A single narrow-band antenna must have at least one null direction in which there is no coupling with the other antenna. The link in the OP's post is to ~two~ UWB antennas driven separately. If they are driven/received correctly (notable time difference between pulses, etc.), then there will be no null direction. There will still be some variation in gain per direction. You'd need a lot of separately driven (with UWB) antennas to make an "antenna" that has close to the same gain in all directions. Technically, two separately-driven antennas is not an "antenna" though, in the context of isotropy. Also, you can take advantage of knowledge of the current link to drive/receive the two antennas separately. For example, you could put all your energy into the antenna with the best coupling.