Open Source OPNET alternative
James Neushul
Miles Fidelman
> Does anyone know of an OSS alternative to OPNET?
There's a lot out there if you google "open source" plus "network
simulator" - it's been a while since I've worked with network simulation
tools, so I don't want to comment on the current state of tools.
There's some good commentary at:
http://www.openxtra.co.uk/articles/network-simulation
There's also been a good amount of network simulation done in
MATLAB/Simulink - and since there are a number of open source MATLAB
alternatives, I expect that there are open source network models for them.
The thing is, it's one thing to look for a general modeling tool, it's
quite something else to apply it to specific situations - at that point
you're looking for a modeling tool that is particularly good for the
application at hand, and probably some existing models that can be
tailored to your situation.
"We are trying to figure out how to network very operator on the
battlefield for data and PLI" is just a little broad (and isn't
everybody trying to do that, or claiming they have a solution? :-).
If you can be a bit more specific about the models you're trying to
build, what level of details you're trying to model, and what questions
you're trying to explore, I have several former colleagues who might be
able to point in an appropriate direction - and I'd be happy to dig a
little if you forward me some details privately (after which we can
summarize to the list).
Miles Fidelman
> We tried going with DISA JCSS (Joint Communication Simulation System)
> to be on the same page with everyone - and thinking that DISA was
> doing something right ....
> Then we found out that to do DES we need to buy OPNET licenses. The
> vendor lock-in tendrils reach deep. Why is JCSS fronting for OPNET?
Procurement vehicle? Just like C/JMTK is a procurement vehicle for
ESRI's GIS software.
--
In theory, there is no difference between theory and practice.
In<fnord> practice, there is. .... Yogi Berra
JDN
The "OPNET" reference is a hopeful cast for stuff that lets us use
existing models for tactical radios that have been built for OPNET.
We have the Harris models for the radios that most Marines have - and
that we plan to employ. I think there are joint requirements for the
vendors to produce these. Unfortunately i don;t see a lot of
acquisitions decisions driven by modeling - only testing after the fact.
It would be great if someone could prove me wrong on this score.
Basically - we have a long range "wimax' type waveform for a high
throughput "backbone" type connector. We want to network in (just using
IP) another kind of wideband tactical network for short/medium range,
and then yet another WIFI net for very short range disposable handhelds
at the edge.
We are hypothesizing that if we could do this -- what kind of
software/network architecture would allow the kind of reporting we need
to do. This is part of the "Combat Relevant PLI" effort which is
predicated by the "Tactical Data Network Extension" concept.
Oops - just saw the "privately" part. Nothing sensitive here. Just the
normal noises.
I will send you some drawings and more details offline.
Thanks Again!
Neutron
Likely we will want to model wifi communication between hand helds as well.
Kyle Harrigan
Maybe not much for detailed military radio models, but for wifi/wimax
type waveforms probably some good stuff.
Miles Fidelman
> The "OPNET" reference is a hopeful cast for stuff that lets us use
> existing models for tactical radios that have been built for OPNET.
>
> We have the Harris models for the radios that most Marines have - and
> that we plan to employ. I think there are joint requirements for the
> vendors to produce these.
If you have to use models written for OPNET, then you are most likely
stuck with using OPNET, or finding a way to translate the models. In
general, these things can be pretty application-specific, unless there
were an open source OPNET clone - which I haven't seen.
On the other hand, OPNET seems to be available at no cost to you via
JCSS - you might have to bite the bullet and use a proprietary package
this time around.
Having said that, I did put out some queries, and if I hear anything,
I'll let you know.
> Unfortunately i don;t see a lot of acquisitions decisions driven by
> modeling - only testing after the fact.
Yeah.... funny thing about that (says the guy who used to work for a
modeling and sim outfit).
> Basically - we have a long range "wimax' type waveform for a high
> throughput "backbone" type connector. We want to network in (just
> using IP) another kind of wideband tactical network for short/medium
> range, and then yet another WIFI net for very short range disposable
> handhelds at the edge.
>
> We are hypothesizing that if we could do this -- what kind of
> software/network architecture would allow the kind of reporting we
> need to do. This is part of the "Combat Relevant PLI" effort which
> is predicated by the "Tactical Data Network Extension" concept.
>
> Oops - just saw the "privately" part. Nothing sensitive here. Just
> the normal noises.
>
> I will send you some drawings and more details offline.
When I see it, I'll take a look - it might be that you can run part of
your model in OPNET, and develop your network level stuff on an open
source platform. But if you really are just doing a one-off to support
an acquisition, OPNET might be your best bet.
Miles
Miles Fidelman
I asked around about open source platforms that can run OPNET models,
and this is what I got back - from a former BBN colleague, turned
university professor, who works on things like Airborne networking for
the flyboys:
"I seriously doubt it. This is one of the reasons we use ns-3."
Nothing from anybody else.
Looks to me like your choices are to reimplement your models in
something like ns-3, or hold your nose and take advantage of the fact
that DISA has already paid for OPNET licenses.
Best,
Miles
Neutron
JCSS only provides terrain and network loading. To do DES we need to
buy licenses. I am just not able to by stuff in my current job.
We only need to do 2 radios (AN/PRC-117G and RT-1944 Sea Lancet) along
with handheld short range wifi.
I would MUCH rather do this with ns-3 .. but the radio models are for
JCSS. It is good to know that there is an OSS solution out there.
Maybe after we get rolling we can start some OSS modelling efforts.
Modelling is, of course, technically required for DOD acquisitions.
Neutron
Miles Fidelman
> JCSS only provides terrain and network loading. To do DES we need to
> buy licenses. I am just not able to by stuff in my current job.
Figures. I knew they had to make their real money somewhere.
>
> We only need to do 2 radios (AN/PRC-117G and RT-1944 Sea Lancet) along
> with handheld short range wifi.
>
> I would MUCH rather do this with ns-3 .. but the radio models are for
> JCSS. It is good to know that there is an OSS solution out there.
> Maybe after we get rolling we can start some OSS modelling efforts.
> Modelling is, of course, technically required for DOD acquisitions.
I wonder what it would take to re-implement those models (speaking as
someone with no experience whatever in building radio models).
Interestingly enough, I found a 2010 SBIR topic that focused on re-using
radio models from across simulators - you might find this interesting:
http://www.dodsbir.net/sitis/archives_display_topic.asp?Bookmark=40327
(might be worth making a few calls - maybe somebody over there might
have some ideas, or some models).
Miles Fidelman
>
> Interestingly enough, I found a 2010 SBIR topic that focused on
> re-using radio models from across simulators - you might find this
> interesting:
> http://www.dodsbir.net/sitis/archives_display_topic.asp?Bookmark=40327
> (might be worth making a few calls - maybe somebody over there might
> have some ideas, or some models).
>
I should add - the topic includes some references that point to groups
that might have some ideas/pointers/models/...