Assuming it's still the old course correction stuff:

1. radar tracks projectile/rocket
2. computer predicts miss, calculates required correction
3. radio tells projectile to brake
4. projectile begins to brake at perfect time to minimize range error (creating a roughly circular dispersion; dispersion left/right is usually smaller than in range for unguided munitions)

Sorry, don't remember the correct arty terms for "range" & "left/right".


The simple correction for follow-on munitions has been used with modern rocket artillery afaik since at least the 70's and merely makes registering unnecessary (the first rocket can be set to timed self-destruct in order to maintain surprise for the full salvo).