Multichannel Detection Using Higher-Order Statistics
Abstract
A methodology was developed for the detection of an unknown signal observed simultaneously over multiple channels. The model-based approach was adopted, with a binary decision space. In model-based detection, the parameters of a model are identified from the multichannel process, and the identified model is used to facilitate detection of the desired signal. Identification methods based on higher-order statistics were adopted to estimate the model parameters. The formulation developed in this program is generic but, in Phase 1, emphasis was placed on airborne surveillance radar array applications. In such systems, the array elements constitute the channel outputs. Ground clutter, interference sources, and noise sources are present in the multiple channel outputs along with the target signal. Applicability of the technique to surveillance radar array systems was established by identifying several radar operational conditions wherein target and/or clutter exhibit non-Gaussian statistics. Additionally, a processing option was identified to modify the channel output data and enhance its non-Gaussian characteristics. Simulation-based analyses were carried out to investigate key technical issues, and to validate fundamental aspects. Results indicate the methodology can discriminate between target-present and target-absent conditions.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1995
- Accession Number
- ADB198116
Entities
People
- D. W. Davis
- J. R. Roman