The ISS is a moving target that orbits the Earth about every 89 minutes. To maintain a constant data link would require several TDRSS-style satellites out at geosynch. The RF data transfer rate would be limited by the ISS antenna size, which is limited by physical factors like clearance with the solar arrays and other ISS structures.

Laser data relay is certainly viable but not necessary. A old DSCS-III satellite is capable of relaying very high data rates using ordinary antennas. Commsat technology has advanced considerably since the DSCS satellites were designed. A satellite optimized for high speed data relay, such as the ones used for DirectTV and Dish network, routinely transmit many megabits of data per second. Since those satellites were designed to cover large geographic areas, their downlink antennas aren't optimized for this particular mission. A spot beam antenna on this satellite could easily gain several db of gain allowing for much higher data rates due to the higher EIRP. This is off-the-shelf technology available today. A satellite at geosynch wouldn't even need this kind of data throughput using simple RLL compression. A reasonable CCD resolution, say 10 megapixels, with 256 gray scale depth, could transmit an uncompressed individual image in just 40 seconds at a moderate 250 kilobyte per second data rate. 250 KB/sec is only 30% faster than T-1 data rates, a trivial number with today's technology. With typical RLL compression, the image could be transmitted in less than half that time without losing a single bit of data.

Building a space-rated high performance computer to put on the ISS makes no sense at all. It's far more practical to simply transmit the data to the ground for processing. On the ground, you aren't limited by the volume, electrical power, and heat dissipation limitations of the ISS. On the ground, you can use the fastest processors available without having to deal with SEUs, either, nor do you have to go to the expense of qualifying the computer for use on the ISS. You're not limited by antenna size, either. That factor alone would more than compensate for atmospheric losses.

Any whole sky search telescope has to have a good field of view. It does very little good if your field of view is so narrow that you can't return to a particular piece of sky in a timely manner. It'd take close to 1000 images to do a spherical scan at 3 degrees FOV. If you allow only a minute for each scan phase, then you're able to scan that part of the sky once every 16.67 hours. If you go to a 1 degree FOV, it'd require many times as many images to scan the whole sky and consequently the rescan frequency is much lower. As a result, a moving object might not be detected as readily because it could've moved into an adjacent frame, thus failing the XOR detection. The only way to circumvent that would be to use higher overlap between adjacent images which only serves to complicate the processing.

Ed, I know you're in favor of a lunar based observatory but it just isn't necessary and isn't likely to happen due to the extraordinary cost. I'd be willing to bet any amount of money that a lunar NEO observatory will never be funded by Congress in my lifetime. The perfect (and it isn't perfect) is the enemy of the good, Ed. Your obscession with a lunar observatory seems to be blinding you to reality - political, economic, and technical. Why are you so obscessed with a notion that would cost untold billions of dollars and wouldn't be available for decades? It simply makes no sense.