Disclaimer:  I've only used variations and hybridizations of RUP.

I think you may have it backward.  Each iteration does not have 4 phases.  Rather, the project has 4 phases and each phase has iterations.   You do your initial iteration planning in the Inception phase.  If you google something like "RUP iteration planning," you should get some results that describes the formal methods and will probably find some iteration plan templates.

Your iterations should be based on a few considerations:

• Use cases or components with highest architectural risk are ususally included in the first iteration, so that any challenges or problems can be identified and addressed as early as possible
• Each iteration builds upon the development of the previous iteration(s). If 'Function B' is dependent upon 'Function A', then 'Function A' should be developed first and 'Function B' should be assigned to a later iteration. For example, presumably flights need to be scheduled before a passenger can make a reservation. So you would probably want to build the scheduling components first (if in-scope). Then when the booking component is built and tested, there will be working functionality to setup and maintain the schedules - which makes it much easier to test the booking functionality.
• Core and common functions should be developed first, so they can be refined and finalized early in the project. This minimizes impact and re-work if there are errors in requirements or development for this pieces.

Note that it's probably not possible or desirable for all iterations to be 'identical subsystems'. In fact, the essence of RUP iteration is that early development gets elaborated and refined. So each iteration carries forward some re-work and refinement of previous development. Therefore the scope of new development often gets gradually smaller over each iteration, to allow for carried forward refinements and bug-fixes from prior iterations.

Sandy