Prevent IEEE access restrictions - Don't use option 1
The IEEE's ability to legally restrict access is based on having exclusive rights, which they take from you by strong-arming you into signing those away.
The solution is using the standard trick: to actually grant the relevant rights beforehand to someone else - without losing them yourself - so that when you sign the IEEE copyright form, there will be someone else to give you the rights back afterwards. That's why you should choose one of the last three options, never the first one.
Personally, I would tend not to restrict commercial use from the get-go. It's tempting, but when you think about it - the social damage from having to go through you for economic use of your findings is probably higher than the benefit of sticking it to the large corporations or what-not.
As for public domain - I'd be against it, since attribution is important not just to you but also for academic/scientific work. So, wanna use my paper? Citation please.
Non-SA licenses: I would be worried about derivative work being non-free.
To be honest, though, this isn't a strong recommendation and all three are fine by me I guess.