From a DM perspective, I would prefer it to be a simple contest. You can either go with 4e's "passive perception", or you can make it an opposed check. In either case, all that matters is that the player rolls well enough to overcome the opponent's bonus, but it should never be one roll against seven; doing so only gives stealth an inherent disadvantage.
The benefit to a simple target DC is that it's very flexible, without punishing the player because even tho Guard 1, 2, and 3, didn't roll above a 10, Guard 4 rolled a nat 20. You could easily have guidelines that you could improvise. For example:
Set a base target DC of 10 to sneak past a single guard. Apply +/-2 modifiers based on conditions. Add +2 for each additional guard watching, +2 for each ideal conditions (i.e. well lit), and then add the highest Wisdom bonus amongst the applicable guards. Add a -2 for each condition that works against the guards; i.e. heavy shadows, heavy winds/noise.
This allows both the players and the DM to get rough sense of how difficult/easy a task it is to accomplish. The DM will of course, know the true DC, but at least the party rogue can size a situation up and tell his comrades with some amount of confidence whether sneaking by is likely to succeed.
If you, as a DM, wish to declare that one guard in particular is watching like a hawk, and another is taking drinks from a pocket flask, great. Don't apply the +2 bonus for the drinking guard (as he's not paying attention), and give the guard watching like a hawk a +1 bonus. If the players manage to distract/take out the hawk guard, they reduce the difficulty of sneaking by 3 (or possibly more, if that guard had the best wisdom bonus).
In short, you can model a lot of different scenarios with a simple DC, without creating an overall high level of complexity (as seen in most stealth rule sets).