As a fine first taste of that, check out this overview of the Android multitasking environment, which briefly examines how the process model works and why they made the choices they did. Very neat stuff, and a good design. Frankly, they attacked the problem the right way, asserting that they were going to build Google's own apps "fairly", without resorting to the sort of internal OS backdoors that Microsoft and Apple are prone to. As a result, they had to build an API rich enough to do everything they wanted -- so all apps have access to the same power that Google's own ones do.
(Mind, the process model is distinctly weird, and you have to build your app within its assumptions. But it's a very sensible approach for a realtime environment with varying priorities and hard memory limits.)
