I dont know if this is applicable to you, but I have been contemplating the same thing and wondering how to handle.
Bit of background 1st. Back in 3rd edition days I tried a big battle...it went for weeks, and didnt really keep the players interest the whole time. I wanted it epic, and made it too epic : Too much to do, too much to keep track of, too much to go wrong. So building on that I came to the conclusion that I didnt want to try to emulate a tactical wargame again.
Rather, this time around, I am choosing to represent the battle as a series of skill challenges (Note : We have a kick arse skill challenge system with supporting software

) whereby I will define what is going on in the battle with possible events. Players can choose how to contribute (someone might want to use a heal skill in the triarge, someone else might want to work on inspiring the troops, someone else might want to get into the thick of it).
Based on this dynamics are defined (e.g. How many troops remaining at the north tower?) and influenced by player skill contribution. Triggered events then lead to further skill challenges (e.g. The number of troops on north tower gets to zero...north tower overrun) or even(/inevitably) to pre defined (limited scope) combat encounters. For instance, troops cant really cope with the trolls smashing through the front gates, its up to the heros to seal that breach.
From my position, the point is NOT that troops are litteraly represented as markers on a map, but that the battle is followable (i.e. the current state of play is transparent to the players) and transitions from state to state, and that the players have a real and understood ability to contribute to the result and influence the outcome.
I figure this way I can represent a battle and still be true to the game the
4e actually is.