Top K-leader election in mobile ad hoc networks

Many applications in mobile ad hoc networks (MANETs) require multiple nodes to act as leaders. Given the resource constraints of mobile nodes, it is desirable to elect resource-rich nodes with higher energy or computational capabilities as leaders. In this paper, we propose a novel distributed algorithm to elect top-K weighted leaders in MANETs where weight indicates available node resources. Frequent topology changes, limited energy supplies, and variable message delays in MANETs make the weight-based K leader election a non-trivial task. So far, there is no algorithm for weight-based K leader election in distributed or mobile environments. Moreover, existing single leader election algorithms for ad hoc networks are either unsuitable of extending to elect weight-based K leaders or they perform poorly under dynamic network conditions. In our proposed algorithm, initially few coordinator nodes are selected locally which collect the weights of other nodes using the diffusing computation approach. The coordinator nodes then collaborate together, so that, finally the highest weight coordinator collects weights of all the nodes in the network. Besides simulation we have also implemented our algorithm on a testbed and conducted experiments. The results prove that our proposed algorithm is scalable, reliable, message-efficient, and can handle dynamic topological changes in an efficient manner.