Local Halting Criteria for Stochastic Diffusion Search Using Nature-inspired Quorum Sensing

Stochastic Diffusion Search (SDS) is a Swarm Intelligence algorithm in which a population of homogeneous agents locate a globally optimal solution in a search space through repeated iteration of partial evaluation and communication of hypotheses. In this work
a variant of SDS, Quorum Sensing SDS (QSSDS), is developed in which agents employ only local knowledge to determine whether the swarm has successfully converged on a solution of sufficient quality, and should therefore halt. It is demonstrated that this criterion performs at least as well as SDS in locating the optimal solution in the search space, and that the parameters of Quorum Sensing SDS may be tuned to optimise behaviour towards
a fast decision or a high quality solution. Additionally it is shown that Quorum Sensing SDS can be used as a model of distributed decision making and hence makes testable predictions about the capacities and abilities of swarms in nature.