Cool. This thing is also closely related to Bussard's Polywell fusor. If you have a loose idea of how that works, it's easy enough to make a connection.

I think one of the tricks to it is to make an arc discharge loopback on itself in the presence of a strong magnetic field. Not obviously easy to do from the beginning. (Sometimes an eddy current during a discharge can cause this to happen randomly, but you want it to happen consistently.) One part is probably something like an arc-welder that pulses DC. (Removed from the document, but seems plausible enough.) If you can get the looped current flowing in a path like a string wrapped around a donut, the induced magnetic field should make it semi stable. (The magnetic field produced by the current flow in the plasma keeps it going in a loop.) The pinch point at the center of the plasmid might also be hot enough for fusion to occur.

Now imagine having that little "fusion core" ejected from its reactor by another magnetic field so that you can use it as a bullet. I'm sure that's why the military sees some use for that, particularly if it's scalable.