Adaptive virtual topology control based on attractor selection

Abstract

One approach for accommodating traffic on a wavelength-routed optical network is to construct a virtual topology by establishing a set of lightpaths between nodes. To adapt to various changes in network environments, we propose an adaptive virtual topology control method, which reconfigures virtual topologies according to changing network environments, in IP over wavelength-routed wavelength division multiplexing networks. To achieve adaptability in the virtual topology control method, we focus on attractor selection, which models behaviors where biological systems adapt to unknown changes in their surrounding environments. The biological system driven by attractor selection adapts to environmental changes by selecting attractors of which the system condition is preferable. Our virtual topology control method uses deterministic and stochastic behaviors and controls these two appropriately by simple feedback of IP network conditions. Unlike current heuristic virtual topology control methods developed in the area of engineering, our method does not rely on pre-defined algorithms and uses stochastic behaviors for adapting to changes in network environments. The simulation results indicate that our virtual topology control method based on attractor selection adaptively responds to changes in network environments caused by node failure and constructs operational virtual topologies in more than 95% of simulation trials when 20% of nodes in the physical network fail simultaneously.