“…, c NT,m ] T ∈ C NT . One can obtain the following matrix model according to (1) (see [15], [22] and [24] for similar models) 2 : In what follows, we assume that h, c as well as w are mutually independent and circularly symmetric complex Gaussian random processes with (zero mean and) covariance matrices R H ∈ C NTR×NTR , R C ∈ C NTR×NTR , R W ∈ C NRL×NRL , respectively, with N TR = N T N R and N RL = N R L. We further assume a Kronecker structure for R W [21], [26], viz.…”