The Application of Track before Detect Techniques against Maritime Surface Targets
Abstract
Real radar data containing a small manoeuvring boat in sea clutter was processed using a grid based finite difference implementation of continuous-discrete filtering. An examination was undertaken to determine the appropriate dynamic, target amplitude and clutter amplitude models which should be utilized to allow the successful application of Track Before Detect techniques (TkBD). Both two dimensional diffusion and four dimensional constant velocity models were implemented using Gaussian and Rayleigh sea clutter models. Superior performance was observed for the constant velocity model and significant sensitivity was noted due to mismatches between actual clutter characteristics and Gaussian and Rayleigh models. TkBD performance was examined assuming a Rayleigh sea clutter model with embedded Swerling 0, 1 or 3 target signal models. The Swerling 0 model was observed to exhibit a heightened sensitivity to changes in measured signal strength and provided improved detection of the maritime target examined in comparison with Swerling 1 and 3 targets at the cost of more peaked or multi-modal posterior density. The potential for achieving significant detection performance improvements by utilizing K and KA distributed clutter models in place of the simpler Rayleigh distribution was demonstrated through analysis of simulated data representing spiky sea clutter. However, additional analysis using real data revealed that use of a probability distribution function more closely matched to the observed real sea clutter returns does not necessarily result in improved performance. For the data set examined, significantly degraded performance was observed when K and KA based processing is used in place of a Rayleigh based processor utilizing a simple likelihood limiting step to compensate for model mismatches due to sea clutter spikes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2010
- Accession Number
- AD1001570
Entities
People
- Bhashyam Balaji
- Michael Mcdonald