Geoacoustic tracking for three configurations: (a) Temporal tracking of the average, range-independent environment using a fixed-VLA-receiver and a towed-source, (b) spatial tracking of range-dependence using a towed-HLA-receiver and a towed-source, and (c) temporal tracking of the ocean sound speed profile for a fixed-VLA-receiver and a fixed-source.
Seven-parameter geoacoustic model used in the simulations.
Example 1: Comparison of the tracking algorithms: (a) Evolution of 100 different environments (Monte Carlo trajectories), and (b) RMS errors for the EKF, UKF, 200-point PF, and 2000-point PF obtained from tracking each of these 100 trajectories along with the theoretical lower limit for the RMS error, the square root of the posterior CRLB.
Example 1: Performance improvement of PF as a function of number of particles expressed in terms of (a) filter efficiency, and (b) improvement over the EKF. The dashed line shows the attainable improvement limit.
Example 2: Normalized objective functions for three different frequencies. The cost function for each parameter is obtained by fixing all other parameters to their true values (dashed line).
Example 2: Tracking results of EKF, UKF, PF-200, and PF-5000 for the seven-parameter environment given in Fig. 1 using the long range VLA. True trajectories (dashed) are provided along with the tracking filter estimates (solid).
Example 2: Evolution of the magnitude of the vertical acoustic field in the water column at the receiver array as the environment evolves in time. Hydrophone locations (circle) show the vertical sampling interval of the time-varying field.
Example 2: Posterior probability density evolution for the sediment thickness for a 10000-point particle filter and the EKF: (a) Six snapshots at , 40, 84, 116, 152, and with the local particle histograms representing the PF distribution (solid) and the EKF Gaussian PDF (line) along with the true trajectory (dashed). (b) The continuous evolution of the PF PPD together with the local mean standard deviation of the EKF Gaussian.
Example 3: Tracking results of EKF, UKF, PF-200, and PF-5000 for the seven-parameter environment given in Fig. 1 using the short range HLA configuration. True trajectories (dashed) are provided along with the tracking filter estimates (solid).
Environmental and simulation parameters used in example 1.
Performance comparison for example 1.
Simulation parameters for example 2.
Results for example 2.
Simulation parameters and percent improvement of filters for example 3.
Article metrics loading...
Full text loading...