Authors: G. Vennira Selvi
Addresses: Pondicherry Engineering College, Pondicherry, India
Abstract: In wireless sensor networks, the lifetime of the sensor network is hottest research topic. Efficient usage of energy in each sensor node is very important issue in designing the network topology which affects the lifetime of sensor networks significantly. Clustering provides an efficient method for maximising the network lifetime. In clustering, the cluster head failure can disconnect the cluster members and cluster from the base station. To prevent this and to extend the network lifetime, we propose a fault prediction algorithm. In this algorithm, each sensor node tries to estimate the amount of energy it will spend to transfer k bit data in the near future using integrated double exponential smoothing model. It uses two different smoothing factors to estimate the current energy level and uses it as a forecast for the future value. The proposed algorithm is simulated in NS2.34 and compared with clustering and fault tolerance for target tracking using wireless sensor networks (FTTT). The results show that our algorithm effectively prevents the cluster head and cluster members from failure by predicting the energy level of each sensor node effectively and extends the network lifetime significantly.
Keywords: unequal clustering; fault prediction; reliability; network lifetime; wireless sensor networks; WSNs; energy consumption; simulation.
International Journal of Convergence Computing, 2015 Vol.1 No.3/4, pp.244 - 258
Received: 22 Jul 2014
Accepted: 29 Dec 2014
Published online: 22 Apr 2016 *