Biomimetik.dk

An Internet hosting center hosts services on its server ensemble. The center must allocateservers dynamically amongst services to maximize revenue earned from hosting fees. Thefinite server ensemble, unpredictable request arrival behavior and server reallocation cost makeserver allocation optimization difficult. Server allocation closely resembles honeybee foragerallocation amongst flower patches to optimize nectar influx. The resemblance inspires ahoneybee biomimetic algorithm. This paper describes details of the honeybee self-organizingmodel in terms of information flow and feedback, analyzes the homology between the twoproblems and derives the resulting biomimetic algorithm for hosting centers. The algorithm isassessed for effectiveness and adaptiveness by comparative testing against benchmark andconventional algorithms. Computational results indicate that the new algorithm is highlyadaptive to widely varying external environments and quite competitive against benchmarkassessment algorithms. Other swarm intelligence applications are briefly surveyed, and somegeneral speculations are offered regarding their various degrees of success.