OPTIMUM MULTISENSOR, MULTITARGET LOCALIZATION AND TRACKING

The optimum methodology of estimating and tracking target location parameters (e.g., range and bearing) in a passive multisensor, multitarget environment is investigated. Results of this study show that the localization parameters are best obtained via a two-step procedure. First estimate the time delay vector using the optimum multisensor multitarget time delay processor. Next obtain the localization parameter estimates from the time delay vector measurements via geometric mappings between targets and sensors.^ The optimum multisensor, multitarget time delay processor was studied in detail. The optimum processor was first derived assuming that the observed random waveforms (consist of coherent Gaussian broadband signal waveforms in additive incoherent Gaussian, broadband noise) can be considered as Stationary Parameter Long Observation Time Process. Here the observed wave-forms can be reduced to a finite dimensional, complex Gaussian observation vector. A Maximum Likelihood multisensor, multitarget time delay processor is obtained by reducing and solving the resulting vector likelihood equation.^ The performance of the optimum processor is analysed in terms of the Cramer-Rao Lower Bound. Analytical closed form expressions are obtained for the two-targets, two-sensors, and one-target, M-sensors (for any M) cases. In addition, relation between time delay estimation and localization parameter estimation is explored. Comparison between optimum and suboptimum implementation is considered. A joint time delay and spectral estimation processor is presented. A method for improved time delay estimation using a post correlator matched filter approach is also investigated.^ To account for the moving target environment, a variable multisensor, multitarget time delay processor is derived. The basic derivation is obtained by partitioning the observation interval into smaller subinterval. It was shown that the length of each partitioned time interval must satisfy certain constraint related to time delay doppler and signal bandwidth. The variable time delay is modeled by a finite order polynomial over the whole observation time interval. By estimating the time delay and delay rate at any given point in time, the corresponding localization parameters and the target motion parameters can be derived through a zero memory geometric transformation. This is the integrated approach to multitarget signal/data processing.