Forget about EJB and Struts.
Both are overblown overly complex systems which won't work well for your problem domain (EJB might get interesting if you're planning to seriously scale up and use a distributed server architecture storing the data in an RDBMS).
JBuilder is a good choice for an IDE, it's decently fast and easy to use.
It's of course not the only choice but as others have noted there's about as many preferences for IDEs as there are programmers out there.
Leaving IDE choice up to the individuals is nice but in many environments not an option ("corporate standards" etc.).
Patterns are integral to successful development, but beware of using them for their own sake.
Using patterns just to use patterns (which happens a LOT in applications designed up-front) can lead to disaster as the code tends to get more complex than needed.
MVC is just another pattern (or a compound pattern really) that is recognised as being at the core of writing flexible maintainable code these days.
What tech to look at?
Tech should be led by requirements, not the reverse
You're looking at 2D graphing, which suggests Java2D or maybe some API to generate graphs from numerical data as bitmapped images (probably something like that exists that you can get either for free or commercially, don't reinvent the wheel without good reason).
If you want the data visualised on screen, you'll need some sort of GUI. Swing is a natural choice for that.
Communication between the machines you're using might be simplest using RMI or object serialization over network sockets.
You're going to need multithreading to ensure the machines are responsive enough to talk to each other.
You might want to think about building the actual number-crunching and data-acquisition code as native code in C(++) and calling it using JNI but that's up to you (you might do it in Java first and see if that's fast enough, quite possibly it is).
Your team of 5 could well split responsibilities on development.
Split among the lines of networking/communications, user interface and data processing and assign 2 people to every part (primary + backup) with yourself having a secondary role as system integrator taking charge of the interfaces between those application tiers.
Don't overdesign, rather set guidelines and let them do their thing as long as they can defend their choices.
