DACA: Dynamic Advanced Clustering Algorithm for Sensor Networks

Abstract

The clustering of sensor nodes is an effective topology control approach that can balance the load on sensor nodes and increase network scalability and lifetime. But the environments of sensor networks are inherently dynamic and they are continuously in change (e.g. military environments). In addition the sensor nodes have limited energy supply and after consumption of it they will die, so it is necessary to replace these dead nodes with some fresh nodes occasionally. Thus the cluster structure of sensor networks has dynamic nature. At the Other hand the clustering process will impose some overhead to the network and it is not efficient to execute it more than once. Therefore an algorithm that can dynamically cluster sensor networks makes clustering more efficient. Our approach in this paper doesn't need to be executed periodically all over the network. Our proposed protocol DACA (dynamic advanced clustering algorithm) mainly would be executed at set up time all over the network and any time any changes are sensed in the network it locally reconfigures the cluster structure in the clusters that influenced by that changes and correct the cluster structure in a dynamic manner. This method reduces the overhead of clustering and prolongs the network lifetime. In addition the DACA can support multi-hop clusters. In multi-hop clusters the members of a cluster can be connected to their cluster-head in an indirect way with more than one hop. This is useful when communication cost in multi short hops is more efficient than one long hop (e.g. when the noise in the environment is high).