On the linearly constrained minimum mean square error estimate

It is shown that the minimum mean square error estimate under a linear constraint is a linear sum of the unconstrained estimate and the constraint. This result is derived without further restricting the estimated random variable or the observations.