Well, you might have mechanics for facing, or a "combat focus" that requires the player to declare that his attention is focused in one place (at be at a mechanical disadvantage against his other opponents).
That's really beside the point I'm making, though. The sort of awareness I'm thinking of modelling is instinctive (and developed with increasing skill), and doesn't rely on "focus" or "facing". Indeed, I really don't think "facing"
per se is meaningful at all in melee combat - either your perception is on your entire surroundings or you're dead meat. Being engaged by multiple foes has an effect, for sure, but never do you choose to focus only on one of them. I actually think the "flanking" system brought in by 3.5e D&D or the "-10 per opponent above one" of HârnMaster are the ideal solution, there.
You could create a deeper action economy that requires players to make more detailed choices between guarding themselves, helping others, moving or attacking.
You could, but many seem to cry foul at the complexity of the action economy already. Besides which, the sort of perception I'm talking about operates outside the "action" system; if it takes an action to instinctively note an opportunity for a particular type of manoeuvre, the action system is getting pretty "dissociated"...
You could impose tighter dice rolls that always leave a chance of failure (rather than the 3.5 rogue that never fails tumble after about level 2).
The point is that there should be times when the chance of success is minimal, and there should be times when the chance of success is pretty good - and a skilled character should know the difference instantaneously.
You could then design character abilities that make skilled combatants better able to focus their attention on multiple enemies in their area, or less likely to fail certain checks, or grant them more power within the action economy.
All very nice - but missing the point that what (I think) they need is the ability and necessity to make timing decisions based on a constantly varying "option space" that they can perceive and are thus aware of. In other words, the opportunity to use some moves should come up relatively seldom, and even more infrequently when considered useful by the character, but the character (and, by extension, the player) should be aware of those opportunities with minimal effort such that they are able to exploit them should they choose to do so. If they are spamming the ability because it always works, or ineffectually trying it repeatedly because it hardly ever works, then the system has seriously missed its mark. A skilled operator will attempt the moves only when they are likely to work - but that will not be all the time.
The power concept models none of that stuff.
It models making timing and opportunity decisions, which is more than any other system we have so far does. It's not perfect, I'll agree - but at least it makes an actual attempt to model this aspect of skill.
Well for D&D I would certainly do it a bit different , but in my own games the way I do it is make maneuvers conditional on simple things like whether a foe did damage, hit but didn't do damage, whether you hit the foe, whether you closed in on the foe this round, etc. it definitely comes a bit more from my boxing background and again I would do it differently for D&D but for my game it works pretty well (though I am revising and play testing alternatives for the next variation).
That makes it sound like it's mostly "attack" manoeuvres you are modelling - is that the case? Can you give any examples? I can see some options, here; the rogue's "tumbleset", for example, which can move past enemies that are engaged by at least one other ally without attracting opportunity attacks, making the precise positioning on the battlegrid do the work of generating the "opportunity", but it's not going to work for all cases.