Deepak Sethi and Partha P. Bhattacharya Pages 248 - 259 ( 12 )
Background: According to nodes’ participating style, routing protocols can be classified into three categories, namely, direct communication, flat and clustering protocols. In direct communication protocols, a sensor node sends data directly to the sink. Under this protocol, if the diameter of the network is large, the power of sensor nodes will be drained very quickly. Furthermore, as the number of sensor nodes increases, collision becomes a significant factor which defeats the purpose of data transmission. Under flat protocols, all nodes in the network are treated equally. When a node needs to send data, it may find a route consisting of several hops to the sink. Normally, the probability of participating in the data transmission process is higher for the nodes around the sink than those nodes far away from the sink. So, the nodes around the sink could run out of their power soon. In the clustered routing architecture, nodes are grouped into clusters, and a dedicated cluster head node collects, processes, and forwards the data from all the sensor nodes within its cluster. One of the most critical issues in wireless sensor networks is represented by the limited availability of energy on network nodes thus, making good use of energy is necessary to increase network lifetime.Methods: Review of literature survey and patents has been conducted on the use of artificial neural networks with wireless sensor networks. While working with wireless sensor network the key concern is energy conservation. It is quite important to extend the network’s lifetime and to reduce the energy consumption as the nodes used are battery powered and provided with a fixed amount of initial energy. So for such a target, we are localizing the base-station of the network. The base-station of the network should be placed in such a way that the energy consumption of nodes is reduced to some extent. We are using the artificial neural network approach for finding the optimized position of the base station. Results: The proposed base-station localization algorithm using artificial neural network is when embedded to LEACH protocol, then it gives better result than the LEACH protocol in which the basestation is positioned at the corners of area where sensor nodes are deployed. The results are better as shown in above figures, in terms of – number of nodes dead to number of rounds, energy consumption (in Joules) to number of rounds, number of nodes dying in terms of number of rounds and the number of packets that are being transmitted to the base-station and the cluster heads. It can be concluded that LEACH with ANN provides an energy efficient scheme. Conclusion: In this work, we have applied ANN in base-station localization and embedded in LEACH protocol. But in future, we can further enhance network lifetime by applying ANN in cluster head selection mechanism of LEACH and LEACH-C protocol.
Artificial Neural Networks, Base-Station Localization, Radial Basis Functioning, Supervised Learning, Unsupervised Learning, Wireless Sensor Networks.
Department of Computer Science and Engineering, College of Engineering and Technology, Mody University of Science and Technology, Lakshmangarh, Rajasthan