We outline the first steps of an effort to start defining the communication architecture for the next generation of space missions that will support NASA's "faster, better, cheaper" concept and will enable new types of collaborative science, where investigators can access their data from space "anytime, anywhere" via direct communication with the instruments on the spacecraft.

We discuss the building blocks for a conceptual design of a network architecture that could support and take advantage of IP-capable spacecraft.

We show that access from a large number of ground stations (that could be directly connected to the existing Internet infrastructure) could increase spacecraft availability time by a significant factor.

We discuss possible multiple access techniques that could enable the transition to an on-demand operation, where spacecraft share space spectrum dynamically. We calso discuss the particular requirements of a next generation of missions consisting of constellations of several small spacecraft and introduce a number of new complex network control, scheduling, routing, data management and communication problems that need to be addressed for this topology.