certi-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [certi-dev] Weather in HLA simulations?


From: Eric Noulard
Subject: Re: [certi-dev] Weather in HLA simulations?
Date: Tue, 29 Apr 2008 15:59:07 +0200

2008/4/29, Gotthard,Petr <address@hidden>:
>
>
> Hello,
> another question: does anyone have ideas concerning weather models connected
> to HLA?

No I have no experience in that domain.

> I found two bad solutions:
>
> ** Define weather as an object which provides a VERY huge matrix which
> defines weather for the whole world. Other objects would subscribe for this
> information with a region, so only relevant information would be delivered
> to the objects.
> Disadvantage: It would require RTI to store the VERY huge matrix that change
> in time as the weather changes.

RTI will have a great task to handle UAV in this scheme.

> ** Use interactions between the weather and the objects. The objects could
> use an interaction to "request" weather data for specific coordinates.
> Disadvantage: Its not very natural to interact with weather. It may also
> generate unnecessary overhead.
>
> Do you have any ideas you could share regarding this issue?

If you want to "Request" information you may use "RequestAttributeValueUpdate"
and not Interaction.

I do not see why the interaction solution is not that natural?

Your weather model will CreateRegion
then SendInteraction with Region.

Your objects will SubscribeInteraction with Region

I think its pretty "natural" since the objects knows which Region is of interest
for him.
Weather object may loop over region and SendInteraction embedded
needed informations.

The RTI should distribute Interaction only if there is a subscriber which
should not generate too much overhead (this should be benchmark though).

I don't know how CERTI would handle this scheme since I did
not use Region in my simulation but theoretically this could be done.

On the performance-side the number of region will certainly influence
the RTI performance.

The problem I see here is that Weather cannot know which region
is of interest to somebody such that it may compute information for nothing
because there is no subscriber.

You may use both approach (Regionned UAV + Interaction) if object
send an interaction when they "enter" a region and another "one" when
they leave. this way Weather may know which region should be UAVed.

These are just ideas since I haven't done such experiment.

-- 
Erk




reply via email to

[Prev in Thread] Current Thread [Next in Thread]