Assumptions
We know the signal
- its channel-to-channel properties are modeled by the “steering vector”, s: x = Aejwts (|s| = 1)
The noise and interference are zero-mean Gaussian random processes
We know the noise + interference covariance matrix
- by modelling, or
- by estimation from existing data
- estimation requires that the noise+interference be stationary over the samples
- estimating C can be anywhere from a minor annoyance to a major problem!