Biologically-inspired path selection scheme for multipath overlay networks

In this paper we discuss the application of a biologically-inspired approach for path selection in overlay networks under the constraints of in-order packet delivery. We apply the concept of attractor selection by modeling the path selection process in multipath overlay as a stochastic dynamical system that instantaneously converges to a sufficiently good solution utilizing the system inherent noise to provide a robust and adaptive mechanism. Our method is inspired by a mathematical model for the stochastic dynamics of gene regulatory networks. Simple numerical simulations will show that this proposed method is stable to changes in the environment and adaptable to the objective function, which is in our case the buffer occupancy level of the destination node in an overlay network.