We study the robust link scheduling problem (RLSP) based on a physical interference model with errors in channel state information. The objective of RLSP is to find a robust minimum length schedule using spatial time division multiple access. We compare two approaches to RLSP, one using channel gain estimates and the other using location information. In both cases, we formulate the RLSP as a binary integer program and solve it by a classical column generation technique. Our comparison reveals that both approaches yield similar performances, but with different overhead.