LAUSR

Diversity has long been used as a design tactic in computer systems to achieve various properties. Multi-agent systems, in particular, have utilized diversity to achieve aggregate properties such as efficiency of resource allocations, and fairness in these allocations. However, diversity has usually been introduced manually by the system designer. This paper proposes a decentralized technique, clonal plasticity, that makes homogeneous agents self-diversify, in an autonomic way. We show that clonal plasticity is competitive with manual diversification, at achieving efficient resource allocations and fairness.