A predictive mechanism for enhancing adaptability of self-organised routing

To tackle problems emerging with rapid growth of information networks in scale and complexity, bio-inspired selforganisation is a promising design principle for future networks. However, self-organising systems fall into local optima or converge slowly under some environmental conditions. This can make self-organising systems slow to adapt to environmental change, despite robustness against environmental change being an important feature expected from self-organisation. To adapt to dynamically changing conditions while retaining its distributed nature, each component predicts the future state of its neighbours from past behaviour, and proceeds according to the predicted states. We take AntNet, an ant-based routing protocol, and add a mechanism to accelerate path convergence with prediction. Simulation results show that introducing our predictive mechanism reduces recovery time by up to 60%.