After a long search I did find papers claiming a
finite KK tower or
KK tower with irregular spacing. The finite tower is on a "fuzzy sphere", the irregular towers are obtained through special boundary conditions. And here is a reference on
KK towers on general manifolds.
Let's suppose you want to get fermion generations as the three lightest massive states in an irregular KK tower. What is the Higgs, in this picture? In the standard model, fermion mass is obtained from a yukawa coupling between left-handed particle, right-handed particle, and Higgs condensate. The
higgs-tau coupling is now being measured. So it looks like your model will separately need a scalar that couples to the KK modes in proportion to their mass. Well, maybe you can build it out of the fermions themselves, by using some kind of
four-fermion interaction.
You also have to worry about "charged lepton flavor violation" - e.g. a muon emitting a neutral Z boson and becoming an electron. How is the Z supposed to discriminate between the KK modes with sufficient precision, that they can play the role of different flavors? Find or impose an
A4 symmetry in the interaction, and
maybe you could solve part of that problem.
Such difficulties would be why people don't try to do this. Here is an
unusual paper which does seek a kind of KK explanation for the generations, but deals with mass in the orthodox way. The extra dimension is "apple-shaped", with a cusp, and produces three
massless KK modes. These then acquire mass through a version of the usual Higgs mechanism, and their higher excitations are posited to all be far above standard-model scales. This is also how things work in string phenomenology - the observed particles all come from massless states of the string.
I haven't managed to understand the finite KK tower of the fuzzy sphere paper.