Intelligent Automation, Inc. — Department of Defense SBIR Phase I: A18-044

Description

We propose to design and implement the Resource Aware Information Sharing Intelligent Network (RAISIN). The objective of RAISIN is to improve communications in a tactical network with situational awareness and provide the commander with actionable intelligence that is represented by the value of information (VoI). The tactical network consists of artificial and human agents each equipped with embedded sensors and operating with different communication links subject to computational (processing power), bandwidth and other (such as energy) constraints. Each agent generates metadata obtained with deep learning methods describing the recorded content and decides what metadata to transmit depending on its resources. RAISIN uses deep learning to characterize the VoI for the generated metadata and assist commander to produce queries according to the mission. RAISIN also applies deep learning based information retrieval methodology to find low-cost models according to agent’s processing power and battery lifetime to generate metadata and handle potential errors in metadata processing. The goal is to deliver the best VoI such that the commander can make the decisions faster and with higher fidelity.