How can I get all actors in a camera's view frustum?

I guess I probably will stick to my current implementation unless performance problems arises when the number of “units” increases.