mark & Zurima McKinney]

Robert Munck wrote:
> Come on! Everybody knows that the source code is compiled into
> binary object code, and that the important thing is to maximize the
> number of 0's because they weight much less than 1's.
> 
No MAybe they're going to use some interpreted 4gl thingy :-)

MArk McKinney

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Marin David Condic, 561.796.8997, M/S 731-93]

Dale Pontius <pontius@BTV.VNET.IBM.COM> writes:
>
>Really, I suspect the 20% of weight they're talking of is, of
>course storage media, be it RAM, ROM, or tape. But what may have
>really happened here is that a threshold has been crossed. Perhaps
>the storage technique that worked with less sophisticated (or
>bloated?) software has become insufficient. Perhaps the issue is
>no longer "more of the same" storage, but a new hierarchy or new
>media.
>
    It could also be a matter of "large" software demanding more
    compute power which can demand more or bigger (physically)
    processors. Increases in clock speeds can also translate into
    energy consumption (bigger heat sinks, etc.). It might be an issue
    of increasing the size/complexity of the software causing a whole
    redesign of the computer which increases the weight.

>Is it possible to use a hard drive on a small satellite, or will
>the rotating platters foul things up. I'd expect that the shuttle
>is so big that the effect is insignificant.
>
    When rockets fire, you get *lots* of vibration. Head crash on a
    disk drive on a satelite would be *very* bad! And don't forget
    that spinning a disk platter is going to consume electricity -
    moreso than an EEPROM (unless you had a *lot* of EEPROM!). It may
    be that somebody has figured out how to use a disk drive in this
    sort of application, but it wouldn't be my first choice. I'd be
    real curious to find out about any space-based embedded computers
    that used a disk drive in an unmanned vehicle. (I wouldn't be
    suprised if there were some in the shuttle, but that gets to come
    home rather quickly and breakage there may have less critical
    impact.)

    I doubt the rotation of a disk say the size of one in a notebook
    computer would significantly impact the course or attitude of a
    satelite - even a "small" one. (Even "small" satelites are pretty
    heavy in comparison to the weight of a notebook sized disk.) And
    most satelites I know of contain some ability to adjust their
    course & attitude in flight to compensate for drift from any & all
    sources.

    MDC

Marin David Condic, Senior Computer Engineer    ATT:        561.796.8997
Pratt & Whitney, GESP                           Fax:        561.796.4669
West Palm Beach, FL                             Internet:   CONDICMA@PWFL.COM
===============================================================================
    "It's a small world, but I wouldn't want to have to paint it."

        --  Steven Wright
===============================================================================

Re: Will Boeing Use Ada To Build $9 Bln Network For Teledesic?

[Marin David Condic, 561.796.8997, M/S 731-93]

>Mark & Zurima McKinney <mckmark@US.NET>
>Robert Munck wrote:
>> Come on! Everybody knows that the source code is compiled into
>> binary object code, and that the important thing is to maximize the
>> number of 0's because they weight much less than 1's.
>>
>No MAybe they're going to use some interpreted 4gl thingy :-)
>
    But consider the fact that while the zeros are of lower mass, the
    ones can be packed much closer together because they take up less
    space. Hence you can put more of them into a smaller memory chip,
    thus saving all the weight of packaging.

    MDC
Marin David Condic, Senior Computer Engineer    ATT:        561.796.8997
Pratt & Whitney, GESP                           Fax:        561.796.4669
West Palm Beach, FL                             Internet:   CONDICMA@PWFL.COM
===============================================================================
    "It's a small world, but I wouldn't want to have to paint it."

        --  Steven Wright
===============================================================================