Distributed Learning Automata-Based Clustering Algorithm in Wireless Ad Hoc Networks

In ad hoc networks, the performance is significantly degraded as the size of the network grows. The network clustering is a method by which the nodes are hierarchically organized on the basis of the proximity and thus the scalability problem is alleviated. Finding the weakly connected dominating set (WCDS) is a well-known approach, proposed for clustering the wireless ad hoc networks. Finding the minimum WCDS in the unit disk graph is an NP-Hard problem, and a host of approximation algorithms have been proposed. In this paper, an approximation algorithm based on distributed learning automata is first proposed for finding a near optimal solution to the minimum WCDS problem in a unit disk graph. Then, a distributed learning automata-based algorithm is proposed for clustering the wireless ad hoc networks. This clustering method is a generalization of the algorithm proposed for solving the WCDS problem, in which the dominator nodes and their closed neighbors assume the role of the cluster-heads and cluster members, respectively. The proposed clustering algorithm, in an iterative process tries to find a policy that determines a cluster-head set with the minimum cardinality for the network. For both algorithms, the simulation results show that they outperform the best existing algorithms in terms of the number of hosts (nodes) in the cluster-head set (dominating set).