Learning automata based energy efficient data aggregation in wireless sensor networks

Recently, many algorithms have been proposed for data aggregation in wireless sensor networks which try to find routes towards the sink through which data can be aggregated. In addition to data aggregation, two more criteria are also used in many of these algorithms for finding the routes; remaining energies of the nodes and their numbers of hops to the sink. But to the best of our knowledge, no data aggregation algorithm has been presented in which all of these three criteria are considered together. In this paper, we propose a novel data aggregation algorithm, called LAG, which tries to mix all of these criteria for finding the routes. Furthermore, by considering the fact that the remaining energy of a sensor node and its possibility for aggregating data received from other nodes may change during the operation of the network, the proposed LAG algorithm tries to dynamically adapt itself with such changes and to select new routes towards the sink accordingly. The adaptive behavior of LAG is the result of using learning automata (LA). Each node is equipped with an LA which helps the node selects its next hop for forwarding data towards the sink considering all of the three mentioned criteria. The learning automaton used in LAG algorithm, called INCASE-LA, is introduced in this paper for the first time. Using computer simulations, we demonstrate that LAG aggregates data better, consumes less power and achieves higher network lifetime in comparison to other existing algorithms such as SPT, TAG, and ES LA.