Covariance Matrix Estimation
In a real system, we don’t know the exact covariance matrix of the interference.
Therefore, we have to estimate it from the data itself.
A reasonable estimate of E{xixj*} (the i, j component of the covariance matrix)
is the average of m independent samples of data from the same distribution
To extend this approach to the entire covariance matrix, we can represent a
collection of M independent “snapshots” of data from the N channels as an
MxN matrix X. Then we have:
Each of these is an outer product
(product of a column with its conjugate transpose)