Its very unlikely a planet-killing asteroid will be spotted on final approach to earth and so that there will be enough time after its discovery to improvise a defense for it. What such a system could do, however, is build up an extensive catalog of orbital elements for large numbers of solar-system objects and that simulations and modelling of these bodies will allow us to predict their orbits years ahead of time and which objects may be a potential threat in the future. In other words, we will have years ahead of the collision to develop a technology to mitigate the effects of the collision, or avoid it all together.

In addition, such a system could monitor all those bodies that are not currently considered a threat and detect any changes in their orbits (caused by perturbations from other bodies) that might suddenly cause them to become a threat.

I envisage a suite of telescopes and radars coordinated by a permanent institute and dedicated staff, all feeding data to massive computers. Projections of potential encounters could be made decades ahead of time, and there would be plenty of time to be able to devise a planetary defense, whether it would be destroying the object, or perhaps nudging it into a safer orbit. This project would have the facilities to detect and map these bodies, and if necessary, communicate with other observatories to gather more specialized observations. This institute would publish bulletins which could then be scrutinized for potential collisions, and prioritize their risks. It would also act a clearing house for follow up and coordination of other observations collected at other observatories engaged in other types of research.

The solar system is a dynamic place, things are constantly changing and interacting. The good news is that we would have enough warning to do something about it. All the methods currently conceivable of protecting us from collisions are usually much easier if done years ahead of time, instead of rushed at the last minute. This facility could also be designed so that other types of research could be done there which could help defray its cost and extend its usefulness.

I don't visualize a big telescope or radar, but several of them, with links to other, existing instruments and institutions, permanently funded for the long term. It would operate like NOAA, or the USCGS, or the Weather Service--except in space.