you know, I keep thinking about this.
With the advent of SIDI, fuel isn't added until the compression cycle, and only directly into the combustion chamber. This reduces the chance of pre ignition, should the inlet gasses be oxygen rich.
SIDI would also reduce the chances of pre ignition further, as after the exhaust cycle there is very close to zero unburnt fuels remaining in the cylinder to support pre-combustion.
What would happen if we controlled the O2 content by directly injecting LOX into the combustion chamber after the fuel was injected SIDI style ? I think there would be limited inter cooling effect so that big bonus might dissappear.
Intercooling by lox injection in the head might work. Directlly port inject LOX as close to the start of the compression stroke as possible, to limit cold O2 going out into a hot exhaust system. Flame front propagation would be much faster, so you could retard the timing quite a bit - in fact it would be necessary. Pre ignition could be dealt with if the incoming charge was cool enough, and the LOX injection was late enough.
Still it's an awful lot of engineering to achieve a similair result to NOS.
I stil think that small amounts of oxy would be interesting. Increase percentage in the atmopsphere by a couple of percent, retard the timing and you should be ok, in a port injected or SIDI engine.
The problems of O2 injection really only become insurmountable (at least to my mind) when you have very rich burns, over advanced allowing the flame front to actually propagate past the valves, or pre ignite mid compression stroke.
Oxy doesn't cause combustion, it's required to support it. More oxy will make all fuels burn faster and more completley. Just have to match the burn characteristics to the engine.