METRON INCORPORATED — Department of Defense SBIR Phase I: ABSTRACT: The goals of this project are to (i) develop a theoretical framework for a full

Description

ABSTRACT: The goals of this project are to (i) develop a theoretical framework for a fully adaptive radar (FAR) system that includes specification of the feedback mechanism from the receiver to the transmitter and specification of performance metrics to assess FAR system performance, and (ii) to develop analytical as well as computer simulation methods for determining the performance improvement achieved by the FAR system over standard fee-forward radar (FFR) systems. We propose to develop a general FAR framework that can be applied to a variety of missions. Using a first principles approach, we develop a framework that consists of seven components: environment, transmitter, receiver sensor, adaptive detector, covariance matrix estimator, mission processor, and controller. In our formulation, the mission performance objective induces component performance objectives on the adaptive detector, covariance matrix estimator, and mission processor (tracker). We will also develop a computer simulation and experimental test to demonstrate the performance of the model, and will analyze the performance of the FAR system in terms of global performance metrics, component performance metrics, training data support, and computational complexity. BENEFIT: Radar systems are crucial for robust surveillance, target acquisition, and reconnaissance in all weather conditions and over wide ranges of interest. Most radar systems employ a feed-forward processing chain in which they first perform some low-level processing of received echo data and then pass the processed data on to some higher-level processor, which extracts information to achieve a mission objective. The application of artificial cognition to radar systems offers much promise for improved sensing as well as the creation of new sensing modalities. Specifically, FAR offers the potential for two to ten times performance over state-of-the-art in terms of output signal-to-noise-ratio (SNR), and error variance in parameter estimation. This translates to a 3-10 dB improvement in target detection performance over the state-of-the-art. The framework developed in Phase I will be quite general. As such it can be applied to a variety of radar systems for a variety of missions, and can be translated into other domains such as computing, autonomous vehicles, and perhaps even neuroscience and evolutionary biology. The project will develop the fundamental tools necessary to design and analyze a cognitive processing system, and will enable further research in the abovementioned diverse fields.