The properties of statistical error covariance matrix (ECM) with incomplete measurements were studied for the phenomenon of missing measurements which often occurs in target tracking. The relationship between the steady filtering error variance and the measurement noise with incomplete measurements was analyzed. While the detection probability was assumed to be known, a new filter algorithm with variance-constrained index was proposed which allowed the measurement noise intensity as high as possible, so the requirement of the sensor could decrease. In order to demonstrate the usefulness of the proposed design approach, a numerical example was presented as well as Monte Carlo simulation.
