Using Autonomous Agents to Improve Efficiency and Robustness in Slow, Unreliable Networks

A number of environmental factors commonly present in military and particularly tactical military contexts affect the ability of applications to operate robustly and efficiently: ¯ Machines hosting agents may shut down due to damage or power loss ¯ Hostile forces may disrupt communication links ¯ Communication links may be exceedingly slow, since some signals are transmitted over tactical radio links with maximum data rates as low as 4.8 kbps ¯ Tactical processors and network services may be over-loaded, impeding the completion of agent tasks. Mobile agents can significantly reduce load on networks in environments where bandwidth is limited and only a small percentage of the information available in the net-work is needed by an application. (Kotz et. al. 2000). However, to exploit this capability in military tactical net-works, which are particularly fragile and dynamic, agents must be highly robust and fault-tolerant. In this abstract, we present a brief case study that illus-trates problems common to military environments, and we describe ideas for improving agents ’ decision-making capabilities to enable them to operate efficiently and robustly in the face of degraded network conditions