An Adaptive Topology Management Model in Self-organizing Overlay Network

In large-scale dynamic network Gossip-based unstructured topology model is an efficient method of overlay topology management, which can better match with the underlying physical network and supply more supports about robustness and flexibility. In the process of topology self-organizing the convergent speed of topology is the most important evaluation criterion of topology management model. Increasing the number of Gossip packets exchanged can accelerate the convergent speed, but the data volume exchanged is limited to the network bandwidth. In this paper we analyze the characteristics of data exchange in topology evolution and propose an adaptive model to control the dynamic transmit cycle of Gossip packets and improve the efficiency of data exchange to accelerate the topology convergent speed. We use the Logistic Growth Curve as a basic grow function and create a feedback control to cope with the particular events. By simulation the validity of the adaptive model is verified and the results show that this model has its particular advantage in dynamic network environment.