I strongly suspect that any cryptographically suitable problem derived from finding the intersection points of external tangent lines would reduce to the closes

I strongly suspect that any cryptographically suitable problem derived from finding the intersection points of external tangent lines would reduce to the closest vector problem on lattices.