Defence Applications of Agent-Based Information Fusion

The defence landscape is populated by multiple information sources and consumers. The information sources must correlate and combine their information in order to derive accurate situation awareness. This motivates information fusion solutions. The information consumers must be proactive in seeking relevant information and managing their joint decisions in response to the information. This motivates an agent-based approach. The need for accurate and timely situation awareness drives decisions about where to gather new information. This is fused with previous information and the fused products influence future decisions, creating a feedback loop. Agent-based information fusion (ABIF) is the result of combining information fusion and agent-based methods in order to control multiple such feedback loops executing within a large-scale system. The desired effect of ABIF is to drive positive emergent behaviours that increase the operational effectiveness of the system. This paper describes the modern defence landscape and states its main requirements for ABIF solutions. Core functionality is indicated in relation to specific defence application examples. Then the paper highlights two case studies in the areas of situation awareness and resource management. The first demonstrates how information fusion techniques developed in the ALADDIN project can be used to ensure robust estimation in sensor fusion networks when some of the sensor data may be biased and faulty. The second study demonstrates that decentralized agent-based solutions for resource management outperform a distributed (i.e. replicated central) solution when the communication networks are sparse and subject to delays.