An Entropy - Based Approach to Integrated Information Needs Assessment

With an overload of sensory inputs, fusion processing must ultimately be scoped based upon the requirements of the consumer of the fused data. This idea, called computational steering, allows the fusion system to process only the type of information relevant to the consumer’s needs. Simple approaches include filtering based on spatial proximity, latency, and data source. Although these methods are useful, the amount of data left for processing after applying these filtering methods may still be enormous. An integrated method for assessing the needs of a fusion process is required to make sure the informative data is processed first. Lockheed Martin Advanced Technology Laboratories is developing an entropy-based approach to identifying information needs so that computational steering can be performed in an intelligent manner. Using the expected reduction in entropy, data is dynamically selected for fusion processing. As a result, the maximum discrimination gain is obtained each time data is consumed by the fusion process. This entropy based approach ensures that fusion processes are processing the most informative, minimizing the processing of less informative data. 1. Motivation With the rapid growth of deployable sensor assets and network connectivity, the battle space has become inundated with information. As a result, the fusion community is faced with increasing amounts of heterogeneous information to process. In the past decade, efforts have been made to reduce the computational load for Level 1 Fusion systems by processing only the data that is necessary and relevant. In some cases, attempts have been made to quantify the value of the available data so that the most informative information is processed first. This concept is called Information Needs Assessment. In the Level 1 Fusion context, Information Needs Assessment takes the form of real-time sensor cueing [1,2,3]. The success of Level 1 Fusion approaches to Information Needs Assessment has inspired Lockheed Martin Advanced Technology Laboratories (ATL) to develop analogous approaches to the Level 2 and 3 Fusion problem space.