D. GopalaKrishnan, D.Santhi Jeslet
The research we study the performance of semi-cluster based network algorithms in networking , a new algorithm to address the shortcomings of both routing and clustering based as well as the traditional algorithms. Due to the unstable nature of MANETs memory based search algorithms such as clustering based methods become efficient and practical as the network mobility increases. Our simulations show that ant based routing algorithms suffer from lack of accuracy while incurring extensive memory usage as well as valuable processing power and bandwidth costs in highly dynamic environments. The proposed semi-cluster based routing algorithm exploits a dual-mode approach. Each node can independently operate in a local mode as well as in global mode using the information provided by rule based mining. The network topology changes are constantly monitored. When the rate of topology change gets too high for converge efficiently, a node switches to local mode to rely less on the information learnt through ants. Our simulations show how this switching approach significantly improves the performance of the network by quickly adapting to the dynamics of the environment.
-