SBIR-STTR Award

The mission of passive undersea surveillance is to classify potential threats by examining the acoustic signatures of detected sources. Current methods are based on visual examination and operator translation of the acoustic scene. With the ongoing reduction in Navy personnel, an automated system requiring minimal operator interaction with improved classification capabilities will be needed to support current and future automated surveillance requirements. In the past many attempts have been made to develop and automated classification system. These systems generally failed, primarily due to lack of acoustic expert involvement in the development of these classification systems. Additionally, front-end processors were inappropriately designed and did not provide information required by the classifier to produce high-confidence results. ORINCON has successfully developed and demonstrated prototype automated undersea surveillance systems utilizing expert processors on both SPAWAR and DARPA programs, such as the Automated Surveillance Information Processing Systems (ASIPS) and the 7100 Site Evaluation System. In Phase I we will apply our extensive experience in developing automated detection and classification systems to design an information processor (IP) and utilize in-house experts to develop the associated rulebases required to perform automated classification. In Phase II, a prototype system will be developed to demonstrate this concept.

Keywords:
INFORMATION PROCESSOR AUTOMATED CLASSIFIER DATA FUSION AUGMENTED TRANSITION NETWORK (ATN)

The U.S. Navy's complex passive acoustic surveillance and ASW systems depend significantly on an acoustic operator/analyst's ability to quickly recognize potential targets of interest, analyze the parameters and characteristics of the acoustic data presented to the operator, evaluate the meaning of the parametric data and classify the target. With the advent of numerous friendly and threat and neutral forces operating in the world, the task of consistently classifying numerous acoustic targets has become extremely difficult. Acoustic Intelligence data on hundreds of target types is impossible o memorize or quickly reference. Operators need a passive acoustic data classification tactical decision aid which can help them reference important parametric data and assist in accurate and timely classification of the targets. This proposal effort will research, design, develop, test and implement an expert-based Passive Acoustic Classification System (PACS), incorporating unique acoustic signature features extraction algorithms, ACXINT Specialist-developed Target Characteristics Data Base, use of SRC/Unisys-developed Fussy-Conditioned Dempster Shafer Classification Reasoning algorithms; Machine-assisted PACS MMI routines to optimize operator/algorithms-generated input data characterization; and FCDS algorithms-based data base Query, Match, Score and classification code/algorithms, migrated to IUSS-standard work station environment.