Use of Trust Vectors for CyberCraft and the Limits of Usable Data History for Trust Vectors

The CyberCraft initiative is designing a system of agents to provide command and control of future air force information systems. In such a distributed environment, trust between agents becomes pertinent to provide a combatant commander an accurate picture of the network by merging conflicting data and giving the commander confidence in the control of their network. The trust vector model breaks from traditional trust models as it provides a range of trust values for data, enabling the system to describe partial trust in data rather than binary values of trust or no trust, and from that build confidence intervals in fused data. The trust vector model builds trust based on current and historical data, but does not address the limits of utility of the historical data. This research provides implementors of the trust vector model input as to how much historical data should be stored to balance accuracy of trust with limits of storage