MAP-PF wideband multitarget and colored noise tracking

The maximum a posteriori penalty function (MAP-PF) approach is applied to tracking the bearing and bearing rate of multiple wideband sources in unknown colored noise. The track estimation problem is formulated directly from the array data using the maximum a posteriori (MAP) estimation criterion. The penalty function (PF) method of nonlinear programming is used to obtain a tractable solution. A sequential update procedure is developed in which penalized maximum likelihood estimates of target bearings, spectra, and noise parameters are computed and then used as synthetic measurements in a set of Kalman filter trackers. The two steps are coupled via the penalty function. During parameter estimation, the penalty function prevents erroneous outlier estimates which can cause the tracker to lose track. During the tracking step, it determines the influence of the parameter estimates on the final track estimates by adaptively adjusting the measurement error variance. Performance is demonstrated on a typical sonar scenario.