Comments on ´Cramer-Rao lower bound for tracking multiple targets´

In a recent paper, Ristic et al., 2004, proposed a simple and exact derivation of the theoretical Cramer-Rao lower bound (CRLB) for tracking multiple targets using intensity maps as measurements. Although the formulation and the theoretical derivations presented in their work are indeed correct, the analysis of the proposed bound and the conclusions that follow are wrong. Here, a correct analysis is presented with the key observation that [as opposed to the claim made in the work of Ristic et al. (2004)], the proposed CRLB reflects the interaction between the targets, even if their motion is independent. The multi-target CRLB is also compared with the error performance of a multi-target state maximum likelihood estimator; a remarkable agreement is observed for high signal-to-noise ratio