Data Fusion for Ship Self-Defense Using Massively Parallel Processing

The capability of combat systems in littoral warfare is decreased by conflicting data. ORINCON proposes a solution that integrates on- and offboard sensors to provide a better understanding of the situation. In particular, we are proposing a solution for the anti-air warfare mission. ORINCON proposes to start with its multiple hypothesis tracking (MHT) algorithm. MHT provides a solution for fusing (or associating) data together. Sometimes the associations are unclear and require additional information not currently available. In order to postpone a decision, a number of hypotheses are allowed to form, indicating the possible associations. Current sequential computers limit the number of hypotheses by pruning, perhaps removing the "true" association too soon. Use of a massively parallel processor (MPP) in this situation allows more time to examine the associations and maintenance of more hypotheses. The objective of this project is to prototype a scalable data-associator and multihypothesis tracker that can be inserted into the AAW mission. The prototype MPP is the 336-processing-node Intel Paragon at NCCOSC with the Scalable Programming Environment (SPE) developed at NCCOSC. The SPE reduces programming complexity by hiding details of parallel communication. Together, SPE and MPP have shown that image processing times can be reduced linearly by increasing the number of processing nodes.