Ecologically Inspired Agent Control Model

In a typical agent based system, a number of mobile agents cooperate to achieve a desired goal. The efficiency of the agent system in reaching the goal, and the completeness of the result depends on the number of agents in the system. Too few agents will not achieve the full potential of parallelism, and will lead to decreased system efficiency. Too many agents can overburden the system with unnecessary overhead, and also can result in significant delays. The task of finding the optimal number of agents required to achieve the desired effect is difficult and problem-specific. In this paper, we propose an ecosystem inspired approach to this problem. Similar to a real ecosystem, our solution will exhibit properties of emergent stability, decentralized control and resilience to possible disturbances. In our work, we propose to solve the technical problem of agent management using an ecological metaphor. In Section 2 we describe the current state of research in the fields of simulated ecosystems and multi-agent control and stability. In Section 3 introduces the problem of managing the number of agents populating a physical network, as well as explain a proposed solution. Lastly, Section 4 demonstrates the initial experimental results and conclusions.