Multichannel Detection Using Higher-Order Statistics

Period of Performance: 08/30/1994 - 08/30/1996

$198K

Phase 2 SBIR

Recipient Firm

Scientific Studies Corp.
2250 Quail Ridge
Palm Beach Garden, FL 33418
Principal Investigator

Research Topics

Abstract

Many sensor-based systems are inherently multichannel. This includes radar, sonar, and communications systems. Multichannel processing techniques offer enhanced signal detection performance over less optimal, single-channel methods. In this program, multichannel identification and detection is pursued in the context of radar systems via a model-based approach. A methodology was developed in Phase I which uses higher-order statistics (HOS) for the identification of time series model parameters. The model pseudo-innovations sequence is used for the detection decision. In general, HOS-based algorithms offer several advantages over conventional time series methods based on second-order statistics. HOS-based algorithms specifically address the cases vhere the radar return does not exhibit Gaussian statistics, which is the situation in many radar system applications. Additionally, HOS-based algorithms are immune to additive Gaussian noise, such as receiver noise and interference sources. Feasibility of the proposed detection methodology was demonstrated in Phase I. In Phase II, a detailed assessment of the detection performance of the HOS-based methodology will be conducted, and an optimized candidate architecture for algorithm implementation vill be defined. Also in Phase II, the algorithm will be implemented on a processor development system having a parallel processor architecture.