Cramér-Rao Lower Bounds in Signal Processing Applications

The Cramér-Rao Lower Bound (CRLB) is a lower bound on the covariance matrix of the error of unbiased vector parameter estimators. It represents a bound on an information content of data about an unknown parameter in a statistical model of the given data. CRLB is a classical tool stemming from the works of Cramér, Rao [8], [9], and other researchers in 1950's. CRLB has many extensions and modifications: a Bayesian CRLB for random parameters [10], [11], a hybrid CRLB for a mixture of random and deterministic parameters [12], a CRLB for biased estimates [13].